User talk:WereSpielChequers

From Wikimedia Commons, the free media repository
Jump to: navigation, search
Welcome to Wikimedia Commons, WereSpielChequers!


Category:Churches in England by patron saint by county[edit]

I noticed I was close to stepping on your toes, so I backed off and populated Category:Churches in England by patron saint by county with empty sub categories for nearly every county. Still not sure if it should be in or on the Isle of Wight and having Yorkshire as well as West, North, South and the East Riding seems less well used. Scillystuff (talk) 18:42, 14 October 2011 (UTC)

Hi apparently churches are on islands and in counties. Yorkshire is complicated, if I remember rightly the three historic ridings are East North and I'm sure whether south or West. Then the early 70s reorganisation created West, South and North plus the northern half of Humberside, which I presume is now East. And I'm not sure what happened in Tyne and Wear. So I'm mainly operating in counties which haven't been reorganised. WereSpielChequers (talk) 18:56, 14 October 2011 (UTC)
I have knocked up a little script that creates most of the Religion churches in county links for churches of a religion and suggests the text for the category. I have put an example for the LDS in my sandbox here - User:Scillystuff/Meetinghouses of The Church of Jesus Christ of Latter-day Saints in county. So if I find an image of a meetinghouse which does not have a county category I can quickly and easily create it. This is only a beta test, so let me know if you might find it useful and if it could be improved. Thanks. Scillystuff (talk) 12:13, 15 October 2011 (UTC)
That sounds like a good idea, though I'm nervous about creating empty categories as though it is much more efficient to set every county in one pass there is a risk of them being deleted. Oh and I think you need on the Isle of Wight rather than in. I'm slowly doing quakers, RC, URC Baptists, Methodists and so forth so this sort of script can be useful provided we then populate the categories quickly enough to avoid the deletionists. WereSpielChequers (talk) 12:50, 15 October 2011 (UTC)
I understand your concerns about deletion, so I haven't created any LDS county categories yet that were empty - I am only creating categories as I need to fill them and it is a lot easier to do based on the script created hints page. I think I have covered Methodist and Anglican and did RC some months ago. I'll have a look at Quaker. Scillystuff (talk) 13:04, 15 October 2011 (UTC)
Thanks, your list includes some of the old counties, Middx, cumberland etc, I think we are doing this on current boundaries, but I'm not sure and I've seen some stuff that was wrong on any boundary. WereSpielChequers (talk) 13:08, 15 October 2011 (UTC)
I took my list from the input for the Template:Counties of England which does include some of the old counties. I've only populated the entries based on the suggestion from MaPit (from mySociety) on the containing areas (ward, constituency, etc.), an example being the Cumbrian church you have just categorised. This suggestion can be accessed from the geographical coordinates of the image (camera location) - MapIt containing areas either on the right, or listed under Great Britain. I've now done the hints for User:Scillystuff/Quaker meeting houses in county. Scillystuff (talk) 14:10, 15 October 2011 (UTC)
Thanks, I don't suppose you have a semiautomated way of creating the categories for the patron saints by county? It's much quicker to move them to the counties if the county is already populated. With Mary, Mary the Virgin, Michael, All Saints and a few others we can be pretty confident that every county will need a category. WereSpielChequers (talk) 14:15, 15 October 2011 (UTC)
Just doing User:Scillystuff/All Saints churches in county. Do you want all the modern counties creating for the saints you mentioned? Scillystuff (talk) 14:34, 15 October 2011 (UTC)
Thanks and yes please. Also "Saints Peter and Paul" together and separately as well as James, Giles, Mark, Mary Magdelene and John the Evangelist all seem ubiquitous if its not too much trouble. Once we've got these cats its much easier to use Catalot to give a rough sort to those categories. WereSpielChequers (talk) 14:39, 15 October 2011 (UTC)

Saint name churches in England (by county)[edit]

So far I have created modern county sub-categories for:

Scillystuff (talk) 12:45, 19 October 2011 (UTC)

Naming of church categories[edit]

Hi, I notice you've been creating a lot of useful categories for individual churches: Many in Category:Saint Mary churches in England for instance. However, I do have a couple concerns with the exact category names you have are creating. As an example, Category:St Mary, Plympton. The church is universally called St Mary's (included by its own website), and my understanding is churches dedicated for saints are typically referred to in this form (as being of the Saint). In addition to this, this style of category name introduces ambiguity - does the category refer to the church or the surrounding area? There are lots of villages, districts of larger towns/cities, and electoral wards, named after the dedication of the parish church. IMO it would be better for the typical church category to be of the fuller format "St X's church, town" - it is much clearer what its of, and there's no potential for conflict with a category for the surrounding area. The use of the possessive "'s" depends on circumstance, as some churches do not use it, but as that is the more common form - it should be the default unless there is an indication that the without-"'s" version is preferred for that specific church.--Nilfanion (talk) 10:44, 20 October 2011 (UTC)

Hi Nilfanion, I agree we need to avoid conflict with categories for the surrounding areas, though I'm not sure that including the word church is often necessary for that. I usually just use the information already in Commons to set categories, and I try to minimise punctuation because I know that some punctuation characters get changed in some character sets. WereSpielChequers (talk) 11:14, 20 October 2011 (UTC)
The ASCII apostrophe (') is a standard character, and should not be affected in any character set (any more than a letter or number character would be). Minimising punctuation is reasonable (dropping the . from St. is standard for instance), but changing "St Mary's, Plympton" to "St Mary, Plympton" is a fundamental change, as the first is a valid name for the church, the second is not. Dropping the apostrophe can actually make the category name invalid. The formal name for the church is "The Church of St Mary", but I wouldn't advocate using that either ;)
Avoiding conflict with the area is important, but that's not the reason for including the "church". Our categories should use the common name, and in the general context the average church is called "St X's church" ([1], [2] and [3] etc). Once that context has been established, the word "church" is then dropped.--Nilfanion (talk) 12:19, 20 October 2011 (UTC)
It may be a standard character in ASCII, but it does cause problems. File:Rose Ash, St Peter’s church - - 272396.jpg is an example of an image which presumably had an apostrophe before it was imported from the Geograph. There's another one in an image that catalot has been hanging on for five minutes. I don't know if its the reason why Catalott has hung on me but I'm loathe to pad categories out with them. WereSpielChequers (talk) 16:10, 20 October 2011 (UTC)
As for whether the category names I've used are "correct", I'm pretty sure they all pass the test of The category name would be enough to guess the subject, but some extra text can be useful to precisely define it. WereSpielChequers (talk) 16:19, 20 October 2011 (UTC)
The problems from Geograph result from files using the non-ASCII apostrophe (’) instead of the ASCII ('). For instance, look at File:The Old Granary, St Mary's - - 488816.jpg. If the Geograph user who uploaded this had used a standard apostrophe on their upload, like the uploader of this, there would be no issues with ours. An ASCII character is universal whether its any of the following characters (between the quotes): "A", "b", "2", "$", ",", "'" or " ".
As for correctness, if something is not its name, its not its name and it is wrong to use it. That church is not "St Mary", it is "St Mary's" - "Manchster" is enough to guess its about "Manchester" but you wouldn't use it as the category ;) As for including the word "church" is extremely helpful, and helps address the fundamental point of a category name "enough to guess the subject".--Nilfanion (talk) 21:17, 20 October 2011 (UTC)
To expand that point - which is the most appropriate name for the category of this obscure church? "St Paul", "St Paul's", "St Paul's Cathedral" or "The Cathedral Church of St Paul"? The locality is the optional element and ideally should only be included if necessrary to disambiguate - with the average church its probably best to include by default. Bear in mind the name of the church normally doesn't include the place name, its a descriptive addition we have to include to enable identification.--Nilfanion (talk) 00:06, 22 October 2011 (UTC)
I'm aware that the placename isn't part of the name for churches dedicated to particular saints, that's why I use a comma. For churches that aren't dedicated to a saint the placename is very frequently in the name. But I always include the placename because I've only found a couple of saints that have only one church dedicated to them. WereSpielChequers (talk) 08:16, 22 October 2011 (UTC)

In more general terms, I am concerned about UK church category names as there's no consistency: There's some St X, town; St X Church (town); St X's Church, town and so on. I'll probably start a COM:CFD on this to try and get some consistency set up.--Nilfanion (talk) 00:06, 22 October 2011 (UTC)

Consistency is an overrated virtue, and not necessarily helpful here. Many dedications are not to Saints. Some saints are very common as are some placenames. St Mary, Newcastle would be unlikely to be unique, Category:St. Edmund, Fenny Bentley probably is and I'm quite happy with that format of church name which as you've noted I was using heavily myself, then there's Category:Christ Church, Oxford which follows your preferred pattern of including the word church, despite being instead an Oxford college. If we were to create an approved format for the names of church categories then it would be logical to reserve that format for churches, and then we get into sillyness such as Category:Christ Church, Oxford (Oxford college). Keeping the wiki non-bureaucratic and open is more important than making it consistent, I have tried to follow precedent and naming conventions that work for various types of church, but clearly not every categoriser is working to the same system, nor would the same system necessarily work equally well everywhere. Our categorisation backlog is far too great to worry about imposing conventions and thereby raising further barriers against editing. Better in my view simply to reaffirm The category name would be enough to guess the subject, but some extra text can be useful to precisely define it, then I can get back to categorising these churches. WereSpielChequers (talk) 08:16, 22 October 2011 (UTC)
The name of the church rarely includes the place name. The only real exception is for cathedrals and the like. We have to include it to disambiguate (and I have no objection to doing so), but its still not part of the name. However, when a church is "St Edmund's" its flat wrong to call it "St Edmund". You can't call the town of St Alban's, St Alban for the same reason. Look back at the example above (for St Paul's) which is the most natural name for its category? And getting the structure right is important - and its better to do it before the categories are created: Its a lot more effort to fix things later. 3 or 4 different category schemes is just not helpful. I'm analysing situation now and will start the CFD later.--Nilfanion (talk) 09:52, 22 October 2011 (UTC)
By the way, its irrelevant what Christ Church in Oxford is. The existence of other terms that end "x Church" should not impact a category scheme for churches; might just need a side note on how to handle conflict. "St Edmund" could be anything. "St Edmund's Church" is clearly a church - so better meet the enough to guess the subject, but some extra text can be useful test.--Nilfanion (talk) 09:59, 22 October 2011 (UTC)
(ec) Re the name rarely including the placename. For Anglican churches that is beginning to be almost true. Calling an Anglican church after the village is in many areas now an archaism, a relic of the times when the Anglican church could be expected to be the only church or at least the "official" one. But it still persists for many abandoned or near abandoned villages where there is unlikely to be another church, and a lot of people including many of the Geograph photographers still follow an "established church" naming convention (including I notice one of the UK board members). Amongst several protestant denominations including the place name in the name is far from rare - Baptists, Methodists and the URC very frequently follow the placename, denomination format. Getting the structure right and imposing unnecessary uniformity are not necessarily the same thing. The names of the categories for individual churches are a very low level of the structure, often the lowest level. I think if we were to change the structure it would be better to replace the "by religion" convention with "by denomination". Many Christians consider their religion to be Christian and the other denominations to be other denominations of Christianity - not separate religions. WereSpielChequers (talk) 11:39, 22 October 2011 (UTC)
Yes, certainly true with regards to non-Anglican protestant churches, bear in mind RC also tends to name its churches for the saint, not the town. The by demonination/religion thing isn't really relevant to this - besides the religion is "Christianity". If "Methodism" is being called a religion that can be fixed easily.
As for church names and my key point: Please don't remove the "'s", as it is an integral part of the name. The following categories are all incorrectly named: Category:St Luke, Torquay its St Luke's, its not Category:St Leonard, Southminster its St Leonard's, its not Category:St Margaret, Lowestoft its St Margaret's.
If the church's name is of the form "Place Denomination Church" then that should be the category name (unless the place is ambiguous). If the church name is of the form "St Mary's Church", then that should be the category name (with addition of place name for disambiguation). Those two possibilities cover 95% of churches (at a guess).
Also bear in mind that the sort of names you associate with churches "St X", "Holy Trinity", "All Saints" are commonly used for other institutions such as schools and hospitals - so dropping the word "church" for the church makes the category less clear, and increase the change of a conflict. There are also St X churches that do not use the St X's form, those churches shouldn't use it, any more than churches with it should drop the 's.
The reason for me wanting to start a CFD is there are a number of points to consider here, and that is appropriate venue (rather than your talk page). That could include the denomination variations, correct capitalization (is it "St Mary the Virgin" or "St Mary The Virgin"?), whether to use St., St or Saint and so on. The reason I started this thread, is because you should not remove the 's.--Nilfanion (talk) 12:06, 22 October 2011 (UTC)
Taking your first example I didn't start or rename that category. I'm aware of your opinion on this, and for the time being I have stopped creating categories for churches. But I have twice referred you to the policy on this, and while I appreciate that you have a particular opinion on the naming of church categories there are clearly policies and other opinions in play. If you are going to criticise me for adding a further category to a category that is valid within policy but that you think is misnamed then I'm not sure if any of my time here is worthwhile. As for coming up with rules as to whether one should use St., St or Saint, you are assuming that it is a good idea to set such rules and thereby declare other variants invalid, not because they are in anyway incorrect but because only one is our house style. My response is that the wiki does best when it avoids having a house style, if there is a particular church that one person has categorised as st. and another as saint then one can always create a redirect as redirects are cheap. WereSpielChequers (talk) 12:28, 22 October 2011 (UTC)
Yes, the St Luke's one has nothing directly to do with you - its just an illustration of the (IMO) flawed style you have been following. My problem is this is an incorrect name. Do I really need to cite policy saying we should not use incorrect names for categories? You can't just shorten "St Mary's" to "St Mary" merely to avoid the apostrophe, as Mary is not the same as Mary's. In most you can avoid it entirely by use of an alternative name. For instance for this church - "Church of St Mary, Beachamwell" and "Beachamwell Parish Church" are both ok. "St Mary's, Beachamwell" or "St Mary's Church, Beachamwell" are both also fine. I have a preference for the last of these, but all 4 are workable options. Category:St Mary, Beachamwell on the other hand is not.
Whether to include Church or not is a seperate issue (sometimes it will be necessary, sometimes it is not but may still be a good idea). Things like standardising on St, Saint, or St. (or not) are trivia and don't really matter, but there's no harm in setting one standard style. By the way, category redirects have significant drawbacks compared to normal redirects, we shouldn't create them if we can avoid them entirely.--Nilfanion (talk) 13:25, 22 October 2011 (UTC)
I'm not keen on long category names, or ones that start with what I'm unlikely to be keying into hotcat, so I don't like the ones that start "church of". That said I'm currently sorting out the LDS churches in the UK and with some difficulty following the style that someone else has set there. I think we can agree that sometimes the inclusion of church is necessary, most obviously in my view when either the place or the local school is named after the same saint as the church. There's no harm in setting one standard style providing that one does so by example, but there is a real risk in setting a standard style in that you deprecate all else and create another barrier against editors. It is a very slippery slope from agreeing a common preferred style to finding people are having their work rejected for nor following house style, and I'm not sure how one can safely start down that road without causing such problems.
I didn't know that there were drawbacks to category redirects, my understanding was that they were the preferred solution where one found duplicate categories or made a typo, and these are probably the only occasions when I use them. I certainly don't aim to create redirects for the sake of it. What are the drawbacks to them? WereSpielChequers (talk) 14:08, 22 October 2011 (UTC)
I'm not keen on overly long names either - so I agree with you on the "Church of" style. It is the formal name of the building for 99% of Parish churches, but I'm not a fan of it at all, for similar reason as you. LDS = Latter Day Saints I assume? "X Parish Church" is a necessary fall back for certain churches, primarily Cornwall, as there the village church has the same name as its village in many cases (St Kea, St Endellion, etc). My intent with trying to find a standard style is not to set a slippery slope but try to get some consistency. The one that sticks out to me is the use of bracket disambiguation for the place, comma disambiguation is standard for UK subjects.
Drawbacks of cat redirects - see Commons:Only use category redirects where necessary. I don't agree with all that essay's points, but they are certainly more problematic than normal ones; if we can avoid creating them great. The most obvious difference is unlike article redirects on en, we should never pre-emptively create category redirects: So if we use "St Mary" we shouldn't create redirects for "St. Mary" and "Saint Mary" just for the sake of it.--Nilfanion (talk) 14:20, 22 October 2011 (UTC)
LDS = Category:Meetinghouses of The Church of Jesus Christ of Latter-day Saints in the United Kingdom which I find a complete pain to follow as a convention because it is longer than the edit windows on hotcat. I'm familiar with that essay and believe I've been following it, I've certainly not been preemptively creating category redirects for the sake of it. If you've got any examples where you think I've created unnecessary category redirects then feel free to point list them here. I'm sure there've been some that were just my typos and I'd have no qualms seeing them go. I prefer comma over bracket disambiguation as the primary one for the UK, but when I need two types I use both - so "All Saints, nonunique placename, county or city (stained glass)" makes it clear that commas are for geography and brackets something else. I'm fairly sure I picked that up from observing the formats that others were using rather than inventing it myself. WereSpielChequers (talk) 14:50, 22 October 2011 (UTC)
I suppose you could use "Mormon meetinghouses in..." as a redirect, but that kind of goes against what we've just been saying (and not all Mormons are LDS either, though it might be OK in UK context). Likewise I agree on comma vs bracket disambiguation, but there are a number of inappropriate bracketed places such as Category:St Agnes (Cawston). You haven't created any (as far as I can tell) but they should be deprecated IMO as part of any standarising effort. Btw there's a few bad ones suchas Category:St. Giles Church!--Nilfanion (talk) 15:28, 22 October 2011 (UTC)

Mass renaming[edit]

By the way, I am thinking of running over all the UK church categories to rectify this problem as I see it: the use of St X instead of St X's. I will take care not to move any which really are called St X (some are). I don't want a rather pathetic edit war over this of course, so does that sound OK with you?

