User talk:The Anome

This user is one of the 400 most active English Wikipedians of all time.

Page move

You have just moved a page to Charitable dispensaries in Manchester. That might be a bit unfortunate because they are rarely referred to as charitable dispensaries and in fact generally were subscription-based. Many became provident dispensaries and some, such as that at Ancoats, became hospitals. I've got a lot more to add to that and related articles but wonder if it might be best to stick to the original title. The period when these thing were formed is generally referred to as the "dispensary movement". - Sitush (talk) 14:43, 11 August 2013 (UTC)

Thanks for the feedback. You're right, Charitable dispensaries in Manchester is wrong for this. I've renamed the article to Dispensary movement in Manchester, which I hope is clearer than "dispensaries in Manchester", which might be taken as referring to present-day hospital dispensaries. -- The Anome (talk) 13:36, 12 August 2013 (UTC)

Yup, that makes sense. Thanks. I'll try to catch up with the writing soon - I've become a bit burned out (singed?) with events elsewhere. - Sitush (talk) 16:39, 12 August 2013 (UTC)

"FOO First Nation" item should not have "coord missing" added

I just reversed the addition of coord-missing to "Saddle Lake Cree Nation", which is not a place, it is a government. Individual reserves and communities of this government will have coords, but putting coords on government pages is not wanted or needed. I don't know how many such items you've added this to, if you did it on a global setting across certain categories, please reverse it so I and other's don't have to manually.Skookum1 (talk) 01:59, 15 August 2013 (UTC)

Hi -- thanks for spotting that and fixing it. The bot uses a variety of filters that inspect article content and categories to try to stop errors like this from happening: I'll take a look to see how this one slipped through. -- The Anome (talk) 09:21, 15 August 2013 (UTC)

OK, I've found why the bot tagged the article as a place: it was categorized as if it as was an article about a specific community, with no category that was sufficient to tell the bot that it was about a First Nation people, not a specific place. I've changed the category to reflect the fact that this is an article about a people, not a place. -- The Anome (talk) 09:27, 15 August 2013 (UTC)

Images missing by place

Hi,

It strikes me that we could reuse the logic you used in applying {{Coord missing}}, and add articles with coordinates, but no image, to a series of Category: Articles without images in [foo place]. People visiting or living in a county or city could then look at the local category to see what images are needed nearby. Someone could write an app for mobile users to do the same, based no the coordinates themselves and Wikipedia's "nearby" function. Would you be interested in that? I'll assist with grunt work. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:23, 17 August 2013 (UTC)

Yes, I would. The coord tagging has got a bit dull lately as the bot is running out of low-hanging fruit to tag, and I'm looking for more challenges. Certainly, almost anywhere coord-worthy is also image-worthy, with some obvious exceptions (historical events, subterranean geological features and crash sites come to mind, for example. Rather than create a new in-article template, I think we should use the existing {{image requested}} mechanism on talk pages to do this. There's a fair bit of infrastructure to create to make this happen: I will need to create:

• a classification of which types of articles are clearly image-worthy, using a whitelist, as opposed to a blacklist, approach
• a reliable way of detecting pages that are missing pictures (as opposed to maps, or icons such as flagicons or stub notice icons)
• a way of detecting the correct place to put an {{image requested}} tag on a talk page
• a way of working out what "in" parameter I should be using in the tag to bring it to the attention of relevant WikiProjects

There are also other projects we can liase with, to make it, for example, easier for volunteers to find nearby sites to photograph or hunt down public domain images.

I'll think about this some more, then get back to you.

-- The Anome (talk) 14:38, 17 August 2013 (UTC)

OK, that all sounds good. Please ping me when you're ready to move. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:39, 17 August 2013 (UTC)

Stalking you ;-) I saw User:The Anome/types of features eligible for photographs. How can I help? Why wouldn't it be easier to come up with a list of "things with coordinates not eligible for photographs"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:20, 18 August 2013 (UTC)

Given the number of articles that are eligible for images, I'd much rather take the cautious approach than be on the receiving end of hundreds of complaints for mis-tagging. Hence the use of a large whitelist of known good features, instead of a blacklist.