I'd also ask that you consider if St X or St X's is the most appropriate form, when creating further cats in future - not just use St X. We might not agree on whether "St X" is valid, but I think we are in agreement than "St X's" is?--Nilfanion (talk) 14:01, 22 October 2011 (UTC)

That would be a big change with thousands if not tens of thousands of edits, so it sounds a rather timewasting thing to me, especially with thousands of uncategorised churches where one could be creating a category instead. Though presumably a bot could do the donkey work. If there were consensus to enforce that as a particular house style then this would be a predictable subsequent step. Provided you leave category redirects behind I would have no strong objection, but it does undermine one of the values of Commons if we rename such things. I know a lot of churches have links to their Flikr page on their website, and I've been talking to the new UK CEO as to how we could encourage them to do the same with Commons, so the last thing I want is instability in these category names. Oh and yes I agree that "St X's" is a valid format which would work for most RC and CofE churches - I'm just reluctant to use it myself because of what I've seen happen with the geograph and other files, but where others have created a category that way I will happily add more files to it. Presumably you have no objection to the formats "Holy Trinity, placename" and "All Saints, placename"? WereSpielChequers (talk) 14:36, 22 October 2011 (UTC)
Actually the scale isn't that bad, which is the major reason why I want to tackle this now rather than wait for the completion of the good work in creating church categories. I'm only proposing changing one type (the "St X" style to be "St X's") as this is the only one I have a real objection to (due to it not being "correct"), the rest just relating to standardising. The total is only about 700, and some are ok as is. I'd estimate 500-600 moves required which isn't that much. If you are concerned about stability: I'd suggest inserting the word "Church" is probably a good idea, as with it the category is never going to be ambiguous, without it there is a risk of it being so. Its easier to just create with at that point than investigate to see if there is a school, district etc that could cause a name clash. the most common style looks like St X's Church in any case.
And yes, no objection to Holy Trinity, All Saints etc. The apostrophe problem with Geograph is unfortunate, but that's because certain Geograph users used a non-ASCII apostrophe (but better typographically) and GeographBot mangled it on upload. A regular apostrophe that you type on the keyboard with the ' key will cause no more problem than a letter key.
By the way, I know we seem to be disagreeing a lot here - but I certainly applaud your viewpoint (barriers to editors etc) and wish there was more of your common sense on Commons. Just look at Category:Ships by name and ask why Category:Titanic (ship, 1912). No new editor in their right mind would be using that convention!--Nilfanion (talk) 15:22, 22 October 2011 (UTC)
OK if the system isn't going to screw up my apostrophes I'll default to using them in church names, though no guarantees as I often just follow the Geograph photographers. 500 to 600 categories is an awful lot of images, as some churches have dozens of images and even separate sub cats for churchyard interior and stained glass though if you've got a tool that moves the images as well then that would all follow the same rule. Providing you are moving them and leaving a redirect behind then I've no objection, and the ones that I've done recently don't need a redirect but I'd be uncomfortable if longstanding ones didn't as they might be being used by someone - especially if they've only created one or two cats. BTW if you can move categories, I was thinking of moving Orpington to Orpington (chicken breed) and turning Orpington into a dab page as half the category are wrong. WereSpielChequers (talk) 15:41, 22 October 2011 (UTC)
Yeah that's fair enough. Besides if the Geograph user has called it plain "St Mary" that's an indicator that its one of those churches that is plain "St Mary" in any case. As for tools, the combination of User:CommonsDelinker/commands, Cat-a-Lot and AWB means I can do this reasonably easily. The time consuming bit is making sure I only move the ones that need moving! Orpington is a mess I agree, I'll get that sorted out tonight. Its not possible to move categories, but you can move the content.--Nilfanion (talk) 15:56, 22 October 2011 (UTC)

A church near you[edit]

I've started to do moves, predictably finding some teething problems. First is a nuisance: The C of E church directory ( lists churches at "St X, town", even when its really emphatically obvious that it is St X's (St Andrew's, Plymouth for example).

Second is more serious: When adding the placename, please ensure it matches the location's own category. If the location is itself disambiguated, by addition of the county, this should also be included in the church category. It may seem redundant, but without it the category can become confusing (Category:St Nectan, Stoke sounds like its in Stoke-on-Trent to me), or worse, could still be ambiguous: Category:St Andrew, Brockley was for the church in Brockley, Suffolk. There is also a URC St Andrew's Church in Brockley, London (and several other organisations there called "St Andrew's"). The last example does not directly relate to you - but do bear this in mind when creating church cats. I probably should mention this stuff to the other editors involved in these categories; any idea who?--Nilfanion (talk) 19:16, 23 October 2011 (UTC)

If one can't place faith in "A church near you" then we do have a problem as I've been using them today. I'm aware of the geographic problem, but I think one can combine this with the church - there are very few St Nectan's so that should be quite safe. "St Mary's, Stoke" or "All Saints, Newcastle" would be bound to cause confusion. As for general discussions re churches we could post something on the Commons mailing list, or create a project for UK churches. User:Scillystuff is certainly involved in the area and created lots of the patron saint by county categories that I'm populating. WereSpielChequers (talk) 19:30, 23 October 2011 (UTC)
The problem with A church near you is it lists everything as St X. This is because the formal name is "The Church of St X", even when its completely clear that the church itself, the diocese, its congregation, local press... all refer to it as St X's. As for the geographic issue, like I said - it may seem redundant but there's no harm, and is a real benefit. It won't affect HotCat etc as the additional disambiguation is at the end. I'm all for short categories, but when you shorten too far you risk ambiguity again. Its not practical to ensure that there is no other "St Peter's, Tiverton" before creating it as the church cat.--Nilfanion (talk) 19:37, 23 October 2011 (UTC)
well I think you've found the source of your missing apostrophes! In the meantime I'm having fun disambiguating Stretton - I've got a pair of duplicates in west Cheshire and the core cat needs to become a dab page. WereSpielChequers (talk) 20:53, 23 October 2011 (UTC)
Yeah certainly looks like it, I'll contact them tomorrow about this, and see what they think. Regarding Stretton - yuck! Suggest you follow en.wp for disambiguation there - the villages are probably mixed up?
On a different note, are you aware of the info at ? With practice, you can find the older churches and most of the info we need for categories. For instance this tells me St George is Grade I listed, built in the 14th C of slate rubble...--Nilfanion (talk) 21:06, 23 October 2011 (UTC)
Thanks, my focus was on getting the churches categorised, listing, architects, denomination and construction material are all things one could add, and where one is on one image I put it on the category for the church. But if I'm focussed at all it on organising what we have. BTW category:St Nectan's Church, Hartland, Devon is the wrong name as the church is in the adjacent hamlet of Stoke, Devon. WereSpielChequers (talk) 22:01, 23 October 2011 (UTC)
BTW would you mind splatting Category:Christ churches in the Greater Manchester for me? Thanks WereSpielChequers (talk) 00:54, 24 October 2011 (UTC)
You can have a passing editor splat it for you, just add {{badname|Category:Christ churches in Greater Manchester}} and someone will delete it, sometimes only minutes later. Scillystuff (talk) 11:30, 24 October 2011 (UTC)
Passing admin, not editor (its gone) ;) On topic of deletions, there are a number of category redirects floating about. I'd like to get rid of as many as practical. If a category is redirected the day of its creation it should be deleted, if its lasted for a year it should stay.. Any thoughts on where the line is?--Nilfanion (talk) 12:28, 24 October 2011 (UTC)
I wouldn't say that it was as simple as that. I created one today by mistake, but it usefully establishes that a particular building is variously known as both a chapel and a church. I suppose if one must cull them then you could search the web and see if anyone has posted it as a link on their website. But if they are harmless then it is probably safest to let them be. There's also the issue of attribution which gets somewhat mucked up when such things are deleted - though there the answer is to get "move category" as an admin right, I'm not bothered if you delete any I've created because of my mistakes and I'd be surprised if any such were being linked to by anyone, I doubt if others would be worried if you deleted category redirects which had a "whoops" in their edit summary and were hastily turned into category redirects the same day they were created. But I'd worry if you delete one created by someone with few edits or a very local focus as they may well be using it as a link or encouraging others to do so. I might raise the move category business in Bugzilla. WereSpielChequers (talk) 12:42, 24 October 2011 (UTC)
Yep, its complex (sometimes 3 year old categories should be deleted because the redirect is downright misleading) - I'd only hit the delete button if I think its better to remove than keep it.
As for the other bit: I'm sure there are lots of requests on bugzilla about moving categories about (its high on the wishlist); the problem is if you could move a category, you aren't moving any of the content - which would still be in the old category. That said being able to move a category page like any other has definite benefits: It ensures proper attribution of the content and means any watchlists are adjusted accordingly. Those two factors alone are enough to suggest a "category move lite" (which moves the page without worrying about what happens to pages placed in the category) would be a good idea.--Nilfanion (talk) 12:54, 24 October 2011 (UTC)
One reason why I'd restrict it to admins is that I'd default it to amending all the pages currently in the category...... WereSpielChequers (talk) 12:57, 24 October 2011 (UTC)

Village signs[edit]

FYI I just reverted a couple of your moves from "Signs in Devon" to "Village signs in Devon" -[4] and [5]. Reason is simple enough - neither Plymouth or Teignmouth is a village :) I'd imagine similar may apply to other recent edits--Nilfanion (talk) 22:13, 22 December 2011 (UTC)

Unsure about a recent category[edit]

You recently moved a bunch of stuff from Category:Wiggenhall St Mary Magdalen into Category:Magdalen, Norfolk, and made the latter a subcategory of a former. I'm unsure about this. As I understand it (as someone who lives less than a mile from the place), "Magdalen" is simply the short name for "Wiggenhall St. Mary Magdalen". Wiggenhall St. Mary Magdalen is a village; civil parishes can be villages (or towns, or cities). Is there something you know which could contradict my understanding? Lewis Collard! (lol, internet) 18:35, 26 January 2012 (UTC) :)

Hi, I'm not there myself but I looked up on maps and it seemed to me that the civil parish had two villages each with its own church. They have a village sign for the civil parish but also signs for Magdalen. WereSpielChequers (talk) 18:42, 26 January 2012 (UTC)
Not just a short name
Two churches? If we're talking Church of England churches, I only know of one, current or historic. There is indeed a sign in the village marked "Wiggenhall St Mary Magdalen", within the village itself (inside the area that modern road signs indicate as "Magdalen"!), but that doesn't prove much one way or the other. I live in Watlington, which is a civil parish, and also a village. I haven't found any map that has Magdalen and Wiggenhall St Mary Magdalen as distinct entities. (Unless you search Google Maps by place name, in which case it puts Magdalen in a field four hundred metres from any buildings.) Lewis Collard! (lol, internet) 05:23, 27 January 2012 (UTC)
Some further support:
"Over the years, Wiggenhall St Mary Magdalene has become a bit of a mouthful, particularly given the presence of the nearby parish of Wiggenhall St Mary, where there is no real village. So, Magdalen it has become, incidentally giving its name to a railway station on the Cambridge to Kings Lynn line."' -- Norfolk Churches (the railway station, in case you wondered, is Magdalen Road, now known as Watlington)
"Wiggenhall St Mary Magdalen is about 6 miles S.S.W. of King's Lynn. It is also called just Magdalen." GENUKI
"The hall, which is an imposing property with chimneys and constellations lies on the outskirts of Magdalen (as it is known locally)." -- Literary Norfolk
Again, though, I'm happy to see any evidence to the contrary. Lewis Collard! (lol, internet) 05:59, 27 January 2012 (UTC)
OK so if it is two names in use for the same place we just need a redirect. Done. WereSpielChequers (talk) 07:19, 27 January 2012 (UTC)
OK. :) I don't like that you had to undo your work, but that happens sometimes. And that was an easy thing to be confused about; anyone else who didn't live here would probably assume what you did, too. And for what it's worth, I've definitely seen you on my watchlist a lot of times and I know you're doing great (and mostly thankless) work. So thanks. :) Lewis Collard! (lol, internet) 12:47, 27 January 2012 (UTC)
Thanks, actually I'm happy to have my mistakes pointed out, on this occasion I took a little convincing. but as I use both hotcat and catalot fixing that took very little time. And at the end we do now have a category redirect for Magdalen, Norfolk so others are less likely to be confused. The surprising thing is how much one can do here without any feedback, I find Wikipedia rather more "noisy" though I suppose categorisation is pretty uncontentious there as well. WereSpielChequers (talk) 12:54, 27 January 2012 (UTC)
:) Lewis Collard! (lol, internet) 15:39, 27 January 2012 (UTC)

I'm not stalking you, honest![edit]

But after the meeting at the weekend, I thought I'd go and have a look at some of the Geograph stuff, and User:Rillke was kind enough to write me a script to remove {{Check categories-Geograph}} when using HotCat on the image, and to provide a button to press to more easily remove the template manually. The details are at User talk:HJ Mitchell#Check-categories-autoremoval when categorizing with HotCat if you;re interested. Best, HJ Mitchell | Penny for your thoughts? 23:54, 22 February 2012 (UTC)

Skidby, maybe others[edit]

I see what you did here - for some reason cat-a-lot didn't remove the "needing categories" tag when I categorised the image..

Even now it still has the tag - though it's been categorised 3 times..

If you are going to move images from grid squares to places (which I understand) is there a way you can check to see if the page has already been categorised first.? Hot-cat only removes the tag some of the time and there is no way I can work and manually remove the template every time...

Possibly you could (in the future) move the lot to "Category:Images in grid square XXXX" and add that category to the relavent town and village categories.. ? Mddkpp (talk) 00:15, 24 February 2012 (UTC)

Summarise in other words - why not put Category:Images from the Geograph British Isles project needing categories in grid TA0133 as a subcat of Category:Skidby instead of the block move - then leave it to the peons (me!) to move the images out into the right holes - that would be really helpful - at least for me. Mddkpp (talk) 00:19, 24 February 2012 (UTC)

Hi mddkpp, sorry if that caused you work, I do quite a bit of recategorisation and I hope that such incidents are rare. As I understand it Catalot won't remove the uncategorised template if the image already has a non-hidden category. Hotcat will and will also give you a nifty option to confirm that an image is categorised. But you need to load up a line of js code from User_talk:HJ_Mitchell#Check-categories-autoremoval_when_categorizing_with_HotCat. I've been using it for the last couple of days and it works well. WereSpielChequers (talk) 01:01, 24 February 2012 (UTC)

Re: File:All Saints' church, Malborough - - 1076721.jpg[edit]

Could you please explain this removal of a category? Kiko4564 (talk) 19:10, 5 March 2012 (UTC)

Sure. its redundant as I've put the image into the category:All Saints, Malborough. Cheers WereSpielChequers (talk) 19:12, 5 March 2012 (UTC)

CheckCategories auto-removal-script now included in HotCat[edit]

Hello WereSpielChequers, thanks for testing User:Rillke/checkCat.js. This script is now available by default with HotCat. Please remove this line from your User:WereSpielChequers/common.js. When now categorizing with HotCat and a {{check categories}} is found, script will prompt what to do. To restore the old behavior without prompting, simply click on Don't ask again and then on Remove {{check categories}} when editing using HotCat without prompting, wait until the yellow box disappears and confirm. That's all.

Kind regards RE rillke questions? 12:31, 21 March 2012 (UTC)

Denbury and Torbryan[edit]


As I'm sure you've noticed I've reverted several of your edits to content in Category:Denbury and Torbryan. This civil parish is slightly unusual, in that it doesn't have the name as a core settlement, but there are two distinct villages.

Category:Denbury and Category:Torbryan, Devon should only contain media relating to the villages themselves. So some files like File:Cider mill (^) - - 142724.jpg do not relate to either village, but still belong in the CP's cat. It won't be possible to empty the CP cat into the two village cats.

And if something is "near" somewhere, it doesn't mean its of that place - don't trust that word. File:Copse near Broadhempston - - 40503.jpg may be marginally closer to Broadhempston than Torbryan, but everything in that photograph is in Denbury and Torbryan CP. Thankfully, its easy enough to compare Geograph images to OS maps and the borders.--Nilfanion (talk) 09:35, 23 April 2012 (UTC)

OK, that makes sense. I take it your happy that I created categories for Torbryan and the churches? WereSpielChequers (talk) 15:11, 23 April 2012 (UTC)
Yep, of course no issues with the new categories :) I'm half-inclined to create separate categories for parish and village, even when they share the same name, as I can see benefits to that. But the time and effort involved is prohibitive...--Nilfanion (talk) 21:04, 23 April 2012 (UTC)

Churches in Nottinghamshire[edit]


I see you have changed some of the new category entries I created removing Category:Churches in Nottinghamshire, and adding Category:Saint Laurence churches in Nottinghamshire. I can see the advantage in the new category but I don't see why it was necessary to remove the Category:Churches in Nottinghamshire. It isn't now possible to easily see which villages in Nottinghamshire have entries for their churches (unless you know the patron saint of the church). Would you agree to reinstatement of the Category:Churches in Nottinghamshire? Andrewrabbott (talk) 11:56, 24 May 2012 (UTC)

Hi Andrew, the difficulty of including things in a higher level category is that such categories can become cluttered to the point of being unusable. So as I understand it, it is good practice and possibly even policy not to include things in the parent category as well as a sub category. At the moment only a few dozen churches in Nottinghamshire have categories, so including them in the parent category wouldn't currently make it unusable. But if the categorisation gets to the same level as say Norfolk, and if you follow the default display of 200 items, you would have to page through multiple screens to get to the subcategory that you want. Hence the norm for English counties is to have subcategories for churches by religion, by patron saint, by century and so forth. All I've done with those categories is to move them to the patron saint, redirect a couple of duplicates, add a few other images and where I can I've added the location. If I find any in a listed building or denomination category then I usually move that to the church category. Hopefully others will in future add more of the denomination, conservation and architectural categories. As for making it easy for people to find and use the categories, I would suggest that adding the locality to the category makes it easier for people to find them. As for finding villages that don't yet have categorised churches, I'd suggest either going to category:Towns and villages in Nottinghamshire and looking for villages without subcategories (I've just done this and created Category:All Saints, Babworth - very quick and easy if you install both hotcat and Cat a lot). Or looking at the category:Churches in Nottinghamshire by patron saint and look at the subcategories with uncategorised files. WereSpielChequers (talk) 12:59, 24 May 2012 (UTC)
PS One of the advantages of looking via the village or the saint and county is that very often multiple images of the church are grouped together. I usually don't bother to create a category unless we have three images already. WereSpielChequers (talk) 13:41, 24 May 2012 (UTC)

Fire Stations[edit]

Thanks for your excellent session yesterday. I looked around a bit on WP and there are pages for all the fire service authorities. While many of these are for areas which follow the county boundaries, in a number of cases this is not the case. My view is that the firsetations generally have a county tag, it would make sense to re-organise the "Firestations in xxxshire" into "Fire Stations under the xxxx Fire and Rescue Service. Then appropriate galleries can be put in place on the relevant pages. What do you think?Leutha (talk) 12:22, 29 June 2012 (UTC)

Hi Leutha, one difficult thing there is that the internal workings of the Fire Service may not be obvious to someone taking photographs of Fire Stations. Another is that a former Fire station in one county may have been a library since long before that part of the county was annexed into a neighbouring fire service. A third is that the emergency services might themselves get reorganised. We have a similar situation with churches and diocese boundaries. One working compromise which will keep things in places where people expect them is to have a few categories for anomalies "Fire stations in Barsetshire that are under the Muppetshire Fire service", another is to have separate categories for fire services where they don't overlap. That said I picked fire stations for my training session because it seemed an unloved category - if you want to reorganise it I may use hill forts or market crosses for my next one :). WereSpielChequers (talk) 12:33, 29 June 2012 (UTC)

OK, i don't think this quite grabs me. Getting it down below 200 images was enough excitment to last me a while.Leutha (talk) 19:22, 2 July 2012 (UTC)


As I have flagged it for more feedback on the Village pump, you may want to comment at Commons:Bots/Requests#Faebot or even make some suggestions on how to word the scope of the request better to address issues raised by others about the open-ended-ness of the bot (esp. for future work on Geograph categorization). Cheers -- (talk) 10:59, 5 August 2012 (UTC)

Geograph imports tidy-up[edit]

I have proposed an automated tidy-up of imports from Geograph with unreadable HTML data - see Commons:Bots/Work_requests#Proposal_for_Geograph_raw_HTML_tidy-up. If you know of others who may have a view of what is important to preserve in the imports, please copy them this link. Cheers -- (talk) 17:43, 5 September 2012 (UTC)

Adding month and year in the UK to Geograph images[edit]

As a small test, I have populated Category:1955 in the United Kingdom by month using Faebot going through all Geograph uploads. I can run this for a year at a time (the script takes several hours to run through my off-line copy of the Commons xml index each time, so it's not a fast process). In this case 98 images were newly added by the bot. I am currently running 1956 but wondered if you could take a look and if your intuition tells you that the results look accurate or if images have been missed somewhere. The fact that only one photographer has been matched - he is one of the older Geograph contributors, so the images were taken when he was a teenager - may well be correct. Thanks -- (talk) 12:04, 29 September 2012 (UTC)

Hi Fae, I had a quck look. The first thing that struck me is that you have annexed Ireland back to the UK. I've no objection to that myself, provided you first clear it with our Southern Irish editors. WereSpielChequers (talk) 12:09, 29 September 2012 (UTC)
LOL, good catch. I have cancelled my run for 1956 until this is cleared up, I don't want to cause a political argument! Could you point me to any examples you have spotted and I'll ponder how to make Faebot more intelligent. Cheers -- (talk) 12:23, 29 September 2012 (UTC)
Check out Category:1955 in Ireland - I only checked the one month, the other months could have the same problem. WereSpielChequers (talk) 12:27, 29 September 2012 (UTC)

Okay, I'm doing a test run on 1956, hopefully this will give me some results to look at tomorrow and I will probably put the example list in a sandbox if you want to take a look. FYI I have added a filter on Categories of:

ireland=r"\[\[[Cc]ategory:[^\]]*\b(Cork|Dublin|Limerick|Waterford|Kilkenny|Galway|Fingal|Laoghaire-Rathdown|\ Wicklow|Wexford|Carlow|Kildare|Meath|Louth|Monaghan|Cavan|Longford|Westmeath|Offaly|Laois|Waterford|Cork|Kerry|\ Tipperary|Clare|Galway|Mayo|Roscommon|Sligo|Leitrim|Donegal|Ireland)\b"

Hopefully this will catch all the Southern Ireland matches (it may skip others too, but near enough without driving me mad). I did a small test on images currently under 1955 but nothing popped up, I may have cocked up the test but only so much time in the day to keep doubting myself. :-) If I can get this working adequately, it would be pretty cool to have, say, 1950 through to 1980 indexed this way. I suspect anything later than that might be large and a bit more unpredictable, but we'll see, no hurry.

I may be free to discuss other matters in the next couple of days, but still limited right at the moment. Cheers -- (talk) 22:25, 29 September 2012 (UTC)

I'm afraid this will run into problems. It ought to capture those in county categories or the "X in County/Ireland" cats. However it cannot detect those files that don't mention localities. For example, File:Thomas Street Entrance to Guinness - - 303595.jpg is only in "Guinness Brewery".
A more limited run could be viable, which will also get more specific info at the same time. If the file is in a county cat or X in county cat put it in "<county> by month year". That should avoid the Britain/Ireland mistakes and get more detailed cats at the same time.
By the way isn't that regex liable to capture "Northern Ireland"?--Nilfanion (talk) 22:51, 29 September 2012 (UTC)
Guinness Brewery is a sub category of buildings in Dublin, and while I don't understand the code Fae has listed I would hope that it would be possible to exclude everything in Republic of Ireland and all its subcategories. However whilst that would solve the guinness brewery problem it wouldn't help with those images that are categorised on a completely non geographic basis such as species of plants and more pertinately our significant backlog of uncategorised geograph images from south of the border.. I think that the only safe way to do this would be to use the Geocodes that all these images have, obviously you wouldn't be able to follow the border exactly, but if you err on the side of caution and exclude part of Northern Ireland as well that should work. Perhaps I should put in writing my verbal suggestion at the GLAM wiki conference that the UK chapter invest in some boundary files from the ordnance survey or even open street map, that would make it straightforward to at least put everything from the Geograph into the correct area thus solving this problem and dealing with many of the confusing duplicate villages that I've been resolving on commons. WereSpielChequers (talk) 23:15, 29 September 2012 (UTC)
Hm, I may have to go back to the drawing board. With a more complex way of going about this, by doing more than a simple regex, but instead pulling out the geotags like {{Location dec|51.151907|-0.967229|heading:22}} and looking at it matching a couple of relevant boundaries, my filter would be 100% accurate rather than assuming so much about inconsistently placed categories. I haven't done quite this sort of thing in Python before, but it is probably worth delaying this categorization problem and trying to do this as having these skills would probably provide some quite nice future little categorization projects for Faebot. Anyway, I'll let my test for 1956 finish, it might be okay for what it is. Cheers -- (talk) 23:26, 29 September 2012 (UTC)
It won't be 100%, but hopefully the anomalies would be ones you could attribute to geocodes being incorrect:) As for the 1956 run, if it is similar to 1955 then the easiest thing is to create a geograph 1956 category and then go through manually sorting them into countries. That's my preferred use of bots - go through a huge list and produce a much shorter pprospect list for manual use. WereSpielChequers (talk) 23:36, 29 September 2012 (UTC)
Oh, that sounds easy peasy. How about 1940 to 1970 being done by hidden categories of the form [category:month year Geograph images] which would be nested under the [category:year Geograph images]? I could kick off something like that in a day or two, and ponder the geotag template method in some backburner moments. :-) -- (talk) 23:43, 29 September 2012 (UTC)
BTW I'm not volunteering to go through those myself - I've got several subjects to go through that I find more interesting. But if te batch that I did go through are representative it shouldn't take much work, though presumably each year will be more time consuming as you move forward. WereSpielChequers (talk) 23:48, 29 September 2012 (UTC)
Don't worry mate, I have loads of time now that so much is being taken off my shoulders. :-) -- (talk) 23:50, 29 September 2012 (UTC)
Okay, I have started with Category:1955 Geograph images which is being populated now. If that looks okay, I'll do the rest of the 1950s by year only. After that I'll think about the 1960s which may need a month level breakdown. -- (talk) 15:02, 30 September 2012 (UTC)
(I love edit conflicts!) With regards to exploiting the geocoding and boundary data, OS OpenData is freely available and gets us down to Civil Parish level in GB. I used that for the maps that most UK articles now use. What's missing is a proper lookup that a script can use. MySociety's MapIt project provides exactly the look-up service, but not in a bot-friendly form (due to throttling): For example it gets these areas for this image.
WMUK cooperation with MySociety may get allow us to use their API quickly. If so, that would solve almost most localisation issues in GB at a stroke. The errors should be restricted to cross-border images (if the subject is in one country, the camera in another - the geocoding is for camera location, categorisation should be for subject location), and incorrect geocoding by Geograph users.--Nilfanion (talk) 23:51, 29 September 2012 (UTC)
Thanks Nilfanion, Fae maybe this is something you could float by your board? If Mysociety need to throttle then maybe they could be persuaded to let us mirror their data onto toolserver - and if it was there that should speed up the IO on some of the processes that run there. WereSpielChequers (talk) 00:09, 30 September 2012 (UTC)
No problem, though I suggest a volunteer shape up the idea by having an initial conversation with MySociety to sound it out and perhaps confirm with the Toolserver folks that there is no particular reason why it should not be done. It looks like this is not a question of money, and the trustees should do more to empower volunteers to handle relationships with potential partners, as good working practice. Now, all you have to do is find the right volunteer. ;-) -- (talk) 08:00, 30 September 2012 (UTC)
Contacting MySociety may not be necessary here. MapIt is an open source project, with the source on github and detailed installation instructions at [6]. Their public installation is understandably throttled, but it ought to be easy enough to set-up our own, either locally for a botop or on toolserver. Only reason we'd need to contact is if tech support is required.
That could just leave the coding so the bot can call the lookup service and extract the info it needs from the output. Note that none of this is any help outside England, Scotland and Wales...--Nilfanion (talk) 11:04, 30 September 2012 (UTC)

Dealing with licenses for older Geograph images[edit]

Hi, my categorization by date has thrown up an unexpected problem. I have now gone as a far back as the 1930s (just started running that script yesterday). I had thought that the result would be nil, however you can see that photos dated as made in 1930 have been listed at Category:1930 Geograph images. There are not that many different Geograph photographers involved, and it might be straightforward to confirm the background of the licenses for each author. I did check the background of some 1950s photos and it was entirely credibly that the photographer involved was more that 70 years old, took these himself and in his profile mentioned taking early photos as a teenager after the war. However this cannot apply to the 1930s photos which seem more like family album photos that some Geograph members have uploaded. Would you like to take a look? My time is limited but we should probably take the initiative to sort these out rather than having them forced to deletion review first. More photos will be added through 1930 to 1950 over the next few days, so it may be as well to lay the process to assess the credibility of the licenses on these fairly promptly. Cheers -- (talk) 10:44, 6 October 2012 (UTC)

At least one of them is credited to an individual other than the Geographer, so methinks they have a process in place for this. I've dropped them an Email, but this highlights another migration issue. Would it be possible for you to produce a list or hidden cat for images only in a subcategory of Category:United Kingdom by decade, or better still migrate the poorly categorised ones to category:Otherwise uncategorised United Kingdom by decade? It is a bit of a shame if we have no other categories for these images. WereSpielChequers (talk) 13:06, 6 October 2012 (UTC)
It can be done, but don't we have the problem of then falsely categorizing images in Ireland as in the United Kingdom? Could you clarify if you are suggesting a decade parent cat (easy, started setting this up now) or whether you are looking for the uncategorized images to be automatically spotted (harder) or perhaps your expectation is something else? Cheers -- (talk) 10:04, 7 October 2012 (UTC)
I don't know how the original Geograph categorisation run was done, unlike in your test I didn't see any obvious Irish ones so it is possible that the original run was OK. But even if not you wouldn't be classifying them as UK - you'd be re categorising a bunch of images already classified as UK so that they were available for manual review. If there aren't very many of them I may go through them myself - I'll certainly go through a few dozen. So if there are any Irish ones in there you are making them more likely to be corrected. WereSpielChequers (talk) 18:37, 7 October 2012 (UTC)

Finding place addresses using JSON calls[edit]

I have played around a little bit more in Python with the above suggestions today and managed to squeeze the following little test run using Google Maps:

Getting Category:1945 Geograph images list...
File:Duntulm Castle - Caisteal Dhùn Tuilm - - 688945.jpg
Extracted coordinates: 57.683067 , -6.347740
Formatted address from Google maps: A855, Portree, Highland IV51, UK
File:Ferryhill Market Square 1945 - - 88616.jpg
Extracted coordinates: 54.68675 , -1.55944
Formatted address from Google maps: 6 Neale St, Ferryhill, County Durham DL17 8NB, UK
File:Interior of St Bridget's Church, Bridgerule - - 685467.jpg
Extracted coordinates: 50.802445 , -4.440310
Formatted address from Google maps: Under Rd, Holsworthy, Devon EX22, UK
File:Selborne Road - Victory Street Party Sep 1, 1945 - - 447551.jpg
Extracted coordinates: 51.377607 , 1.402905
Formatted address from Google maps: 14 Selborne Rd, Margate, Kent CT9 3SP, UK
File:St Bridget's Church and Church Cottage - - 1340809.jpg
Extracted coordinates: 50.802277 , -4.439734
Formatted address from Google maps: Under Rd, Holsworthy, Devon EX22, UK

I could not get this to work using the Mapit project, but maybe Google maps will be sufficient for the moment. Google allows a maximum of 2,500 api calls in a day for this service, but that would be pretty cool if we have specific parts of Geograph that need addresses adding. There is no easy mapping for long addresses into Commons categories, but even if this were an extra field in the description text, it gives very useful information such as county, postcode or town for users to make easy use of (or for later bot scripts to use too).

Left a message back on my talk page for your other problem. -- (talk) 21:47, 18 October 2012 (UTC)

Thanks for this, 2,500 a day adds up to a lot over a year or two. It isn't a magic wand but it is pretty close. WereSpielChequers (talk) 22:07, 18 October 2012 (UTC)
I am trying an addressing test run on Category:1969 Geograph images (97 images). Take a look when you have a moment. I will probably raise this test run for discussion and suggestions at Commons:Bots/Work requests before trying a large run. There will probably be some ways of using microformats that I am not aware of and we are likely to find ways of it falling over with odd use of the Information template. No hurry, this is a nice back burner job for Faebot. -- (talk) 05:45, 19 October 2012 (UTC)
Note, Faebot will skip images with no Location dec template or those where Google returns a short address like "United Kingdom". -- (talk) 06:12, 19 October 2012 (UTC)
I hadn't appreciated that this was just going to add an address without improving the categories, or was that just because those were all already fully categorised? I'd be inclined not to go for a house name or number as it looks a bit creepy to me. Also there will be some people writing to that address which may not be correct or geared up for post. For example File:Clayton St. John the Baptist - - 368471.jpg Like many churches if you google them they don't give that address for postal correspondence. There may also be licensing issues with the Royal Mail if you were to start a process that effectively extracted the whole Postal address file at address level. WereSpielChequers (talk) 08:25, 19 October 2012 (UTC)
I don't know of a way of checking a standard address against what might be available as categories on Commons, as there is no standardization, it would be hard to even predict how categories with names of towns would be represented - at the end of the day, having the location in words is an aid to the categorizer to quickly see if road or town is missing (for example Category:Grand Avenue is a musical group rather than a road in Cardiff). I agree with the house number thing, and I have tweaked the script to take these off and Faebot is trimming house numbers now... even if not creepy, it gives a false sense of accuracy which might be misleading. I'll play further with Google's data to see if it is a bit more intelligent. There would potentially be licensing issues if we were creating a postcode database or a postcode map, but in this case there is no system apart from looking up a standard address from Google one point at a time, so I believe that is not a concern. Testing is throwing up other encoding problems, so this will take more than a week or two anyway. -- (talk) 09:47, 19 October 2012 (UTC)
I don't know of a way of checking a standard address against what might be available as categories on Commons, as there is no standardization, it would be hard to even predict how categories with names of towns would be represented - at the end of the day, having the location in words is an aid to the categorizer to quickly see if road or town is missing. I agree with the house number thing, and I have tweaked the script to take these off and Faebot is trimming house numbers now... even if not creepy, it gives a false sense of accuracy which might be misleading. I'll play further with Google's data to see if it is a bit more intelligent. There would potentially be licensing issues if we were creating a postcode database or a postcode map, but in this case there is no system apart from looking up a standard address from Google one point at a time, so I believe that is not a concern. Testing is throwing up other encoding problems, so this will take more than a week or two anyway. -- (talk) 09:47, 19 October 2012 (UTC)
OK that makes sense, Most UK residential postcodes cover up to a hundred addresses within one street - streets with more than 100 letterboxes eventually get subdivided so postcodes with over 100 properties are rare. About a tenth of UK Postcodes are unique residences, usually farms and so forth that have no other homes on their road, so for categorisation purposes you could drop the two letters from the end of the Postcode, the first half of the Postcode in combination with the number that leads the inward half usually equate to about two thousand homes, though things get different in business areas. Alternatively we could take the view that these are all publicly released photos, mostly taken in public places, and the grid reference actually positions them more accurately than the full Postcode would. Another issue to remember is that Postcodes are not permanent identifiers, the Royal Mail has a steady process of reviewing and reorganising Postcodes to reflect their structure and convenience. Most of the changes are demolitions and new construction, but they do sometimes re Postcode whole towns, or at least change the first half of the Postcode, so if we are Postcoding our data we are eventually going to get anomalies unless we watch for and perform those updates. On the other hand we have some areas like London where we are using the Outward Postcode as part of a parallel geographic categorisation system. I guess this is all rather messy and we need to think through where we are, where we want to be and how we could get there. In the longterm I'd like to see collaboration with both OSM and the Geoograph so that our metadata is more compatible and all three sites benefit from the same data. Tom Morris is looking at getting Commons cats and possibly even hill forts as objects within OSM, and it would be cool to get contact with the Geograph, though I've yet to get a reply to te Email I sent them several days ago. WereSpielChequers (talk) 10:58, 19 October 2012 (UTC)

Fixing categorisation errors[edit]