For an example of other kinds of things that need to be taken into account, see the Edray, West Virginia article that I found at random by clicking "random article" a few times until I found an eligible article. By looking at the output of http://wiki.riteme.site/w/index.php?title=Edray,_West_Virginia&action=render , you can see that it contains four images: two maps, a flag, and a dot. We need to be able to ascertain that none of these is a photograph before we decide to request one for the article. Now, we can try to do this by a number of different methods: analyzing the wikitext and template parameters, analyzing the HTML, looking at categories, looking at filenames, or by looking at the images themselves. In this case, for example, file analysis would show that two of the images are too small to be photos, and the other two are low-colour PNG images. However, just looking at the filenames of all four images in this particular would show them to be derived from original SVG images, making it unnecessary to look at the images at all. But that might not be true of other images; for example, some article might have a full colour image saved as a PNG or a GIF, and the only way to detect that would be to analyze its colour histogram. Or another article might use an image that is a full-colour JPEG photograph, but only use it as a tiny icon. Or an SVG might embed a bitmap image of a photograph. And so on... -- The Anome (talk) 10:49, 18 August 2013 (UTC)

After some more thought, I think I can work my way through much of this entirely from dumps, by combining information from the template links, image links, external links and category links tables to generate a shortlist of pages for more detailed inspection. Once that's done, I can then work on devising a set of more detailed checks to be performed on-line before making the final decision of whether to add a photo request template, what the content of that template should be, and where on the talk page to insert it. -- The Anome (talk) 12:25, 18 August 2013 (UTC)

Typically, an {{image requested}} would go just below the WikiProject banners, but outside the {{WPB}}{{WPBS}}, if present. Some banners provide parameters for the same purpose; for example, {{WikiProject Trains}} has |imageneeded= and |Imagedetails=, as seen at Talk:Brookfield (Cumbria) railway station. --Redrose64 (talk) 13:57, 18 August 2013 (UTC)

Yes. I think the best way to do this is to add {{image requested}} tags to articles first, then, with that work done, optimize later by adding project-specific parameters to WikiProject templates. Finding out what is and isn't a WikiProject banner, and dealing with the various grouping templates, is a subproject in itself.-- The Anome (talk) 17:10, 18 August 2013 (UTC)

Further to my comment above: I've remembered that there are some WikiProject banners where there is a |photo= parameter which doesn't work as you might expect (yes doesn't mean "yes, an image is needed"). See for example {{WikiProject Derbyshire}} where we find that this parameter may be set to |photo=yes if at least one photograph is included in this article; set |photo=na if this article does not require a photograph; set |photo=needs if this article needs at least one photograph to be added. --Redrose64 (talk) 16:29, 5 September 2013 (UTC)

Fair enough; is it OK to edit your whitelist, or would your prefer suggestions elsewhere? Or should I go away and stop bugging you? ;-) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:26, 18 August 2013 (UTC)

No, all suggestions are welcome! I've created a "put suggestions here" subheading on that page to put them under. Entries should be in the same format as the existing entries: words or phrases, exactly as used in existing article categories (and therefore typically plural), all lowercased, and with spaces replaced by underscores. -- The Anome (talk) 17:04, 18 August 2013 (UTC)

Thank you - added some, more later. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:49, 18 August 2013 (UTC)

@Lydia Pintscher (WMDE): made a suggestion:

You could check Wikidata for items that have a geocoordinate but no picture associated with them and generate your list. Then encourage building more templates that automatically use the picture if set. See http://ultimategerardm.blogspot.de/2013/08/the-occitan-wikipedia-and-wikidata.html for example.