  • By the way, with the scripts I now have, it would be entirely possible for me to distinguish whether images were in a particular rough area (a rectangle, other shapes would be more complicated) or distance from a particular point using the existing Location dec data. Consequently I could test if images were in the Orkney Islands or in Wales, though it would be slow, and best if on a relatively small category (i.e. thousands rather than millions). As boundaries are less of an issue for islands, I might try a test to create a problem category like Category:Geograph images geotagged as not in the United Kingdom or Ireland or a positive category such as Category:Geograph images in the Isle of Man (possibly testing for several islands at the same time as this would save a lot of processing power). Separating Northern Ireland from Southern Ireland might be a headache, though if we can rely on Google's data which included political regions and countries, that would be a solution without having to have a detailed database of lat/lon coordinates. Lots of interesting future experiments here. :-) -- (talk) 10:54, 19 October 2012 (UTC)
This would be really useful for certain areas where we know there were cockups in the past Geograph categorisations. For example if all the images categorised in Wigtownshire that are actually in Northern Ireland could be moved to the relevant Northern Ireland county, or if they also have a category that is a subcategory of Northern Ireland we could just strip out Wigtownshire. IOW/Hampshire is another little mess along with the West Sussex part of Greater London. The bigget one is the West Midlands, West Midlands region just needs a few high level links to pull in Herefordshire, Staffs, Shrops, Warks, etc, but there are tens of thousands of images where you see roads in Worcestershire and also roads in West Midlands etc etc. The West Midlands county categories are often contaminated with >50% being other counties with the West Midlands region. WereSpielChequers (talk) 11:02, 19 October 2012 (UTC)
Good practical suggestions. I'll cut my teeth on one of these next week. For the moment the WCA is urgently calling! :-D -- (talk) 11:11, 19 October 2012 (UTC)
Thanks, keep me posted on any results and if I come across further blatant ones where a bot would help I'll tell you. WereSpielChequers (talk) 11:16, 19 October 2012 (UTC)

Potential automatic identification of county[edit]

Okay, this could be interesting. I think counties have been identified by Google Maps as the data type "administrative area level 2". These are probably unique names throughout the UK. I believe that we could usefully add a hidden Geograph category with this as the lowest useful and user friendly regional level that could then be easily used to find incorrectly placed images and recategorize in interesting ways. For images such as this still marked with the Uncategorized-Geograph template, it would seem a no-brainer to add the same district name (Category:Cornwall) as a normal visible category too. I have run two short tests on the following categories:

I have started a discussion at Commons:Bots/Work requests#Adding addresses and regions to Geograph images to get feedback, and potentially some form of consensus, by bot-aware contributors. -- (talk) 00:14, 20 October 2012 (UTC)

Very useful, and the examples I checked looked good. I was wondering if you would be able to use this to get a data feed for one or two projects of mine? The first and hopefully least contentious would be for our hill forts project. I was thinking of an upload of images from the Geograph of images not yet uploaded which contain the word "hillfort" or "hill fort", using your county based routine to add them into the relevant category such category:Hill forts in Somerset. It would be easy to do the final stage of categorisation if they were uploaded by county, and I'd have thought it would be easy to get consensus for such a targetted program. The other would be to assist with some workshops I'm running in London. Rich Farmborough has created a list of geocoded articles in the UK and Ireland which lack images, I'm starting to add images that we already have on commons, and am planning to run a workshop to get more done, but it would be nice if we could first suck in a few thousand images that are in close proximity to those locales. Even if said images were only categorised by county. WereSpielChequers (talk) 15:25, 21 October 2012 (UTC)
Should be do-able. I'm not experienced in batch uploading but even if they were dumped in a large holding category, it would not take long for a script to sort a few thousand into sub-categories by location. I am finding glitches in the OSM data, to compensate I'm using to double check the data (I could use mapit but it only has a rather complex JSON service rather than XML and I found it a bit hard to use, a bit of persistence and a strong coffee in a few weeks time might crack it — I probably need to do this before trying a serious run, as having 3 sources of data means that they can "vote" on the answer; and it may rather confirm that this part of OSM is too flaky to be useful). This is probably a good thing to spend a few weeks of testing and thinking on. As for the London boroughs sorting, if there is any doubt as to location, the exceptions can be consigned to a manual check category (or skipped) so that auto-categorization has a high level of confidence by the Commons community. Similarly for postcodes, though sometimes the mis-matches are between W10 or W11 so I may opt for just looking for a match of the first two alphanumerics and skip if these match.
From testing out the London Boroughs, it seems obvious now that each UK region would need special attention to make sure glitches of naming were ironed out. For example though there are only 32 London boroughs, the names used can vary quite a bit (though with some databases more reliable than others), so a robust solution needs to compensate with some matching of names like "Westminster" to "City of Westminster". So the project looks like an interesting series of projects (running in the background through next year?) with less programming time needed every time we do it. It would be neat to think of doing these auto-categorization exercises a month or two before some major push in that region.
By the way, I am watching my London script test itself out at the moment, and it takes over 15 seconds per page due to time to pull in data from different sources. Not unreasonable, but it would not be possible to randomly filter all 2 million Geograph images. I would guess the script is about half ready to do something serious, though it may take me a bit of work to get it to be robust, considering that internet connections are likely to drop out, timeout or get blocked if there are too many queries in a day. If you can think of a Geograph category that is likely to be London-centric and good source of prospective images, this would be a good start to beta testing.
My first test is at User:Faebot/SandboxL#London Boroughs and Postcodes Test on images in Category:Shops in Greater London and you can see the Boroughs are reasonable, it is only postcodes that are poor enough in OSM to need more thought on better solutions using other on-line databases. Cheers -- (talk) 21:46, 21 October 2012 (UTC)
(Update) See User:Faebot/SandboxL#Test on images in Category:Shops in Greater London - I think I'm ready for a real beta test for creating London borough hidden sub-cats for Geograph. This test report shows how the script checks if this is a Geograph image (so any category could be used as a source and only Geograph images would be touched), checks for Object location data in preference to Camera location data, and how it also checks for data errors with the London borough name in OSM and uses two other websites to check these exceptions. The resulting categorization should be pretty highly reliable. For the moment I suggest any beta test is limited, later on we can revisit the ideas about adding 'suburb', 'postcode' or adding visible rather than hidden categories. -- (talk) 12:34, 22 October 2012 (UTC)
(Further update) Okay, I'm out of beta testing, for London at least it seems highly reliable (even with my internet connection dropping many times in the last few days). I was surprised that the total appears to be less than 3,000 images for images geotagged as in London (around 0.15% of the Geograph uploads?). Though I am also noticing a fair proportion that have no geodata, mostly the older images. While categorization continues for Category:Geograph images in London, I am running a dump of all images tagged as being taken of the Orkney Islands and anything further North to see if that might be an interesting second region. I did think about the Hampshire / IOW problem, and I may make that another early target along with a South West target of Cornwall, Devon plus Isles of Scilly, Guernsey and Jersey (assuming they are in Geograph). Anyway, further details will be at User:Faebot/Geograph#C1 as they happen. -- (talk) 12:34, 24 October 2012 (UTC)

Images from grid SP5361[edit]


You bulk moved the files from Category:Images from the Geograph British Isles project needing categories in grid SP5361 to Category:Staverton, Devon earlier. When they clearly relate to the Staverton in Northants (SP), not the one in Devon (SX).

Wonder if you made any similar errors...?--Nilfanion (talk) 21:04, 2 October 2012 (UTC)

Whoops, well I certainly hope not. There are lots of categories that I've unpicked where we had two or more villages of the same name, so it would be embarassing if I've caused one like that myself. But thanks for telling me about Staverton, I'll finish what I'm doing and go take a look. WereSpielChequers (talk) 21:08, 2 October 2012 (UTC)
Fixed. WereSpielChequers (talk) 13:15, 6 October 2012 (UTC)

Ambiguous concepts on Geograph[edit]

I've been struggling with how to handle a few concepts where a single name has been used on Geograph for multiple (related) concepts, or the reverse where 2 different names are used has been used for almost identical concepts. Some of these are not properly handled by our own category system either

3 examples of what I mean:

  • Garages can refer to
    • A private, household garage
    • A car parts/repair shop
    • A dealership
    • A petrol station

Many dealerships are service centres too, so there's some overlap. The small shops aren't far removed from a household garage either - just using a similar building differently. Petrol stations are more distinct.

  • Gardens
    • Household gardens
    • Show gardens (eg at NT properties)
    • Civic gardens in public spaces

There's a spectrum from a small plot to the Eden Project. The problem is the third - at what point do they become parks? What overlap is there with the other sense of garden?

  • Pubs (reverse problem)
    • Public houses
    • Inns

There's a distinction between the two, but I'm not sure its useful as who knows where the line is? There's no difference between this and this, indeed both are called "Inns", yet they are marked as PH on the 1:50K OS maps. Geograph calls one a pub and the other an inn, reflecting the confusion. I think it might be easiest to just merge the cats. Its easier to separate hotels from pubs that way IMO.

Thoughts?--Nilfanion (talk) 22:36, 23 October 2012 (UTC)

Hi Nilfanion, I don't see these as a geograph problem, I see these as ambiguities in English. I'd be tempted to add the cenotaph/war memorial divide where I think that part of our problem is a regional thing with some northerners regarding every war memorial as a cenotaph regardless of its shape.
As far as our categorisation is concerned I'd be tempted to merge inn and pub as they are in effect the same. I suppose the key difference was whether they had the facility to stable horses, but few now have retained that or did so into the age of mass photography.
For Parks and Gardens I see the difference as pastoral/silviculture v horticulture. If it is mainly grass and trees then its a park, if it has lots of flower beds it is a garden. Some places may fall into both categories, quite a few stately homes have a house, a garden and beyond that a park. If I think there are enough images to be worth subdividing them I have even given a NT property separate sub category for its Park and for its Garden - but I'm aware of some places where that would be over complex. If there were a debate about merging the Park and Garden cats then I might fly by and put a mild preference in for keeping them separate, but I'd be open to persuasion.
Garages in the car maintenance sense of the word, car dealers and petrol stations are a bit of a continuum. I'm ancient and decrepit enough to remember the days before supermarkets sold petrol when lots of car dealers had petrol pumps on their forecourt, as indeed did the garage where I bought my first car. I doubt if I have any pictures of such subjects from that era as they don't interest me, but it is recent enough that some people will have, and from what I've seen of the geograph some of our more isolated but photogenic communities still have such establishments. More to the point there will be some buildings that have served different roles in different eras - I know of dealers that have always had a garage and have now swapped their pumps for more display space. If I cared greatly about them I'd suggest that if they are selling cars they are a car dealer, if they sell petrol they are a petrol station and if they repair cars they are a garage (I'm not sure how best to disambiguate lockup garages from car mechanics premises). But I don't care sufficiently about them to visit let alone participate in an RFC on the motor trade.
On a broader note I think that we need to bear in mind the following:
  1. What can we reasonably expect our categorisors to do? In most case I think that involves adding logical categories based on the image and its metadata. But we need to help our categorisors, and to my mind that means creating category redirects from logical alternatives. Or alternatively we need some smarts into Hotcat so that it treats in and of and pub and inn as synonyms and offers you a probable category.
  2. What can we automate? We don't have sufficient resources to do everything manually, and we have some massively over cluttered categories. Some of this can be resolved by geocode work - Sheep in England or National trust properties in the UK could be bot recategorised down to county level. Some could be resolved by ratcheting, so if an image is in the category Anglican churches in england, Saint Thomas churches in the UK, flint churches and 18th century churches in Leicestershire it would be cool if a bot could change the other categories to "in Leicestershire" if they exist and "in England" if that is the lowest level.
  3. What are the known and or probable uses of our categories? I feel it is more useful to create a category if there is a corresponding Wikipedia article that one can link that category to via commonscat. But I'm also conscious of some of the other possible uses of our data such as the idea of getting Open street map to link not just to Wikipedia articles but where we only have a commonscat to link to that instead. Since no-one knows what all the possible uses are, I would be loathe to merge categories unless we can't make a meaningful difference, such as with pubs and Inns. I generally assume that if someone else started categorising churches by patron saint. denomination or century then my work will be useful to them or whoever shares their need, and I get to look at lots of lovely pictures. WereSpielChequers (talk) 11:59, 24 October 2012 (UTC)
I agree this is not a Geograph-specific problem this but the limitations of English. The Geograph-specific part is that when the Geograph users run into these problems and make different choices as to what category to record there. Any automated capture also has same problem. I suspect the solution on Commons for this sort of thing is to have the cats nested with sensible names. That would handle garages as "garages" would be the generic, with the various types as subcats.
I do think a straight merger of pubs/inns is the best way forward there, so I'll start a CFD on Inns in the UK later and see if there's any other thoughts.
Personally, I tend to work on getting the localisation right and ensuring its in the right topic trees. The location makes bot re-catting easier, and if I get it in the right tree, it can be sorted more fully. For instance if I at least mark a image of church as such, you have a chance of finding it and can add the patron saint etc. I'll probably switch to topic-based work at some point (such as getting NT stuff down to property level, getting listed status recorded).
Over-cluttered cats are probably always going to be a problem. I don't think its that useful to localise Category:Roads in Derbyshire further (except in the towns), yet it has over 2,000 files. Category:Roads in Devon by cross-section is not something I care that much about, but is something I did to filter Devon's roads a bit. Even so, 1,000 images of two-lane roads in Devon? Being selective in galleries might be only way we can give a useful end-service for such topics.--Nilfanion (talk) 12:38, 24 October 2012 (UTC)
This might be a good discussion to have in a workshop with OSM at some point. They have faced the same problems and could, perhaps, add some value here as their volunteers have a clear set of icons/types to choose from built into the interface. Though on Commons we might not want to be so restrictive, we could piggy-back on their choices of classification of places (for example it should be possible to datamine OSM just for the list of "recognized" pubs in the UK with their given names and coordinates and use this as a reality check on some of the image classification; at least for those that are geotagged). -- (talk) 12:41, 24 October 2012 (UTC)
One problem we will have with working with OSM is that they are mapping the present, and we are recording all knowledge. This doesn't hugely matter with medieval churches, castles and other listed buildings, but it is going to be a huge problem with pubs. Most of the Geograph stuff is pretty contemporary, but as we encourage more people to upload their collections so we will get shedloads of images of pubs that have since closed. So I'm happy to work with OSM - we have at leat one regular at the London meetup who is also on OSM. But I'd be cautious about using current OSM info to tag old images. WereSpielChequers (talk) 12:47, 24 October 2012 (UTC)
True, and we can work together in four dimensions quite nicely. For example, if OSM has a pub listed where we have no photo, that sounds like a new project (Wiki Loves Pubs!) or where Commons has a photo from 2011 showing that a pub has been reopened as a trendy cafe, well that sounds like an opportunity for OSM to update their location type and look to create a wishlist of checks needed, possibly using Commons images as a useful overlay... :-) -- (talk) 13:13, 24 October 2012 (UTC)
@Nilfanion. There are some categories that I use as an opportunity to declutter other categories that I find more interesting. Shops are a good example. Shops in Hampshire would be a huge category that I'm not interested in, but subdividing it into Shops in Portsmouth, etc means you have the opportunity to declutter some of the town categories by moving all their images of shops into a Shops in town category. I suspect that in the longterm we will need to integrate image quality into our selections, perhaps with options to highligt quality and grey out low quality images in Commons searches. WereSpielChequers (talk) 12:55, 24 October 2012 (UTC)
Agree there (and shops can be broken down further by type by people who do care about them). Making Category:Portsmouth look like Category:Plymouth takes a lot of work, and categories like Shops in Portsmouth. My problem is with the county-level topic cats, as I generally find rural images more interesting there. Rural images aren't cluttering their location cats (parish cats aren't bad), but the topic cats get flooded with them. By-location subcats help, as they can remove urban images in the towns. However for rural images, parish-level is far too precise to be useful and IMO district-level isn't that valuable. District-level topic cats will still get flooded in any case. Category:Roads in Dartmoor is about district-level but still has 600+ files. If you are looking at a national park, you are almost certainly not bothered about urban images, but taking them out still leaves well over 400. Filtering that down to a usable level requires something other than location subcats; like sensible sub-topics, subjective choices (in a gallery) or something else such as improvements to Mediawiki...
I do think galleries are an extremely helpful but sorely underused aid here (compare Shaldon to Category:Shaldon). But that is always going to be a high maintenance solution, with low interest - just getting cats sorted takes forever!--Nilfanion (talk) 22:25, 24 October 2012 (UTC)
I can't comment on Commons galleries as I've barely touched them, though I've added a fair few to articles on Wikipedia. I'm conscious that Commons is growing fast - though since the Geograph uploading was put on hold I suspect the UK has the slowest growth rate of any territory covered by commons. But this has lead me both to err on the side of overcategorisation, as there will be more images coming in over the next decade or two; and also to be cautous about including the UK in Wiki loves monuments, as we probably already have photos of many if not most UK monuments, and our need is for categorisation and image use in Wikipedia. I recently got Rich Farmborough to give me a list of 1500 Wikipedia articles that have UK geocodes but don't have images, my estimate from a very small sample is that > 1,000 will have suitable images already available. I can think of other queries which should find us far more. I'm going to run a workshop in London to show newbies how to illustrate Wikipedia with these images, and maybe do another editor recruitment exercise using these as a wow. My experience with previous workshops is that people really like adding images to articles and feel very confident deciding which image is best to illustrate a village with. What I'm not sure about is whether I can get them to categorise as they choose the photos. As for the issue of choice, in the long run we will probably be best off with some sort of AI based quality assessment, and at some point we will have a major task of updating which Commons images are actually used on Wikipedia. My prediction is that by the time the bulk standard mobile phones are taking multi Terabyte 3d images, even the best current 100mb Hidef ones will be looking flat and suboptimal, and many of our 60k Geograph images will still be in use..... WereSpielChequers (talk) 15:25, 25 October 2012 (UTC)

Inns in UK[edit]

FYI - Commons:Categories for discussion/2012/10/Category:Inns in the United Kingdom.--Nilfanion (talk) 10:11, 25 October 2012 (UTC)

User rights[edit]

I've added the filemover and rollbacker rights to your account in case they come in handy. INeverCry 21:55, 12 December 2012 (UTC)

Thanks, yes those will sometimes come in useful. WereSpielChequers (talk) 21:57, 12 December 2012 (UTC)

Gold postboxes[edit]

Can I ask what the point is behind categorising all these images by athlete? And why you appear to have stopped half-way through? I'm not seeing a compelling reason why having them all categorised by athlete is any more important than for example being able to browse through them all in one category, certainly not at the expense of cluttering categories like Category:Type A/B post boxes in the United Kingdom with (19 and counting) sub-categories that only relate to gold boxes. You've effectively created individual categories for individual boxes, even though that's not a schema that's going to be implemented for all images of postboxes, for obvious reasons (not that it couldn't be done, it just serves no purpose). I am about to launch an index gallery/table which should make it easier for people to identify boxes and match them with locations/recipients, and after that I also intended to create a gallery for each athlete's box/boxes, before finally ensuring all the descriptions link to the relevant gallery, so I hope you will agree that once that's all done, this level of sub-categorisation would be unneccessary and it can be restored to how it was before. For an example, see Manchester bombing memorial postbox - those images are grouped on a gallery that details why they are special, but the images themselves are categorised like any other box is, by colour/reign/type etc, indexes with the box number (a detail which with this change, is now completely redundant for gold boxes). Otherwise, I'd like to see this discussed at Cfd, because the categorisation of gold postboxes is a lot more complex than you may have realised before you embarked on it - not all boxes were awarded to single athletes (some to groups, some to places, some to nobody at all), and two boxes are not even in the UK, are just two issues that immediately come to mind. Ultra7 (talk) 18:44, 15 December 2012 (UTC)

Hi Ultra7, I don't think I categorised any of them by recipient - someone had already done that. All I did was create categories for the ones that had multiple images. We already had over 200 images of gold UK Post boxes, and of course more will be coming in, and that is too many for one category. So I just did the sort of consolidation that I've been doing with UK churches - if we only have one or two images of a particular postbox then we probably don't need a category for it yet, but some of those postboxes have seven or eight images, and for them it makes sense to create a subcategory. As for Category:Type A/B post boxes in the United Kingdom it might make sense to subdivide that geographically or to create a subcategory for gold ones, currently it has over 700 images even after the consolidation that I did! Oh and the business about stopping halfway through, it is only 24 four hours since I took a break from gold boxes, I'm fairly sure I've got most of the multiple ones but don't consider that I've finished. WereSpielChequers (talk) 19:21, 15 December 2012 (UTC)
I'm not sure what you mean, but I call [7] categorising by recipient, ie Hannah Cockroft. The image has been taken out of the type/reign/colour etc cats, and dumped in a recipient specific cat. Are you saying you didn't create the categories? It looks like it was you to me based on this, although thanks to User:Fae flooding my watchlist creating a user cat, I can't see any further back than a day, if someone else started this before you. Someone else previously split them by Olympic/Paralymic, but that was unwise and I was going to reverse that in time, given that at least one box has been given to both an Olympian and a Paralympian, and it's ambiguous at best whether a few of the other 'group' boxes apply to one or the other. Frankly, 200 images is not far too many if you're not interested in who the recipient was. Imagine how difficult it is now to browse through all the type A/B gold boxes if you're just trying to match the background of an image you have taken, with one on here. Or trying to find the best image even. By my count, there are 63 type A/B gold boxes alone, so do you really think that's helping people by expecting them to open all those categories of 3,6,10... images? Even if there were 1,000 images in a single cat, that can still be done in just 5 mouse-clicks, not 63! Cat:Type A/B may well require diffusion, but that certainly shouldn't be by geography given that would just re-unify one branch with an already existing parallel branch, nor should it be by colour for the same reason, and because 99.99999% of them are red! A proper diffusion of A/B would be further along the lines of type as a defining quality, such as first separating the As and Bs, then maybe sub-dividing by makers, etc. What you've basically done is take a quality like A/B type, and divided it right down to individual boxes straight away. It's not helping anyone except people who only want images of the box given to that recipient, and they already had that ability by looking at the person's own cat, which rarely contain more than 20 images as far as I've seen (Andy Murray aside maybe). At the very least, subdivision by recipient can be restricted to just the scheme branch, without bringing in type/colour/reign too. If you can't understand the issues you've introduced, then I'd ask you to either reverse what you've done so far and take it to Cfd, or (rapidly) complete what you've done and I'll take it to Cfd. There's no point in not categorising single images, as more will eventually come, and in my experience it's not likely that the people who add the 2nd image are going to do it themselves. Ultra7 (talk) 20:04, 15 December 2012 (UTC)
If you look at Category:Hannah Cockroft's gold postbox, Halifax you will see that the name of the recipient is only one of the categories that that category is in. All four of those images are of the same gold postbox and before I created that subcategory all four were in the category for that recipient - as well as many other categories. I've used the name of the recipient as the first part of the category name, so I'm naming the boxes partly by recipient. But no I'm not categorising by recipient - so for example where one person has several gold boxes I'm not creating one category for all their gold postboxes. As I explained before I'm creating subcategories for individual gold postboxes for which we have multiple images. Some of the people with multiple boxes have those boxes in different towns, and of different designs. If I was creating categories for gold postboxes by recipient then I would sometimes be bringing together objects in different counties in the same category, and that isn't what I've been doing. As for how one would subdivide AB boxes, well there are many ways to do that. But we need to be realistic in our categorisation. Geography fits with the information that we have, manufacturer may not be so readily available unless the photographer has included the makers name in the photo. As for how one picks the best photo to use, well clearly browsing 700 thumbnails is not viable. My hope is that we can move in the direction of getting image quality as an overlay in our selections, so someone looking for an image to illustratte the article on goldboxes will find it easy to find any that are featured or good images. As for your concerns about the category structure, beyond a certain point it might be worth creating Category:Gold Type A/B post boxes in the United Kingdom as a subcategory of Category:Type A/B post boxes in the United Kingdom and moving the 19 gold subcategories into it. WereSpielChequers (talk) 20:32, 15 December 2012 (UTC)
I've no idea what you're trying to explain in the first part. The issue is, for each box image you've replaced multiple categories with just one, for no apparent benefit that I can see or that you've explained. In the process you've created many issues, not least wiping out any prospect of general browsing irrespective of recipient, dumping the indexing, effectively merging the gold in the UK/box for event branches without realising they're not completely interchangable, and nyumerous other minor issues that only become apparent once you look at the issue in detail. There are 110+ gold postboxes out there, which in this system means that eventually, there will be 110+ categories, leading to nonsensical situations like having 63 sub-categories in Category:Type A/B post boxes in the United Kingdom, or similarly random numbers in the reign/area/postcode branches, for no other reason than they're gold and belong to athlete XYZ (and zero prospect of any of the red images being sub-categorised in a remotely similar way). All this, in spite of the fact that images were already categorised under the athlete name, meaning most athlete categories will presumably now be empty except for a gold postbox subcat!. It's totally pointless. Browsing 700 images is not difficult at all, and people can already get a good idea of likely quality from the byte size if that's their aim. If you want to create a branch by quality, then do so, but I hope you appreciate how silly it would be to then treat that as yet another sub-category beyond what you've already done, leading to bizarre categories like High-quality Type B gold postboxes in Halifax, containing Hannah Cockroft's gold box cat etc. Having 700 images sub-divided on one meaningful quality like recipient/reign/type/area/postcode in parallel, is certainly 100 times easier than opening 110+ individual categories and leaving people to guess whether it's worth their time or not to open them all. That's 110+ mouse clicks! It's never going to happen. Branches like quality should be parallel, just like type/reign etc used to be. It is beyond me that you cannot seem to understand the issues here, but at the very least if you can't explain what the benefit of this change actually is, I am going to take this to Cfd. Ultra7 (talk) 21:30, 15 December 2012 (UTC)
I'm sorry if you have difficulty following my explanation, could you be more specific as to which parts you do or don't get? As for whether each of the 110 gold boxes will ultimately get their own category, that depends on the number of images that people upload for them. I haven't photographed any of them and am unlikely to do so as we already have pretty good photos of the nearest one to me, but as with say war memorials and photogenic pubs these are photogenic objects and I can foresee lots of photographs turning up in the future. creating categories for those that have multiple images is the logical way to respond to that, otherwise we wind up with cluttered categories. WereSpielChequers (talk) 23:43, 15 December 2012 (UTC)
There will be 110+ categories eventually, that's just obvious. Right now there's only 27 boxes where we have no image at all, that will drop to zero once the summer comes around and free license minded people get to the more remote ones and the ones painted well after the games. I wouldn't expect you to photograph any, the number actually photographed by Commons users is minimal. But you're conversing here with the person who's done the heavy lifting with regard to finding, uploading and indexing all the other images out there so far, so I'd appreciate it if you could respond to my concerns in detail and comprehensively. What I do not get is why you chose to unify all the UK postbox branches: type, reign, area, postcode, colour & event, for the gold boxes, if your only apparent goal was to group boxes by precise location (and by necessity, recipient) only. If you only wanted to group related images along those lines, which I'm not averse to doing, you should have only merged the athlete/2012 XX gold/locality categories, not the whole lot. What possible benefit was there to destroying the continuity of the indexing of categories like Category:Type A/B post boxes in the United Kingdom, in this manner? What possible advantage have you produced by actually separating images of the same box when it was red, from when it was gold? Before this change, they sat side by side in trees like Type because they were properly indexed. And what possible benefit have you produced by forcing people to make 110+ mouseclicks, if their goal was previously achievable in 3? I am not interested in vague references to avoiding clutter, that's meaningless. My definition of clutter for example is what you've done to Category:Type A/B post boxes in the United Kingdom, dumping a whole bunch of sub-categories into it, when the only difference between them all and the remaining top level images is one basic feature - colour, which was already catered for in another branch, and the only difference between the actual sub-categories is the precise location (& hence recipient), which for recipient was already catered for by virtue of the athlete's top level category, and location by the associated indexing. I've put many hours (weeks) of effort into categorising and indexing these images, if that's going to be wiped out in the space of 24 hours (and not even finished off properly), I'd like some actual Commons specific reasoning from you which actually talks about the features/qualities of postboxes. Ultimately, based on these replies so far, I'd like some indication that you actually understand how & why the system was organised in the first place, before you decided to change it. Ultra7 (talk) 02:31, 16 December 2012 (UTC)
These categories, such as the Hannah Cockcroft example, are fine and correctly integrated into Commons category structure. The reasoning here is as follows: A category for the specific post box is clearly useful as it allows people after a pic of an individual, significant, structure to have all our media in one place. That category then belongs in all appropriate higher-level categories, such as location, the fact its a gold post box, etc; as these concepts apply to the post box not the individual photographs. With the category in the gold post box cat, it is classic over categorisation to have the file in the gold post box cat directly.
This is no different to say, aircraft. Some individual aircraft have categories specific to that individual aircraft - if so, images of that plane are only in the aircraft's cat - not in the models. And BTW the fact a few individual aircraft have their own cats it doesn't mean all aircraft must. "Index" categories, which allow the user to see all options at once, are not the primary category method on Commons - but may be ok as a supplement. That's why Category:United Kingdom doesn't contain millions of photographs, but several high-level concepts.--Nilfanion (talk) 02:54, 16 December 2012 (UTC)
Thanks Nilfanion. WereSpielChequers (talk) 11:32, 16 December 2012 (UTC)
The primary categorisation method on Commons is to keep sub-dividing related features downwards in the same branch. You don't jump from one unrelated feature to another down one tree, just to separate all images of a particular object from other objects. You don't lump two unrelated features together just to combine a third related one, if that's already been catered for elsewhere. This has nothing to do with aircraft (or churches for that matter), unless you really think that for example, a category that sub-divided aircraft images by their particular wing profile, should then have one further sub-division under that for 1% of the total contents, separating them by a completely unrelated feature like, ironically, colour. It's nonsense. It is actually a perfect example of over-categorisation. The simple fact is, you've sub-divided 4 of the 5 branches of the UK postbox scheme (location, postcode area, reign, type) along lines that are basically meaningless to the parent quality, for a vanishingly small total subset of their potential contents. All apparently because you wrongly think that the most important defining characteristic of a gold postbox, is who it was for, rather than other characteristics like location or even box number (your system for example doesn't group together the boxes painted for different people in the same town, nor does it even group gold boxes in the same town for the same person, nor does it differentiate between different box numbers). There was merit in categorising by 'recipient', but that's not a simple as you both seem to think it is, it's a multi-layered quality, which doesn't need to be applied to all the branches. Performing that under the top level Olympic cat or under the gold cat (excepting the two non-UK boxes) would have been sufficient. Why would someone interested in Type A/B boxes in general, want to first browse through all the red images, then browse through 63 sub-categories of gold ones, just on the off chance the particular feature they want is only present on a gold Type A/B image? Why would someone looking for Hannah Cockroft's gold postbox, be looking for it via Category:Type A/B post boxes in the United Kingdom? Rather than the logical place, the Olympic branch, or at worst, the gold branch. It's a totally ridiculous proposition. Why should a user not be able to see a red image of box xyz, right alongside a later gold image of the exact same box? That is what torpedoes your logic here - you've basically sub-categorised all these branches by box number, but only once they turned gold. It's not proper categorisation, it's just an undue and ill-conceived focus on one aspect, at the expense of many others. And thanks to this change, you've completely destroyed a user's ability to browse all Olympic postboxes without having to click through 110+ categories, hard-wiring into Commons the flawed assumption that uploaders/users already know who the box is for, or want images of a box given to a known person. Quite why neither of you seem to appreciate what a total disaster that is in terms of helping users find anything other than a single box/recipient combination is beyond me. Making Commons completely unusable for whole classes of user is certainly not a goal of the categorisation system. As an example, you've separated images of the exact same postboxes just because of their new found status as Olympic boxes. We quite literally now have some red and gold images of the exact same box, now in completely different places with absolutely no way for users to find them (in the quite frequent cases given the lack of proper titles/descriptions, where the box no was only identified via indexing). Worse, you didn't even bother to finish the job and make this damage uniform across all gold boxes. If you had though, you might have realised some of the flaws it's introduced - 2 postboxes that aren't in the UK are now going to be mis-categorised in any number of UK branches, images of multiple box sites now cannot have their individual types identified without absurd levels of sub-categorisation or worse, duplicate categorisation. Even the one benefit of grouping the images together hasn't even been done properly - with the flat system adopted hasn't for example given the user any hope of appreciating the difference between Team GB boxes and Team Ireland ones, or the boxes not awarded to any athlete, or to groups of athletes, or the launch boxes, etc, etc. The use of box/es plurals is all over the place, it neither corresponds to box numbers or sites or even basic concepts like how many letter holes a 'box' has. I think this is what shows what was behind this - a simple ignorance of how complex the topic is. It most certainly is not as simple as one box per medal as this schema suggests at first glance, that's for sure. If all that wasn't enough, you've completely erased many weeks of work on my part, the one person whose uploaded and categorised the vast majority of these images, and you don't seem to give a flying fuck about it. You couldn't even be bothered to remove the now useless indexes on the categories when you moved them, leaving users guessing as to why for example, Category:Andrew Triggs-Hodge's gold postbox, Hebden Post Office and Category:Danielle Brown's gold postbox, Skipton are both filed under 'B' in Category:Post boxes in North Yorkshire. What was a useful feature properly implemented when they were together, is now totally useless as well as being completely misleading. If you've even bothered to read this far, I'll ask one last time what possible benefit this change actually gives browsers of the type/reign/area etc branches, but I'm not expecting a response frankly. Ultra7 (talk) 14:41, 16 December 2012 (UTC)
You've also managed to place this image in both Category:2012 Summer Paralympics gold post boxes and Category:2012 Summer Olympics gold post boxes in the United Kingdom, and a bunch of others too, in your haste to destroy the deliberate and usefull distinction between gold postboxes in the UK and Olympic/Paralymic scheme boxes. Ultra7 (talk) 16:41, 16 December 2012 (UTC)
You are looking at this the wrong way, and your methodology breaks when switching from high-level concepts to very precise ones as a direct result. You see the 8 categories here as clutter, when in fact having those 8 categories is a good thing for the users of the cat - it removes 41 images of clutter to those people looking for a Type K - who presumably want a red one.
The most important question is "Is having all our images of Hannah Cockcroft's box grouped together helpful?" - a clear yes because that individual box is significant and there's a number of images of it. The next question is "given that category is useful, what properties does that post box have that it should be categorised under?". Its in Halifax, so should be in the Post boxes in Halifax. Its to do with Hannah Cockcroft, so should be in her category. The post box is gold, so be in the gold box tree. Its a QE2 box, so should be marked as a QE2 box, etc.
The category name of Category:Hannah Cockroft's gold postbox, Halifax needs to be concise, so latches onto her name as the only reason the box is noteworthy is the fact its dedicated to her. However, by categorising it in all the trees, the categorisation of that box encapsulates all aspects of it, not just those mentioned in its name, such as the fact its a QE2 box.There are 2 boxes to Hannah, and both are in Halifax. They should be disambiguated by noting which is for which event.
You do not own the post box tree, and if someone makes changes in line with Commons policy which arguably improve it, you don't help your viewpoint by saying they "don't give a flying fuck" about your work. There's room for improvement - eg grouping together the gold boxes and having "Gold Type K" as a specific cat. Errors such as merging of the Olympic/Paralympic boxes can be corrected. Personal attacks are not acceptable, if you persist in those I will take this to the AN.
If you aren't satisfied with this, I suggest you start a broader discussion at the village pump about how having low-level categories breaks usability at a higher-level (due to need to hunt through dozens of cats) - which is your chief concern here? Bit of advice - try to keep it brief or people will ignore you.
Oh, and with regards to the "what about the box before it was gold" question: If we have a picture of that box when it was still red, its still of the box, so put it in the specific cat for that box - you'll see it alongside the images of the gold version without any possible confusion with other boxes. Naturally as it is red in that image, it should still be in the red box tree.--Nilfanion (talk) 17:26, 16 December 2012 (UTC)
Don't patronise me. You know that I know full well that I don't OWN the tree, but I do expect consideration of my efforts in this area when mass changes are effected that cause multiple issues, and dump entire indexes as apparently worthless. And pointing out errors is not a personal attack. Neither you or WSC has offered any policy justification here, you are simply advancing a personal opinion that someone for example browsing the K-Type cat looking for a specific detail, something which has got nothing to do with colour, would consider the gold images as 'clutter' and not worth viewing. It's total supposition, and can be easily defeated with reference to various practical scenarios, that actually do refer to policy concepts like categorising by visual features. Have either of you for example ever actually needed to look for a particular gold box without knowing the recipient? Do you want to try it once this system reaches maturity? Have you got a spare half-hour or a super-computer? And your figures are unrealistic too - there are 20 gold K-types in all, so that's 20 sub-categories eventually, which will consume the whole screen of anyone arriving at the cat looking for red K-types. And let's be clear - all the gold images were already in the correct trees, the issue is their move down one level into a box specific category, across the board. This is appropriate for some branches, but not others. As for the idea that both red and gold images should go in categories with names like 'Hannah Cockroft's gold postbox, Halifax‎', come on. That would look like an obvious error to anyone - if you want those categories to be about the box whatever the colour, then name it accordingly. For that Hannah example, that's Category:Royal Mail postbox HX1 34, Town Hall Wesley Place, Halifax. Ultra7 (talk) 18:56, 16 December 2012 (UTC)
Commons policy is rarely written down more explicitly than the guidance at COM:OVERCAT. Like it or not, but it is not possible for me, or anyone else, to give you clearer guidance than that. If you feel the guidance there is insufficient for this situation, then take it to the policy talk page / VP, as anything else I say here is my opinion isn't it? Sometimes lack of clear policy is a pain in the neck, this may be one. That said, what guidance is there should be sufficient to visualise the issue: Bad "clutter" is too many images in high level categories, and not too many categories. If I quote from that section, replacing the original example with a postbox specific one, the guidance should be as clear enough:

An example of this: Category:Eivør Pálsdóttir Category:Laura Trott gold postbox, Harlow was categorized only in Category:People by name Category:Post boxes in Essex. So if I add an image of her the post box, and know who she is about post boxes, I would also place the image in Category:People of the Faroe Islands Category:Type C post boxes in the United Kingdom and Category:VocalistsCategory:Laura Trott. This is over-categorizing, I've caused clutter in the top categories by adding images directly to them.