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:23, 19 August 2013 (UTC)

Thanks. I've removed football clubs and organizations from your list: they are sufficiently blurry entities that they do not always have something that can be photographed. When they have identifiable headquarters buildings or home stadiums, with their own articles, those can have photographs added. I'm aware this will lose some opportunities to add photos: but I'm happier with having some false negatives, in order to prevent false positives which might cause more harm than good. You other suggestions are great, though -- please keep them coming. -- The Anome (talk) 13:24, 19 August 2013 (UTC)

I've also removed "schlosses" from the list. "Schloss" is just the German word for "castle/palace", and these are all, as far as I can see, listed under existing "castle/palace" categories such as Category:Castles in Germany. Category:Palaces in Germany. Category:Castles in Austria and Category:Palaces in Austria. -- The Anome (talk) 14:26, 19 August 2013 (UTC)

I see you points - sometimes reluctantly - with all those except football clubs, since they all have a strip which can be photographed; as can the games they play, and usually their home grounds. I can't see a false positive in such cases, let alone a harmful one. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:54, 19 August 2013 (UTC)

I take your point. You're quite right: there are categories of things which are photographable, but not geocodeable, and things like chain restuarants (for which a photo of a typical storefront might be appropriate) and football clubs might be among them. Let me think about it some more. -- The Anome (talk) 14:57, 19 August 2013 (UTC)

Football teams (generally) are geo-locatable; I added coordinates to Old Oscott FC yesterday. But if we're tagging things with coordinates only, that's not an issue, surely? Or were you thinking of tagging things with "needing photo and needing coordinates"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:35, 19 August 2013 (UTC)

I'm thinking of things that need photos but cannot be assigned coordinates. For example, a photo of the outside of a Pizza Hut branch would be quite appropriate for the Pizza Hut article, even though it's impossible to assign a single location to that organization; any reasonably typical example of their many branches would do. -- The Anome (talk) 18:06, 19 August 2013 (UTC)

Shops

Perhaps we need to revise our categories, and split those for "shopping chains" and "shop premises", or some such? Then again, if the article has coordinates, is it likely to be a chain?Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:41, 19 August 2013 (UTC)

Ideally, yes. Unfortunately, no-one has cared to make the distinction sufficiently often enough, and I've had real trouble creating sufficiently accurate heuristics to tell the two apart by inspecting other categories. So I'm going to err on the side of caution for these for now. We can always re-visit these later, if needed.

Also, thanks for spotting "cultural_sites". This has led me to find "historic_sites" as another marker of geocodable articles, and I've also back-ported this into the main bot to find new candidates for {{coord missing}}. -- The Anome (talk) 14:49, 19 August 2013 (UTC)

Well, I tried :-( Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:45, 23 August 2013 (UTC)

Progress

I now have a list of 1,004,783 articles which (a) contain valid {{coord}} or {{coord missing}} tags, directly or indirectly, and (b) do not have names that imply they are list articles. The next step is to use the category list to filter out types of articles for which there is nothing relevant to photograph (historical events, defunct buildings, electoral districts, etc. etc.). Then after that I can start working on how to automatically, and reliably, tell the difference between articles with and without pictures. More in a couple of days. -- The Anome (talk) 12:18, 21 August 2013 (UTC)

People can still add photographs of defunct buildings - they may be able to source them from their own or family archives, or talking to a GLAM, or searching the web for something freely licensed or out of copyright (e.g. St Alkmund's Church, Derby). Historic events might have a picture of the site, or a memorial, or a related object in a museum (a military uniform, say) or an artwork depicting the events (e.g. Peterloo Massacre). Electoral districts can benefit from a picture of a civic building, local view or even a polling station (e.g. Sutton Trinity (ward)). Thank you again for doing this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:16, 21 August 2013 (UTC)

Yes, I'm already thinking about it. See User:The Anome/geolocatable articles that are candidates for images, for some thoughts based on inspecting a random samples of 100 test pages, as well as some curious outliers found by bot-grepping the rendered HTML of some articles from a larger random sample of ~500 articles. -- The Anome (talk) 00:40, 22 August 2013 (UTC)

I doubt you meant to do this ;-) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:23, 22 August 2013 (UTC)

No, I certainly didn't -- no idea how your comment got deleted, I didn't spot any edit conflict when I made my edit. I've restored it below: -- The Anome (talk) 06:43, 23 August 2013 (UTC)

Progress indeed. I don't see why we'd exclude list articles - they can still benefit from having an image or two. Regarding 101 Collins Street, I think you can ignore any images which are in navbox templates. Per Radolfzell, you an ignore any image which is in an mbox. I'd include non-terrestrial objects, as NASA may well have PD images. Per my comments above, please do not have an exclusion date for disestablished venues. Apologies if I'm teaching you to suck eggs. If you need to run a trail on a geographic subset (and don't want to give away your location), please use West Midlands/ Warwickshire/ Staffordshire/ Worcestershire/ Shropshire, as that's my locality. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:52, 22 August 2013 (UTC)

Thanks. You're exactly right about the navbox images, and I've taken that on board. I'm still thinking about your other two points as part of investigating the other special cases. -- The Anome (talk) 09:45, 23 August 2013 (UTC)

More progress

I've now got a basic article-inspector working. Using some fairly cautious rules, I'm currently finding about 50% -- around 500,000 -- of the articles on my first list appear to lack photographs. There are still a quite a few false negatives, but I haven't yet come across any false positives so far. -- The Anome (talk) 00:45, 22 August 2013 (UTC)

After filtering down to only articles with a clearly identifiable country associated with them by Wikipedia category analysis so that they can each be unambiguously assigned to a Wikiproject, I'm now down to around 860k articles in my shortlist. I've also tweaked the article scanner a bit more, to be a bit more lenient in some cases where evidence from several different image tag parameters can be combined to rule out an image being a photo. Running the article scanner on a sample of those again suggests that roughly half -- about 430k -- articles are candidiates for photo requests. -- The Anome (talk) 12:49, 22 August 2013 (UTC)

More progress: most of the false positives are being generated by a few special cases, principally:

• images in navboxes -- solution: remove these entirely at the HTML parsing phase  Done
• tweak the logic for detecting stub icons to be a little bit more forgiving for some cases. Done
• images like flags, locator maps and coats of arms in infoboxes -- solution: itemizing the commonest infobox/parameter combinations which generate these, find these by parsing the wikitext, and remove the generated images from the list of images in the page before doing the rest of the analysis  Done
• remove PNG/GIF images within collapsed boxes -- under consideration

-- The Anome (talk) 19:21, 23 August 2013 (UTC)

The hit rate for finding articles which are eligible for images is now around 60% with the new tweaks, up from 50% before. I have yet to see any false positives. This gives 860k x 60% = approx 510k articles eligible to be tagged, and thus slightly more than offsets the filtering by country assignment. I'm sure there are still some false negatives, but we are probably near the point of diminishing returns in terms what can be done without starting doing image processing analysis on included images.

Next step: more hand checking for false positives, and any remaining low-hanging fruit on false negatives. -- The Anome (talk) 20:31, 23 August 2013 (UTC)

The HTML and wikitext analysis stuff is now finished: the only remaining errors I can find so far from reviewing its results are a few remaining false negatives caused by diagrams and maps included as images that would require image content analysis to distinguish from photographs, something which at the moment just isn't worth developing and debugging to get a couple percent higher hit rate. Enough information will be present in the bot's logs to revisit these pages later, if so desired. -- The Anome (talk) 14:04, 26 August 2013 (UTC)

This all sounds good - so what next? Apply for permission? Create some categories? Write some documentation? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:42, 27 August 2013 (UTC)

I'm now working on the final code to filter the articles by type -- building, populated place, etc. -- and to then find the right place in the talk page to add the {{reqphoto}} tag. Then it will be time to request permission from the bot managers to do a test run of say, 100 articles to start with: then 1000, then 10,000, then finally the whole 500,000. I don't plan to integrate the tagging into the parameters of project infoboxes just yet: I see that, and the subcategorization of place tags into subnational entities, as being a job for another bot pass later on. -- The Anome (talk) 21:05, 27 August 2013 (UTC)

This is continuing. Converting the various other categories into reqphoto categories alas requires manual construction of a mapping between the two. -- The Anome (talk) 12:56, 5 September 2013 (UTC)