Therefore following this, Category:Laura Trott gold postbox, Harlow should be in Category:Type C post boxes in the United Kingdom (or a sub-category). Which is what myself, and WSC, have been saying here. The indexing function is preserved through the category structure. As for the choice of category name, there's nothing contradictory with having an pre-Olympics image of a red box in a category called "X's post box". And its obvious these boxes are best known by their connection to the athlete, not the code RM assigns to them (and even applies within RM internal documentation); just as the Enola Gay is at Category:Enola Gay not something like Boeing B-29 Superfortress serial number 44-86292). And do I really need point out that swearing at someone is not conducive to a positive atmosphere?--Nilfanion (talk) 21:03, 16 December 2012 (UTC)
Hi Ultra7 and thanks Nilfanion for the explanations, I'll just cover a couple of other points. Firstly taking Category:Laura Trott as an example, having subcategories for each of her goldboxes now means that all the images we currently have of her can come up on one screen which makes it more convenient for people who are accessing commons with an interest in her or sport. If her career and the gold box tradition continue through future Olympics then I can see a day where she or other Olynpians will need a category for gold boxes, just as we have Category:Monuments and memorials to Winston Churchill - not a category that I created though I created one of the subcategories for it. Not every athlete will get so many images, or images of their gold box that it is worth creating a specific category for them, and if I were to go through all those gold box images systematically creating a category for every one of them even if we only had a single image then I would arguably be overcategorisng things. Also to answer your question "what possible benefit this change actually gives browsers of the type/reign/area etc branches", well because I retained your categorisation work and built on it by moving those categories into categories for individual postboxes, when people add further images of those postboxes they are probably going to go into the relevant subcategory and as a result be categorised by PO box type. I suspect that sportsfans who are interested in who the boxes are dedicated to and people who are interested in their local area and are taking photos of their local landmarks are at least as numerous as those who are interested in different designs of Post boxes. So because I've created those categories there will ultimately be more images loaded into categories for the various PO box types. A second advantage of that grouping multiple images of the same box into one category is that if someone in future decides to segment the "PO box design in UK" categories, either by manufacturer or more likely by geography, grouping images of the same box together into one category means that you need one edit to change a category rather than several edits to change the individual images in that category, which also in part addresses the point you made about User Fae and your watchlist getting busier than you'd wished. WereSpielChequers (talk) 10:25, 17 December 2012 (UTC)
Look guys. I've been on Commons for years, I've uploaded, categorised and editted thousands of images. You can take it as read that I fully appreciate the utlity of categorisiation, especially regarding casual uploaders/reusers. Aside from my irritation at the lack of consideration of my knowledge of this topic, and the annoyance at the many issues this change has caused, my point is a very technical one that I would appreciate you discussing with reference to the actual topic, which is postboxes. Your examples about monuments, churches, aeroplanes, and yes, even people from the Faroes - in every single one of those cases, any particular sub-category can always be fully emptied into the chosen schema for sub-categories, if desired. And in every single one of those cases, there's no particular need to examine images of different objects side by side at the very highest level. In the case of gold postboxes, you've failed to realise that, at least for type/reign/postcode area, there's not one single practical reason why any of the 110+ gold postboxes should be considered different to any of the potentially millions of red ones (or any of the other colours that are in some of those categories). The fact that it ensures newly uploaded boxes are immediately fully categorised is a reason, but it's not one whose practical benefits outweigh the practical problems it causes in this specific case. A case in point - at least on box site has two different types, so already the idea that this schmea automatically categorises these images by type becomes less certain. In another, one box has been awarded jointly to an Olympian and Paralympian, I can only imagine the issues that might cause. And besides, that sort of incomplete initial categorisation and complex intersections is at least something category maintainers and image editors like me can identify and fix in due course, as I'm quite sure is still necessary for planes, monuments, people etc. Category maintainers however can do nothing about a badly designed schema like this. The idea that these categories makes future movements easier is also rather pointless - the indexing ensures that they are together, making movement by catscan or even manually, just as easy as with a separate category. It probably takes longer infact if you need to visually inspect images before deciding where they're going. I would perhaps be less annoyed about those sorts of issues, if you'd at least bothered to create a comprehensive schema - a category for all gold boxes (we do for example have cases where we have a red image but not the corresponding gold one yet). There is a finite set of gold postboxes, each with a known recipient. But no. What you've created is the worst of both worlds - for a category like Type C boxes in the UK, you've both 'tidied' the 'clutter' of some gold boxes away, while leaving others there, for no reason other than number of images (which no ordinary user is going to comprehend, at least at first, and which is certainly of no benefit to maintainers, infact it makes our job harder). Anyone wanting to browse all the gold boxes, or even all the Type C gold boxes, is inconvenienced. Anyone looking for all the Type C boxes regardless of colour, is inconvenienced. Anyone looking for red and gold images of the same box, is inconvenienced (the idea that red boxes logically go in gold categories, is a complete fantasy, it doesn't correlate with any other example I've seen, and I've seen those sorts of things actually be reversed by casual and experienced users as erroneous). Last but not least, anyone who is the first to upload a particular gold box that has no other images here, is due to its non-comprehensive nature, left with a set of decisions and assumptions to make that requires actual knowledge of the schema and perhaps more information about the box than they might posess, making mis-categorisation more likely not less. Ultra7 (talk) 14:49, 17 December 2012 (UTC)

How about trying to work out a viable scheme? The gold post boxes have complications, for instance the non-UK boxes, but if a workable method is found for the simple cases expanding it to the more complex is possible. Here's a starting point:


  1. Category:(Athlete's) gold post box should be created, pluralised when necessary to cover all post boxes for the athlete. To start with this is only a subcat of the athlete, and the 2012 Olympic or Paralympic gold post boxes.
  2. If its a team gold (eg Dressage), then the category should incorporate the team name and event (TeamGB Dressage gold post boxes).
  3. In the case of a single box being for 2 different athletes, call the category for both athletes.
  4. If a single individual, or team, has gold boxes in many locations, use location to disambiguate (Ben Ainslie's gold post box, Lymington), and make these a subcats of the generic one for that athlete. If an athlete won 2+ golds, the event could be usedto disambiguate (Mo Farah's 10,000m gold post box), but I'd favour using location.
  5. If there are multiple boxes at the exact same location, use another method to distinguish them, such as the RM code.
  6. So Category:Nicola Adam's gold post boxes is only in her cat and in the 2012 Olympic gold post boxes. It would contain "Nicolas Adam's gold post box, no LS1 424" and "Nicolas Adam's gold post box, no LS2 1501".
  7. At this point, the individual post box should have a category that applies to it, and only it.
  8. Now apply all relevant cats (reign, type, etc) to the category for the post box, and place all images of that box in that category. At the same, remove from those files any and all of the categories that are now on the boxes cat.
  9. In the case of two boxes at the same location, the location info (the street) applies to both so should be on the generic, as opposed to the two specific cats. Individual files should be placed in the cat for whichever box they show, or in both of them if both are shown.
  10. All images of the gold boxes should now be in the cat for the specific box, and that will be the only postbox related cat. There may be others like date, uploader etc, that are specific to the image itself and not its subject.
  11. Also include all images of the specific box, before it was gold. Those images should be in the appropriate colour cat too.


Some further comments about the above:

  1. These scheme covers most things, and only thing broken is the ability of cats like Elizabeth II post boxes in the UK to display a full list. However, what happens if I create Elizabeth II post boxes in England? Again the index capability is lost, but that new category would help those after English boxes. Geography is an obvious secondary feature for an intersection, but its not the only one. For example: Having this image in Dawlish and Cygnus atratus is sufficient to tell you everything we have about the bird's identity and location, but the meaningful intersections, Birds of Devon and Swans of England add value, as they give you more ways to find it and compare it to related imagery than having to know either the precise species or location.
  2. Indexing can be handled in other ways, such as galleries, which don't get disrupted by improvements to the category tree. Galleries provide low more powerful sorting, meaningful captions, ability to remove junk images, and so on - like the gallery for that box in Manchester. CatScan can also be used to provide a visual index: Simply add the following text - or something similar - to the category description: "For an automatically generated gallery of images in this category, and its subcategories, click here" (This breaks at 1,000 images, but 1,000 is surely too many files, we do users a disservice by giving them too much choice) This preserves the image display regardless of what happens to the tree structure, without the maintenance problems associated with a gallery. Given that, useful development of the tree shouldn't be held back to avoid the loss of that ability - because it won't.
  3. As for colours: A picture of a red box in the generic gold post box cat is somewhat silly, and someone would logically remove it. However, when its in the category for a specifc post box, which happens to be known as a gold box now, its not wrong logically. It doesn't need to have a cryptic RM code as a category name (remember that code is also potentially unstable, as if RM may have renumbered it when the postcode was created or adjusted) - as long as it uniquely identifies the subject, which the above scheme does. As an analog, this image is of Kate Middleton, but not of the Duchess of Cambridge, but that's no reason to remove it from a category associated with the Duchess and maintain an seperate cat for Kate which is not completely independent of that the cat for the Duchess.
  4. Once individual box cats exist, they can be categorised in ways that have nothing to do with postboxes: By athlete, by sport... That's a major benefit.
  5. If a typical Commons user comes along with a photo they took themselves, and its the first of a particular gold box, they presumably know location and athlete. At minimum they will know its gold and what country its in. The scheme above is simple to apply and create the right cat (at athlete's name), or it can be dumped in in the gold box cat - where it sticks out as all properly sorted files will not be there. A knowledgeable editor can come along later and add the postbox technical details which the original uploader is unlikely to know much about. A knowledgeable uploader can do it all at once.

I've deliberately numbered the above points, as there are several issues, please keep any responses similarly segregated, to ease further discussion.--Nilfanion (talk) 23:13, 18 December 2012 (UTC)

Re your points 1-11:
  1. I'm not convinced that we need a category that groups together multiple gold boxes for the same athlete. Take category:Laura Trott as an example, she has three different subcategories, maybe if she is as successful at the next Olympics there will be an advantage to grouping her goldboxes together. The other angle is the geographic one, and if goldboxes continued to be a UK wide category then it would make sense to group by athlete to prevent it becoming cluttered. But there is a longstanding tendency to break up UK categories by nation and even by county, my prediction is that someone will start doing that sometime.
  2. If the gold is for a team not for an individual then it would make sense to name it for the sport.
  3. Agreed
  4. Agreed, I prefer location, but somebody is bound to start using event as well.
  5. For multiple boxes in the same town or village I've just gone to a lower level of location - street if necessary. Not everyone will know the RM code, and geography seems more natural to me. My assumption is that everyone photographing a goldbox knows where they are, and a large proportion will know why the box is gold, but whereas many who photograph red postboxes will be interested in the type of post box this is unlikely to be the case with those who photograph their town's goldboxes. Of course if my assumption is wrong then this may change.
  6. I wouldn't be so prescriptive as to where these will go, if someone creates the category "goldboxes won by UK rowers" under a Rowing in the UK category then I'm not going to say that isn't part of the scheme.
  7. I'd agree that any goldbox for which we have "n" images is worth creating a category for. But we may have difficulty defining "n". Personally I'm unlikely to create a category for less than 3 images, unless those images are otherwise very distant from each other in their category. I'm not convinced that we should be creating subcategories for objects that we only have one image of.
  8. "Now apply all relevant cats (reign, type, etc) to the category for the post box, and place all images of that box in that category. At the same, remove from those files any and all of the categories that are now on the boxes cat." That is just a basic restatement of how the categrisation system works. as per COM:OVERCAT it shouldn't really need restating in a specific guide for Postboxes.
  9. "In the case of two boxes at the same location, the location info (the street) applies to both so should be on the generic, as opposed to the two specific cats. Individual files should be placed in the cat for whichever box they show, or in both of them if both are shown." This only applies if there are enough gold boxes in a place for someone to create category for goldboxes in that place. That is unlikely to happen for places with two goldboxes.
  10. "All images of the gold boxes should now be in the cat for the specific box, and that will be the only postbox related cat. There may be others like date, uploader etc, that are specific to the image itself and not its subject." Again this presumes that it is worthwhile creating categories for goldboxes for which we only have one or two images, I'm not convinced of that.
  11. "Also include all images of the specific box, before it was gold. Those images should be in the appropriate colour cat too." I can see where you are coming from on this, though this isn't what I've been doing, but I did classify as a gold box a goldbox with just its undercoat. I can also see this leading to slow burn edit wars of the yes it is no it isn't variety.
Overall I'm not convinced that we need a specific scheme for goldboxes, and if we do my talkpage isn't the ideal place to develop it. We three are in disagreement as to how big a category can be before it is reasonable to subdivide it; Whether there is merit in maintaining an index by having flat categories where everything is on the same level; and how many related subcategories can be in a category before it makes sense to group them together and add another layer to the category system. I don't see any of them as being specific to Postboxes. WereSpielChequers (talk) 17:26, 23 December 2012 (UTC)
I agree above is overcomplicated, and probably unnecessary, as if post box cats were just created the workflow above is more or less what would result naturally. The scheme is designed to provide a way to uniquely name a category for every last box - regardless of whose box, or how many images of it we have. It doesn't mandate any as essential, but provides a way to place them. And with regards to point 5 - that's about post boxes at the same exact location, so not just the same town, but the same street too. In those cases, location-based disambiguation fails.
There's no automatic reason to create cats for every last box, as there's clearly little benefit to some. (Why split Nicola Adam's two boxes into two cats?). However, there's no harm either - so if anyone wants to they can.
One aspect directly affected by not creating cats to individual box level is Point 8 (the re-statement of OVERCAT). If Category:Nicola Adams' gold postboxes, Leeds is not split into categories for the 2 boxes, then that category should only have classifications relevant to both postboxes, not one-or-the-other. ie It should be in Gold post boxes (both are gold), but should not be in Type C (as only one is). Those aspects specific to one box only, should be on files depicting that box. On the other hand, if it is split into categories for each box, then those properties will be assigned to the cat, not the individual files.--Nilfanion (talk) 23:51, 23 December 2012 (UTC)
Oh absolutely - goldboxes in Trumpington would be the equivalent of churches in Trumpington with denomination, year of build and so forth all at whatever level the individual churches are. I would't assume that street is our lowest level of geography. I can remember one pair of goldboxes which appeared in the same photo and were outside the same building, but otherwise things could well go the way of churches, it is quite possible to have multiple churches on the same street with a category for each and some wide angle shots in categories for multiple churches. WereSpielChequers (talk) 01:11, 24 December 2012 (UTC)

Hampshire and Isle of Wight[edit]

Hi, Hampshire and the Isle of Wight are now complete, so you may want to take a browse and see if tools like catscan might throw up anomalies for other categories. :-) -- (talk) 20:31, 1 February 2013 (UTC)

Import request: Geograph singleton[edit] happens to be of the gunpowder works I'm drafting an article about today. And as November 2012 is not imported. If you have a moment to import it, I'd be very grateful. Or if there is an instruction page to follow, that would do fine. Charles Matthews (talk) 11:54, 14 April 2013 (UTC)

Hi Charles, I've categorised loads of the Geograph images but the only images I've ever uploaded have been my own photographs, and despite a couple of searches I haven't found an easy guide to importing other people's images. Commons:Batch uploading/Geograph explains why the main batch uploading of the Geograph stopped. I'm sure that small uploads are fine providing you categorise the things properly, and uploading from that site is something I could sort out in an hour or two, but I don't know when I'm likely to have that time free. Sorry, may I suggest the village pump? WereSpielChequers (talk) 19:36, 15 April 2013 (UTC)

Church cats[edit]

Umm.. when the church name contains the place, like here, there isn't any real name to add the place as a disambiguator. I realise Ashburton itself is ambiguous, but Ashburton Methodist Church, Devon isn't :)--Nilfanion (talk) 22:34, 5 July 2013 (UTC)

Good point, that would have been shorter and less clunky. I very rarely encounter ones where the county is a sensible suffix, but there it might have been. BTW are you aware of the impending Wiki Loves Monuments contest in the UK? The plan is to target our 60,000 Grade I Grade II* and equivalent outside of England, hence I've recently done a bit more focus on those buildings. WereSpielChequers (talk) 22:42, 5 July 2013 (UTC)
I have been busy with a fair few things, so not even got round reducing my backlog of photo to upload (which include listed buildings of course). Parish-level cats are a major boon to tasks like that one, as then our geographical hierarchy matches English Heritage's. I will check that WLM drive out later, might even give me the kick I need to get more images on here again...--Nilfanion (talk) 22:52, 5 July 2013 (UTC)
Anything uploaded in September can be counted in the contest, provided it is one of the 60,000. You don't need to have taken the photo in that month. So yes they should qualify. WereSpielChequers (talk) 22:55, 5 July 2013 (UTC)


Motacilla (talk) 06:57, 12 August 2013 (UTC)

replied there, WereSpielChequers (talk) 09:29, 12 August 2013 (UTC)


I have a picture like this [[8]] ? Izahias (talk) 00:16, 26 October 2013 (UTC)

Hi, I think you have your answer at User_talk:Herzi_Pinki#picture WereSpielChequers (talk) 08:14, 28 October 2013 (UTC)

File:Celiing joists.JPG[edit]

Commons-emblem-issue.svg File:Celiing joists.JPG has been listed at Commons:Deletion requests so that the community can discuss whether it should be kept or not. We would appreciate it if you could go to voice your opinion about this at its entry.