How're things going? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:51, 12 September 2013 (UTC)

• ahem* ;-) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:59, 24 September 2013 (UTC)

Maintenance template dating issue

Please see this; my error, compounded by your bot. Is that something you can check for? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:51, 20 August 2013 (UTC)

Invitation to WikiProject Invention

Hello, The Anome. You are invited to join WikiProject Invention, a WikiProject and resource dedicated to improving Wikipedia's coverage of inventions and invention-related topics. To join the project, just add your name to the member list. Northamerica1000(talk) 09:24, 29 August 2013 (UTC)

Barnstar!

		What a Brilliant Idea Barnstar
		Don't know what prompted the creation of Catacomb saints, but what a great idea! I'll see what I can add to it - I seem to recall a particularly good story about such a "relic" being toured through nothern France during the 30 Years' War to boost morale among Catholic troops. Anyway, keep up the good work! Stalwart111 11:53, 7 September 2013 (UTC)

Ha ha, mutual barnstars! Thanks! I thought it might make for a good DYK candidate but couldn't think of a hook. Thoughts? Stalwart111 13:39, 8 September 2013 (UTC)

I've acknowledged it needs a better hook but I've nominated it at Template:Did you know nominations/Catacomb saints anyway. Stalwart111 06:56, 9 September 2013 (UTC)

Excellent. -- The Anome (talk) 08:29, 9 September 2013 (UTC)

re: Is there a scientific definition of "shake well"?

I was wasn't as careful about the way I phrased that sentence as I thought. So on second reading I did modify it here specifying shaking " ...the container". I was trying to avoid this. Tsk Tsk! - 220 of ^Borg 16:50, 13 September 2013 (UTC).

DYK for Catacomb saints

The DYK project (nominate) 08:03, 15 September 2013 (UTC)

Teahouse Invitation

Hello! The Anome, you are invited to join other new editors and friendly hosts in the Teahouse. The Teahouse is an awesome place to meet people, ask questions and learn more about Wikipedia. Please join us!

Tariqmudallal (talk) 15:32, 15 September 2013 (UTC)

Precious

missing coordinates and images
Thank you,second-wave Wikipedian, for quality articles such as Catacomb saints, for knowing how little we know on nouns and villages in Afghanistan, for "looking" for missing coordinates and images, for "I take your point. You're quite right" - you are an awesome Wikipedian!

--Gerda Arendt (talk) 09:10, 16 September 2013 (UTC)

Thank you! -- The Anome (talk) 09:14, 16 September 2013 (UTC)

A question about Amplituhedron

Hey bro. If you have any interest in the matter or know any Wikipedian who could help out, a question about Amplituhedron here (http://wiki.riteme.site/wiki/Wikipedia_talk:WikiProject_Physics#Amplituhedron.2C_Scattering_amplitude.2C_Scattering_patterns).

Thanks - 186.221.170.80 (talk) 20:31, 19 September 2013 (UTC)

India coords

Hi. I know it's been a long time, but I thought I'd mention it in case it might be a bigger problem...

With this edit, The Anomebot2 added coords to Bilaspur district, Himachal Pradesh. Unfortunately, there is also a Bilaspur district, Chhattisgarh and it is the co-ordinates for that district that were added instead. I don't know what the source data looked like, but if it did not include the adm1 division in which an adm2 is supposed to lie, it would be good to look for possible dups like this. There is at least one more such duplicate district (Hamirpur district), but it did not get the wrong coords (diff times, sources, etc. though). Anyway, FYI. —[AlanM1(talk)]— 01:00, 22 September 2013 (UTC)

Thanks for letting me know. This shouldn't happen: I'll take a look at the data and see if there's a way that this particular case could have been detected and prevented. Of course, if the source data is wrong, the GIGO principle applies and all bets are off. -- The Anome (talk) 14:32, 22 September 2013 (UTC)

U MAD

LOL

DA WIKI BOSS (talk) 19:26, 24 September 2013 (UTC)