If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!

Afrikaans | العربية | Български | বাংলা | Беларуская (тарашкевіца)‎ | Català | Čeština | Dansk | Deutsch | Deutsch (Sie-Form)‎ | Zazaki | Ελληνικά | English | Esperanto | Español | Eesti | فارسی | Suomi | Français | Galego | עברית | Magyar | Bahasa Indonesia | Íslenska | Italiano | 日本語 | 한국어 | Македонски | മലയാളം | Plattdüütsch | Nederlands | Norsk nynorsk | Norsk bokmål | Occitan | Polski | Português | Português do Brasil | Română | Русский | Slovenčina | Slovenščina | Shqip | Српски / srpski | Svenska | Türkçe | українська | Tiếng Việt | 中文 | 中文(简体)‎ | 中文(繁體)‎ | +/−

Jim Derby (talk) 00:23, 4 December 2013 (UTC)

A small request[edit]

Can you add File:Emily Kinney singing.jpg to w:Emily Kinney when you have a chance? INeverCry 22:17, 22 December 2013 (UTC)

yes no problem, WereSpielChequers (talk) 22:24, 22 December 2013 (UTC)
Thanks. INeverCry 22:34, 22 December 2013 (UTC)

St Luke's church, Peckham[edit]

Unfortunately this is in SE15 not SW15 as the 30+ pictures of it would have us believe. One day I may fix each one - or do you know of a bot that can do this? Best Wishes S a g a C i t y (talk) 06:50, 16 February 2014 (UTC)

Thanks, done. WereSpielChequers (talk) 08:59, 16 February 2014 (UTC)
It is I who should thank you. I ought to have done it when I noticed the errors … but forgot. S a g a C i t y (talk) 15:14, 16 February 2014 (UTC)

Frogmore and Sherford[edit]

Not sure about some recent changes in Category:Frogmore and Sherford - relating to both of its villages. The issue with Sherford is more serious, due to the stupidity of the councils. w:Sherford (and by extension, Category:Sherford) are about the new town that should be built in near future on the edge of Plymstock. When built, that new town will dwarf the village of Sherford, making it clear primary topic on en.wp. Unfortunately, the new town will also be (mostly) in the South Hams - so dabbing the village will be awkward... Would be better if they had chosen a distinct name for the new town! When that new town is built, will be confusion unless its given a better name (or Plymouth expands slightly) - I remember hearing news stories that its already caused sat nav errors.

As for Frogmore, not sure its helpful to categorise images of the Frogmore Creek as in Frogmore - most of the creek isn't in the parish never mind the village.--Nilfanion (talk) 17:41, 23 February 2014 (UTC)

Hi Nilfanion, that is stupid, perhaps Category:Sherford (village) and Category:Sherford (new town) would sort it? WereSpielChequers (talk) 18:10, 23 February 2014 (UTC)
Might need to retain a Devon disambiguation for both (there's a Sherford in Taunton). Semantically, I'm not sure if its correct to call the new town a "new town". That term seems to have a technical meaning in UK, which might not apply in this case (I'm unsure). In any case, its a moot point for now, as the town doesn't exist yet. That said I think the principle behind your suggestions is only workable one.--Nilfanion (talk) 19:20, 23 February 2014 (UTC)

English churches in the 1890s[edit]

Giggleswick Church, 1890s; best seen at full resolution.

I have spotted quite a few rather nice photographs of English churches in 19th century photochrome prints of the UK and Ireland. You may want to have a browse as something like a DYK might pop out of it. -- (talk) 08:10, 8 June 2014 (UTC)

Thanks, nice work. I rarely write DYKs myself, but I have added one or two of those pictures to articles, and categorised lots. Category:19th century photochrome prints of the UK and Ireland (uncategorised) is the temporary holding pen for the uncategorised ones, that way I know which ones have a meaningful category. WereSpielChequers (talk) 18:45, 9 June 2014 (UTC)


Oxyman (talk) 19:29, 11 June 2014 (UTC)

Chesil Beach[edit]

FYI I've just reverted your removal of Category:Isle of Portland from a number of Chesil Beach files such as File:Chesil Beach - - 1311791.jpg. Reason is the eastern end of the beach is in Portland CP, and the cat for the Isle is equivalent to the parish. Without Isle of Portland the files do not have parish-level localisation; and as the two concepts are effectively the same area its a bit pointless to have 2 cats IMO.--Nilfanion (talk) 18:27, 20 June 2014 (UTC)

Emphasising the local[edit]

Good to see you today!

I think you were right about the power of emphasising the local, to chapters and wikiprojects.

I put out a number of notices like this late yesterday evening [9], and they seem to have drawn a few more people. (Well, Perth & Kinross in Scotland has been done, anyway).

I have now put a similar notice on en:WT:USA, so it will be interesting to see whether that gets any interest.

Take care, Jheald (talk) 23:04, 3 November 2014 (UTC)

Thanks, it is a pity we weren't able to do the tagging on commons as that would have enabled people to do far more with the images that were relevant to them. Equally if we had someway of identifying the ships, birds and churches accordingly there would have been people interested in those. I'm going to have a think about commons as a crowdsourcing site, at some point I may draft a proposal. WereSpielChequers (talk) 04:39, 4 November 2014 (UTC)

A barnstar for you![edit]

Original Barnstar Hires.png The Original Barnstar
Thanks for all your help tagging maps in the Mechanical Curator collection - after ten days, we've now done half the books and found 10,600 images! Andrew Gray (talk) 00:49, 11 November 2014 (UTC)

Wikimania 2014[edit]

FYI: File:Wikimania 2014 MMB 51 WereSpielChequers.jpg File:Wikimania 2014 MMB 52 WereSpielChequers.jpg. -mattbuck (Talk) 19:48, 4 January 2015 (UTC)


I don't understand the cat removal. Frampton Mansell is in the Cotswolds.

Saffron Blaze (talk) 23:42, 27 January 2015 (UTC)

Indeed, and Category:Frampton_Mansell is in the category district of Cotswold. But that image doesn't need to be in cotswolds any more than it needs to be in category:Saint Luke churches in England. WereSpielChequers (talk) 23:52, 27 January 2015 (UTC)
Indeed, it is in the district, but it is also in the range of hills with the same name, but doesn't show there: No biggie. Saffron Blaze (talk) 03:41, 28 January 2015 (UTC)

Hamlet cats[edit]


I have a couple concerns with some of your recent edits regarding hamlets. I'll use Category:Ash Thomas as an example.

  1. First point is straightforward - Its just in Category:Towns and villages in Devon. But it is not in the CP cat, and you have removed the CP on some as a result. Settlements should be in their parish, district etc...
  2. The more difficult one is what is the point? The 5 images you placed in Ash Thomas are all completely rural scenes, in the general area of the settlement, but do not show the settlement. That means they are all useless to illustrate anything about Ash Thomas (eg w:Ash Thomas), as they just some stuff in its vicinity.
In contrast the field image is useful in Category:Halberton. It is just as useless to describe the village of Halberton as it is Ash Thomas. However it is in the civil parish. Can sum up my opinion as "stuff nearby = irrelevant, stuff in = relevant".--Nilfanion (talk) 22:27, 12 February 2015 (UTC)
Hi Nilfanion, I've been going through articles on Wikipedia where there is a request for photographs. Ash Thomas was one of them, where I find some images that are right for the category even if they aren't suitable for the purpose of illustrating a wikipedia article then I add them to the category. If Ash Thomas is in the civil parish of Halberton then I've no problem with it also being a subcategory of Halberton. Does that work for you? WereSpielChequers (talk) 23:34, 12 February 2015 (UTC)
Adding as a sub-cat to the parish cat is obviously sensible - and same will apply to your other recent edits I'd imagine.
However, can you explain to me how these images are right for Category:Ash Thomas in any way, shape or form? They don't show any aspect of Ash Thomas, but depict a gate, some fields and a couple roads nearby. To my mind, that's analagous to including this Google image in Category:Palace of Westminster. Well it shows Jewel Tower, which is part of the complex.. but you should get my meaning :)--Nilfanion (talk) 00:11, 13 February 2015 (UTC)
I didn't take or upload those photos, and having found them I chose not to use any of them to illustrate the wikipedia article, so I don't see our positions as that far apart. Where the analogy falls is the difference between a place and a building, that Google image doesn't illustrate the building, but if we were to put it into a place category it might well belong in the same place as the nearby building. WereSpielChequers (talk) 00:25, 13 February 2015 (UTC)
I agree we are fairly close here, we both see limited value of these images (the difference is you presumably see some value, while I see none at all).
A settlement is simply the collection of buildings (and associated areas such as streets and gardens). Its not the buildings plus that area around it that can subjectively be described as nearby. A field outside a settlement is outside the settlement, just as a road outside a building is outside the building. The same effect is true at the larger scale. Its useful to say Watford is near London - but you wouldn't include Watford imagery in London's cats.
If we had useful images of Ash Thomas, the presence of these images would reduce the signal-noise ratio.
In fact, it might actually be detrimental to have this cat at all: Ash Thomas is a logical subcat of Halberton, so images in Ash Thomas would not be in the Halberton cat. However, these images are potentially useful to show the rural landscape of the parish, and if someone wanted a rural photo, they might miss these by ignoring a subcat that can reasonably be assumed to depict the hamlet itself.--Nilfanion (talk) 01:04, 13 February 2015 (UTC)


Motacilla (talk) 14:40, 23 April 2015 (UTC)


Ndalyrose (talk) 20:33, 5 June 2015 (UTC)


You are receiving this message because a technical change may affect a bot, gadget, or user script you have been using. The breaking change involves API calls. This change has been planned for two years. The WMF will start making this change on 30 June 2015. A partial list of affected bots can be seen here: This includes all bots that are using pywikibot compat. Some of these bots have already been fixed. However, if you write user scripts or operate a bot that uses the API, then you should check your code, to make sure that it will not break.

What, exactly, is breaking? The "default continuation mode" for action=query requests to api.php will be changing to be easier for new coders to use correctly. To find out whether your script or bot may be affected, then search the source code (including any frameworks or libraries) for the string "query-continue". If that is not present, then the script or bot is not affected. In a few cases, the code will be present but not used. In that case, the script or bot will continue working.

This change will be part of 1.26wmf12. It will be deployed to test wikis (including on 30 June, to non-Wikipedias (such as Wiktionary) on 1 July, and to all Wikipedias on 2 July 2015.

If your bot or script is receiving the warning about this upcoming change (as seen at ), it's time to fix your code!

Either of the above solutions may be tested immediately, you'll know it works because you stop seeing the warning.

Do you need help with your own bot or script? Ask questions in e-mail on the mediawiki-api or wikitech-l mailing lists. Volunteers at m:Tech or w:en:WP:Village pump (technical) or w:en:Wikipedia:Bot owners' noticeboard may also be able to help you.

Are you using someone else's gadgets or user scripts? Most scripts are not affected. To find out if a script you use needs to be updated, then post a note at the discussion page for the gadget or the talk page of the user who originally made the script. Whatamidoing (WMF) (talk) 19:03, 17 June 2015 (UTC)

You have been randomly selected to take a very short survey by the Wikimedia Foundation Community Tech team![edit]

This survey is intended to gauge community satisfaction with the technical support provided by the Wikimedia Foundation to Wikipedia, especially focusing on the needs of the core community. To learn more about this survey, please visit Research:Tech support satisfaction poll.

To opt-out of further notices concerning this survey, please remove your username from the subscription list.

MediaWiki message delivery (talk) 20:51, 15 October 2015 (UTC)

St James', Spilsby and St James, Southwick[edit]

On 15 December 2012 you seem to have added 50 files to Category:St James, Southwick (interior) in Hampshire that look like they are from St James, Spilsby in Lincolnshire. I don't know how to use Cat-a-lot. Please will you move the files to Category:St James, Spilsby (sculptures) and/or Category:St James, Spilsby (interior), as appropriate?

Thankyou, Motacilla (talk) 09:28, 17 November 2015 (UTC)

Thanks, good spot. I've moved fifty files to Category:St James, Spilsby (interior) may move some to sculptures when I'm next online. WereSpielChequers (talk) 13:29, 17 November 2015 (UTC)

File:Holy Trinity, Landcross- banner (geograph 4583920).jpg[edit]

Commons-emblem-issue.svg File:Holy Trinity, Landcross- banner (geograph 4583920).jpg has been listed at Commons:Deletion requests so that the community can discuss whether it should be kept or not. We would appreciate it if you could go to voice your opinion about this at its entry.

If you created this file, please note that the fact that it has been proposed for deletion does not necessarily mean that we do not value your kind contribution. It simply means that one person believes that there is some specific problem with it, such as a copyright issue.
Please remember to respond to and – if appropriate – contradict the arguments supporting deletion. Arguments which focus on the nominator will not affect the result of the nomination. Thank you!

Afrikaans | العربية | Български | বাংলা | Беларуская (тарашкевіца)‎ | Català | Čeština | Dansk | Deutsch | Deutsch (Sie-Form)‎ | Zazaki | Ελληνικά | English | Esperanto | Español | Eesti | فارسی | Suomi | Français | Galego | עברית | Magyar | Bahasa Indonesia | Íslenska | Italiano | 日本語 | 한국어 | Македонски | മലയാളം | Plattdüütsch | Nederlands | Norsk nynorsk | Norsk bokmål | Occitan | Polski | Português | Português do Brasil | Română | Русский | Slovenčina | Slovenščina | Shqip | Српски / srpski | Svenska | Türkçe | українська | Tiếng Việt | 中文 | 中文(简体)‎ | 中文(繁體)‎ | +/−

Nilfanion (talk) 01:26, 8 November 2016 (UTC)

Geograph British isles project categorization[edit]

Hi, I found several files you've categorized from the nearly endless categories of the Geograph British Isles project, on which I work sometimes too, and would like to point to something: When moving files from one of this categories to approriate categories by cat-a-lot, they are left in the county category as West Yorkshire or Cumbria for example. To avoid overcategorization it would be helpful to remove these main categories from every file after moving it to the correct subcategory, because it is not done automatically. It's quite uncomfortable, I know, but otherwise someone will have to do it later, when cleaning up the overcrowded main categories. I learned that fact only after a while, when I had categorized lots of files without removing it from the main category, too... Best, --Tine (talk) 19:49, 9 December 2016 (UTC)

Or another way of looking at it, whoever simply put those geograph images into the county did no one any favours. The county categories are swamped to the point of uselessness and the workflow of using catalot to move articles out of uncategorised geograph into relevant categories is compromised. Hopefully someone will at some point use a bot to remove anything from a county category if it is also in a subcategory of the county. Removing counties manually takes longer than the catalot categorization I used to do, and is one of the reasons why I largely stopped doing that and am no longer so active on commons. I suggest you point out to anyone who suggests you remove those counties that they are welcome to do so themselves but you are under no more obligation to do so than they are. WereSpielChequers (talk) 20:21, 9 December 2016 (UTC)
Sorry, I'm not a native english speaker, your last sentence is unclear for me. But it seams to me I've annoyed you with my contribution - I'm sorry for that. My request was meant as a hint, not a criticism. Best, --Tine (talk) 21:49, 9 December 2016 (UTC)
Hi Tine, you haven't annoyed me, though the situation has. WereSpielChequers (talk) 21:52, 9 December 2016 (UTC)
Thanks - so than: I agree... :-) --Tine (talk) 22:08, 9 December 2016 (UTC)

Unidentified locations[edit]

Why remove images from Category:Unidentified locations in London without identifying the location? For example File:City of London 00 (31).jpg which you moved into Category:City of London. Not only is "City of London" a too vague a location, but in this instance it is unlikely to be accurate, I am fairly sure that this location is near Old Street but I would need more time to give an accurate geolocation. Images here are often uploaded by people with a limited knowledge of English, they may not realise what the difference is between "City of London" and "London". Oxyman (talk) 18:14, 27 January 2017 (UTC)

Greater London is a thousand square miles, City of London one square mile. So an improvement of two orders of magnitude. WereSpielChequers (talk) 18:34, 27 January 2017 (UTC)
Only if the location in the image is within that one square mile, otherwise this improvement is just inaccurate or plain wrong. Oxyman (talk) 19:12, 27 January 2017 (UTC)
As the policy says always place an image in the most specific categories, and not in the levels above those. If information turns out to be incorrect then files can be renamed or edited to correct that. But I suspect we are more likely to have that discovered after they are categorised than before. WereSpielChequers (talk) 20:41, 29 January 2017 (UTC)
Not really. Look at the thousands (or is it millions) of errors that the Geograph upload resulted in. If they were in the wrong category its very unlikely that fact was picked up, whereas if they were in an overly broad category - like the county - they are either in a correct, but vague, cat, or have been moved down. And that's despite having decent geocoding to assist matters, without that pure guesswork is needed.
With regards to London in general, an "Unidentified locations in London" is not the category above "City of London". In terms of the Commons category structure, it is actually at the same level as a geographic sub-cat of Category:London. If we cannot be positive of the location, its best to leave it in the unidentified location sub-cat rather than guess which sub-cat it belong to. The unidentified location cat is actually perfect for that task, its a slush pile for people to look at when they have spare time; just as the Category:Unidentified birds category sorts birdie pics out.
As for File:City of London 00 (31).jpg - its a view north up Macclesfield Road, St Luke's, in Islington not the City (Thank god for those blue roofs!)--Nilfanion (talk) 23:05, 29 January 2017 (UTC)
I fixed a lot of incorrectly categorised Geograph images. To my mind there is a difference between a bot that categorises an image described as in Wigtownshire as in Northern Ireland because it is geographically closer to the centre of Northern Ireland than the centre of Wigtownshire than my categorising an image that is described as in City of London into the category City of London. In both cases I'm categorising the file according to the information with the file. WereSpielChequers (talk) 13:52, 31 January 2017 (UTC)
Yes, there's a clear difference, but it boils down to trusting the information provided. Trusting the file description as being completely accurate without validating is risky, just as trusting GeographBot to get a location right is. In fact without any real means to validate, it might be even more problematic - geocoding allows detection and correction of Geograph errors. In any case, my opinion is that this file should have been kept in the unidentified location cat as just adding the relevant London borough isn't really a proper localisation for something in the capital - and being in both cats maximises the chance of a good localisation.--Nilfanion (talk) 21:26, 31 January 2017 (UTC)
I'm well aware that sometimes the information with a photograph will be incorrect, I've fixed quite a few where I've spotted errors. But categorisation has to start from somewhere, and it would be wrong to assume all information is wrong. Keeping images in both City of London and unidentified images in London would violate the principle of find the most specific category that fits the file WereSpielChequers (talk) 21:13, 2 February 2017 (UTC)
Seriously, there has to be some sort of logical balance here, I didn't catagorise this image File:Buckingham Palace, London.jpg as what the file name says it is just because "I'm categorising the file according to the information with the file". It was fairly obvious to me that File:City of London 00 (31).jpg and it's companion batch were not of the Square mile. So putting them in an unidentified location category until I had the time to sort it out seemed an obvious way to go. I can't see why you can't just admit to making an error on this rather you are arguing that some kind of procedure overrules logic or common sense. Oxyman (talk) 22:14, 2 February 2017 (UTC)
Unidentified locations in London is NOT a logical parent of City of London, so should not be removed on the principle you state. This is because you are removing information - the fact its an unidentified location! Just saying the file is in the City does not locate it. It would be like if I uploaded a picture of a gull, and mark it as an unidentified bird. You remove the unidentified cat, and use the knowledge you have to mark it as Category:Larus. It is now removed from the unidentified tree, making it harder to find for real bird experts - but is not correctly categorised to its species.
In practice we have Category:Unidentified Larus, and you'd move the gull pic there. That correctly sorts it to limit of information you can provide and marks it as needing better. The equivalent category you should have used is Category:Unidentified locations in the City of London. Oh wait - that cat doesn't exist? So either create it or just have both its logical parent cats.--Nilfanion (talk) 23:27, 2 February 2017 (UTC)
It seems to have escaped your attention yet again that the images you put in Category:City of London had nothing to do with the Square mile, So err not putting it in Category:City of London was not err "removing information". OK so you can't admit you were wrong and keep on with your procedure overrules logic or common sense arguments. I can't see anymore reason to carry on this discussion, there comes a time when you have to realise that you can't just respond to an argument without a real or considered point. Oxyman (talk) 02:01, 3 February 2017 (UTC)
@Oxyman:, I said "removing information" not WSC - I think your comment was aimed at my words, not his? Mistakes can and will happen, especially when misleading information is provided. The most important thing is to correct the errors: Please message me separately with a list of the files in this batch, and suggested new names and I will get that sorted.--Nilfanion (talk) 08:36, 3 February 2017 (UTC)
Taking the example Oxyman gives of File:Buckingham Palace, London.jpg It just so happens that I know that information to be incorrect and I know what that building is. So I have now renamed the file to get rid of the incorrect information "Buckingham Palace". But we are a global site with global subject matter, an editor unfamiliar or less familiar with the two buildings involved would be likely to categorise it by the filename as the only information they had - that image doesn't have a map coordinates. I wouldn't criticise someone else for categorising it as Buckingham Palace - in my view you should only remove information if you are very confident that it is incorrect. For that particular file the information Buckingham Palace now only exists in the history and in the redirect. If the filename for that picture had been "building in London" and the description "Buckingham Palace" I would have altered the description. I recently took several hundred images out of the category churches, in many cases I used the coordinates to simply add a country, US state or Russian oblast, sometimes I was able to do more and those I failed with I put into unidentified churches. WereSpielChequers (talk) 09:46, 4 February 2017 (UTC)
So another editor comes along and can see the file was not of Buckingham Palace by taking the time to look in the Buckingham Palace category and noticing that it is not the same building, the editor doesn't know what building the file shows, so they put it in an unidentified category at least until they work out what the building is that is in the file. Can't that editor expect a bit more effort taken to actually identify what the file shows before it is removed from the unidentified category? Rather then it just being removed in a batch edit using the Cat a lot tool purely on the basis of a filename, a filename that is known to be incorrect. I suppose the editor could risk getting involved in an edit war by reverting the ill considered edits as trying to point out the problem on your talk page is pointless. Oxyman (talk) 14:13, 4 February 2017 (UTC)
The key phrase here is "a filename that is known to be incorrect". Going back to the example you raised of File:Buckingham Palace, London.jpg - as soon as I saw that image I knew it was incorrect and I renamed it. From its history obviously several others had seen it and either did not know it was incorrect or did not have the userright that allows renaming. We would have a big problem if people knowingly left incorrect filenames here. I believe that incorrect filenames are fairly rare, but I don't underestimate the problems they might cause if they weren't corrected. WereSpielChequers (talk) 15:25, 4 February 2017 (UTC)
So what if you are unsure of the location..., renaming a file "unknown subject" doesn't seem to be a good use of filemove userights..., especially if you think given a bit more time you could figure out the location, that is why an unidentified category is useful. As for knowingly leaving incorrect filenames, I prioritised categorising images with no category into a relevant category, There has been a large batch of Panaramio images uploaded recently, many with no category and inaccurate or misleading information on them. I simply did not have time to fix all the details immediately, so started with the category. As for believing incorrect filenames are fairly rare, I had imagined you would have spent more time around here. Truth is filenames from some sources can be consistently inaccurate and I deal with them all the time, Calling "London" a "City of" is fairly common especially when they are coming from sources that do not use English as a first language. What is needed is to use a balanced approach to decide what information is correct especially with media that has been put in an unidentified category...,For a reason. Using a Cat a lot batch command based solely on filenames in this instance is not a balanced approach. Oxyman (talk) 17:04, 4 February 2017 (UTC)
Perhaps our differing experiences here lead to different expectations as to what is typical and what is rare or fairly rare. Much of what I have done has involved working with images from the Geograph, GLAMS and Wiki Loves Monuments and I'm pretty sure that the info with those images is usually correct. Perhaps our perceptions of the frequency of incorrect file names differs, or our uses of terms such as fairly rare differ. In a project of 36 million files evidence that 36 or even 360 meet a certain condition doesn't refute my statement that that condition is fairly rare, it even reinforces my position by ruling out descriptions such as "unheard of" or "impossible". Where we do seem to have a difference in editing philosophy is over what we do when we find information that we know to be incorrect. I'm inclined to prioritise fixing information I know to be incorrect because what I know to be incorrect others are likely to take on trust. It may be much easier to categorise an image than to correct incorrect information, but it is much more useful to do the thing that fewer can do - such as correcting information that you know how to correct. As for "Using a Cat a lot batch command based solely on filenames" When I use Cat a lot I use a combination of thumbnails and filenames each of which has to be individually ticked. Sometimes I can see that the thumbnail is incorrectly described by the filename and in those cases If I can fix things I do. WereSpielChequers (talk) 12:53, 5 February 2017 (UTC)
Projects like Geograph are mainly carried out by people who are interested in Geography, they usually consider the geographical information and other details. London is a Major city with a large tourist trade and which holds events which capture attention of people who are otherwise hardly interested in the city. It's not unusual for files to be uploaded with the only available information being "London". You are not considering the source of the files and expecting everyone to take the same amount of care when uploading as they do at the Geograph project or when they enter a competition. Category:Unidentified locations in London Is not a batch of files from the Geograph project or a competition, it is a category used when some difficulty has been found when trying to identify what location the file shows. Yes I know how Cat a lot works...You clearly did use a Cat a lot command based solely on filenames here, as you did not consider why the files were in an unidentified location category or look at the files long enough to notice that the buildings shown were not typical of those in the square mile. There was no information in those files other then the filename linking them to The City of London. So, What is your suggestion as to what should happen when the file name is known to be incorrect but a correct name is unknown? Oxyman (talk) 13:48, 5 February 2017 (UTC)
If you look again at those images I think you'll find that the only available information wasn't "London" it was "City of London", the much larger group where the only information was London I left in unidentified London. Looking at them in more detail I have tracked down six in terms of where the photographer was, and all six were taken in the City of London. Though admittedly three were views of the south Bank from the City of London (one of those contains Tower Bridge so arguably is partly in the city). Three others are definitely in the City of London this has a bollard in city of London colours, the rest look to me as being in the city but I can't quite name the places. The buildings are typical of the square mile, or at least the square mile as it now is, at least two include the distinctive new building at 20 Fenchurch street. I'm afraid that like it or not we now have to treat the Walkie Talkie as a building in the city of London - whether or not we consider it typical of the city. So the "City of London " information actually looks quite accurate, all six I can be sure of were taken by someone standing in the City of London. As for your question what should happen when the file name is known to be incorrect but a correct name is unknown? I'd say that unless we know what something is, it is difficult to be certain what it isn't. WereSpielChequers (talk) 20:01, 5 February 2017 (UTC)
OK Mr perfect, seriously as I have tried to explain earlier "Calling "London" a "City of" is fairly common especially when they are coming from sources that do not use English as a first language" CAN YOU ABSORB THIS IN FORMATION??? CAN YOU READ THIS INFORMATION??? If I now look at, Category:City of London I can see straight away that the fist 10 or so images shouldn't be there. The other images you talk about were already in Category:City of London they were just waiting for someone to identify their location, If you were able to locate the files then why not do so and then put them in the appropriate category. Rather then just carrying out vandalism with Cat a Lot. So you are taking a dumb cliché as a law to be unquestionably followed. If I found a ship labeled as a car or a train labeled as an aircraft I fink I can identify the ship as not being a car and the train as not being an aircraft. YOU see people who are less perfect then you can use common sense to do such things. If you are going to reply to a post at lest take 2 seconds to check what you are posting makes any sense, because your last post was unconsidered nonsense. Some people are ably to consider they might have made a mistake others are just unreasonable ego maniacs. In future I will just revert your vandalisms as reason can not be found on your talk page. Oxyman (talk) 20:31, 5 February 2017 (UTC)
You disputed my categorisation so I spent some of my time tracking down buildings such as 20 Fenchurch Street, I knew it by sight, I knew it was in the City of London, and because you queried what I'd done I tracked it down, added that category and responded to you. I'm also pretty sure that [File:Unidentified in the City of London 36.JPG]] is the bit of the Thames bank opposite Hays Galleria - that looks very much like a bit of HMS Belfast in the top of the image. As for whether this image needs to be in any category other than traffic bollards in the city of London, it seems we have a difference of opinion. I won't revert you on that one, but I suggest you be a little more cautious as to who and what you describe as vandalism. WereSpielChequers (talk) 20:47, 6 February 2017 (UTC)

Grade I listed buildings / wikidata integration[edit]

Hi Jonathan,

Good to see you today. The experimental templates I mentioned, trying to express the category specification in machine readable form, with a link to a query to see what items on Wikidata appear to match, can be found at the top of the category pages:

At the moment there are some problems with the query: the one for buildings and houses works, the one for churches and bridges times out; and they all seem to be taking longer than they ought to be. I'm not really sure why this is so, or what can be done; it's a known issue since 2015, but it's a key thing to get right if the search service is really to complement categories for content discovery under the Structured Data initiative. Very curiously, the churches one fails, while the buildings one works -- yet churches are a subset of buildings, so the latter query includes everything that should be found in the former. Sorted; now works so long as the order of the specifying properties is carefully hand-chosen for the most efficient search.

It's also not pulling up as many churches as it should do (though pulling up a couple more houses than the Commons cat) -- I think this is because a lot of churches haven't had parish information added yet; or the parishes haven't been identified as being in Buckinghamshire, so breaking the chain of everything to be found in Buckinghamshire. I'm not going to get into that yet, because there's discussion to be had on Wikidata as to whether or not to have separate items for parishes and villages of the same name (at the moment mostly not, in future probably yes), but once that's decided, it should be possible to systematically improve the data for all items located in the UK. So the completeness (at least for these categories) of the wikidata search is something that should probably get fixed really quite soon.

The other advantage of adding this header template to categories, giving a machine-interpretable category contents specification, is that it might help a bit with auto-categorisation. So, if one has an image of something with a known (and reasonably well-populated) Wikidata item (or a category of such images that one wants to slot into appropriate supercategories, systematic use of these kind of description tempates should then make it possible for a tool to start up at some high-up category, then iteratively get all the subcategories of it, then for each one write a WDQS query using the string in the template in the category to ask whether the item fits the specification (something much faster than a query for everything that does); and keep on going until it has built up a list of all the lowest subcategories that have conditions that the item fulfils -- so that one might get not just "Grade I listed houses in Bedfordshire", but also "G1 listed building built in the 1700s", "G1 houses open to the public" etc -- for whatever categories have been defined on Commons (so unpredictable!), so long as the data statements to check inclusion criteria have been filled out for the item on wikidata.

As for images & mapping, it should be easy enough to add those to the queries -- I'll update this paragraph with a prototype. Jheald (talk) 19:57, 12 February 2017 (UTC)

Here's an updated version of the query:
Hit the blue 'Run' button at the bottom to execute.
When it completes, as well as the table view, you can also select 'image grid' or 'map' using the 'Display' dropdown on the right between the query window and the results window. Jheald (talk) 21:06, 12 February 2017 (UTC)
(BTW clicking on any of the map dots brings up a little pop-up showing the other fields from the query -- ie the picture, the item name, etc. Jheald (talk) 22:08, 12 February 2017 (UTC)
For comparison, here's also essentially the same query run for places within 20km of Bedford: I can't yet put up a border of Bedfordshire on the map, but that will come. But it's the previous query above that is closer to what could be run generically for any category. Jheald (talk) 21:21, 12 February 2017 (UTC)
So which form of the template do you think would be a better "sell" to the Commons community, for a template that could generically be put on any category, for example Category:cat ?
  • This original simpler one:
  • or this more complicated one,
that includes images and the possibility for a map view? Jheald (talk) 21:55, 12 February 2017 (UTC)
(For the second hit "Edit on" at the very bottom right, then "Run", then "Display" -> "Image grid", to see all the selected pictures of kittehs :g ) Jheald (talk) 22:41, 12 February 2017 (UTC)
I went with the version including the column for coordinates. Written up on Commons Village Pump here. Jheald (talk) 12:18, 13 February 2017 (UTC)

BL maps[edit]

Regarding the BL maps, I created a page almost a year ago at Commons:British_Library/Mechanical_Curator_collection/georeferencing_campaign/next_steps. I haven't really looked over it since, but there's a to-do list of some immediate house-cleaning tasks at the bottom. Then the next thing is probably to finalise a design as to what the file-description pages should look like (what templates etc), and convert existing material over to it.

In terms of what is georeferenced, a live spreadsheet-like table can be viewed here; clicking on eg "country" gives a drop-down box with an option to find eg "India" and filter the set for it; then clicking on "scale" allows the maps to be sorted from the widest-scale (1) to most local (20). I think we have 290 for places in India at the moment. The whole table can also be downloaded as a spreadsheet; there's also a slightly extended version which includes the bounding-box co-ordinates.

Concerning the U.S. civil way, there's quite an extensive Category:Maps_of_the_American_Civil_War; but it's hard to see what is or isn't there.

Another approach might be a Wikidata query for the battles, to see which of those come up with maps; and to try to fish out maps from the BL set with appropriate coordinates -- but that might need a bit of work making sure the data base is populated; also there may be false positives, if we only know the location not the date depicted. One can have a browse on the spreadsheet by selecting eg country=United States, and look at various scales, eg scale 12.

Alternatively (and closer to what I think you were suggesting), we could extract the maps of various scales from books known to be about the Civil War (which we have tagged), then look for other maps with about the same centre and bounding boxes -- eg these at scale 12. So that could well be possible. Jheald (talk) 19:56, 12 February 2017 (UTC)

A query for what's currently on Wikidata for the U.S. Civil War, cf Category:American Civil War:
As well as the map and image grid views already seen, the dates on this query also allows a timeline view, which is worth a look. (Though this alternative version showing images (P18), rather than maps, is prettier: )
Data coverage for coordinates is pretty good; dates about 50/50; commons category: probably a lot missing; maps - only one added
This was a search for items marked as "part of" the Civil War -- one thing I can't tell is how much is missing because it hasn't been marked up in that way. (Though tools do exist to compare queries against category contents, which I might look at.) Jheald (talk) 23:24, 12 February 2017 (UTC)
Thanks James, an alternative approach would be to identify some categories of maps that enthusiasts are ready to categorise and digest and that cover an area where we have gaps. Then upload at least those subsets of old British Library maps. I'm guessing that our American civil War bods such as @GELongstreet: would be able to tell us whether those maps of civil war battles, sieges and campaigns were a treasure trove, similar to what they already had or redundant to modern scholarship. Equally some of our Indian colleagues might be able to tell us whether those old Indian maps and plans were interesting and relevant or redundant to stuff they already have. I have been out of the loop since my departure from WMUK, but I'm conscious that the Library like any GLAM involved in a release were interested in seeing use of this material, some examples of those maps and plans appearing in Wikipedia articles would be good. WereSpielChequers (talk) 15:18, 14 February 2017 (UTC)
Just for the records, I have little experience with maps and none with wikidata. Suppose for the wikidata stuff you go to the aforementioned civil war bods on the WikiProject Military history talk or, if you want a larger base of real civil war experts for the maps, you´ll leave wikipedia and go to ... GELongstreet (talk) 16:01, 14 February 2017 (UTC)