Commons:Village pump/Proposals/Archive/2022/09

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search

Supporting RTF and TXT files

Hello, wonderful people. I have noticed that commons.Wikimedia.org supports PDF files. Therefore, I suggest we add the support for RTF and TXT as well. What do you think? Jan Honvehlmann (talk) 21:23, 11 September 2022 (UTC)

 Oppose, "Commons does not generally host documents" per COM:FT#Textual formats.   — Jeff G. ツ please ping or talk to me 21:35, 11 September 2022 (UTC)
Oh, I did not see that yet. I thought that if TXT files have such a small size, there should not be a burden. But that's OK, since there is Archive.org. Thank you for your quick response. Jan Honvehlmann (talk) 21:40, 11 September 2022 (UTC)
@Jan Honvehlmann: You're welcome.   — Jeff G. ツ please ping or talk to me 21:42, 11 September 2022 (UTC)
The main point is that for textual content, there is usually another Wikimedia project that is better for the text (entered directly). PDFs are accepted since they can often serve as source documents for Wikisource, where the text gets extracted, but they are not meant as a means to author free content here. Is there a particular use case you had in mind? Carl Lindberg (talk) 00:45, 12 September 2022 (UTC)
Very old public domain text books, and also self-written instruction manuals. Jan Honvehlmann (talk) 11:52, 14 September 2022 (UTC)
@Jan Honvehlmann Public domain books is a prime example for when we would typically keep a pdf (or djvu) of the scanned book on Commons and a digital text on Wikisource. Self written manuals would probably fit best on Wikibooks. El Grafo (talk) 09:00, 15 September 2022 (UTC)
 Oppose per Jeff. –Davey2010Talk 21:42, 11 September 2022 (UTC)

Add yandex to reverse image search gadget

i propose adding yandex.com to MediaWiki:Gadget-GoogleImagesTineye.js. (the website link would be https://yandex.com/images/search?rpt=imageview&url= + image url.)

i learnt from lihkg forum that yandex can be quite useful sometimes when google gives you no clues. indeed, you can try for example File:04.08.21 OSKAR-9.jpg. neither google nor tineye can find it, but yandex can. RZuo (talk) 11:00, 13 September 2022 (UTC)

also, the gadget could be renamed to Gadget-ReverseImageSearch.js , reflecting its purpose regardless which websites are used. RZuo (talk) 11:05, 13 September 2022 (UTC)
 Support, the more the merrier. Optional abbreviations for each RIS service would help manage screen real estate.   — Jeff G. ツ please ping or talk to me 11:05, 13 September 2022 (UTC)
 Support neither Google nor Tineye cover everything. Maybe also consider adding Bing image search - it doesn't have as many features as others, but sometimes it finds things the others miss. El Grafo (talk) 08:46, 15 September 2022 (UTC)
 Weak support Some privacy issues (more with the browser than the web search engine?) but it's very useful for allowing easy searching of a subsection of an image. --Lord Belbury (talk) 11:24, 15 September 2022 (UTC)
 Support - I too have found Yandex to be extremely useful when Google or TinEye haven't been, The more help the better. –Davey2010Talk 01:13, 17 September 2022 (UTC)
 Support This would be a useful addition. De728631 (talk) 10:37, 17 September 2022 (UTC)

Tracking file usage on some Wikimedia chapter wikis

Over at Commons talk:Tracking external file usage I've posted a proposal to start tracking file usage on certain wikis run by Wikimedia chapters in the same way that I already do for Wikitech and the OpenStreetMap wiki. Please leave comments there. --bjh21 (talk) 15:18, 24 September 2022 (UTC)

Checkmark This section is resolved and can be archived. If you disagree, replace this template with your comment. proposal accepted, code deployed --El Grafo (talk) 13:16, 17 October 2022 (UTC)

Lost treasure prize

In line with what I saw in User:Mztourist's comment about uncategorized files, what about something like a "Lost treasure" (or "Hidden" or whatever) prize for finding an image from within Category:All media needing categories as of 2017 or earlier (say uploaded more than five years old to be more broad) than did not have categories, was not used anywhere that is a good quality image or file? I understand you will get a lot of people who will just fight to categorize and show off the few good images in the bunch but I'm hopeful there will be people who slog through the 70k in that category. I don't think this fits into the Quality imports or other things but more like a Wiki Love Itself? Better yet, this could help in the next campaign as a separate campaign focus than adding new media. -- Ricky81682 (talk) 07:37, 12 September 2022 (UTC)

@Ricky81682: Nice idea Mateusz Konieczny (talk) 11:05, 11 October 2022 (UTC)

Quality imports

Actually I want to propose this logo with inverted colours, namely a red seal with a green arrow.

In light of the above failed proposal to expand "Quality images" to include works imported made by non-contributors I would like to propose an alternative system where images originating from non-contributors can be noted for their quality.

This system would be known as "Quality imports" (COM:QIM) and would exist parallel to the existing system.

To quote myself from earlier: "This way we can preserve the unique incentive for photographers to have their own works be promoted to Quality Images and also create a separate incentive with a whole separate process for "Non-Wikimedia Commons originals" (or however we should refer to imported works). While both "Quality imports" and "Quality images" would refer to images, "Quality imports" would refer to images imported from somewhere else (the photographer's mobile telephone, laptop, desktop, camera, Etc.) The names would be recognisable enough to let people know which images were created specifically by Commonswiki photographers and which ones were merely imported by Commonswiki contributors from external sources. This would please those that want to keep QI exclusive and preserve the incentives for photographers and it would create more incentives for importers to find high quality images (due to the dopamine rush)."
Later the original proposer, user "Nosferattus" wrote: "Well I spend a lot of time and effort combing through thousands of wildlife images on iNaturalist and Flickr and picking the few that are really high quality to import into Commons. The fact that I can't get these images designated as "quality" and thus more likely to be found by reusers is very discouraging to me. Instead they are just lost in the sea of poor to mediocre Commons wildlife photos. So it feels like a waste of time. Lumping me in with promotional agencies feels even more discouraging. It seems like my contributions here are not really valued." These are actually good points.

TL;DR Create a system where imported images can be given "a seal of approval" similar to "Quality images".

--Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 17:00, 2 September 2022 (UTC)

Quality imports (Votes)

Quality imports (Discussion)

Stupidly enough I forgot to note the advantages for people, for Commonswiki contributors this can be a motivation to search for more images to import to then propose this for (some people really like to do these things and likely experience a dopamine rush), and for Commonswiki users (in the broad sense) this will make it easier to find images of high quality. Let's say that there are a thousand images from one museum (let's say that all these images were imported using Flickr2Commons), then it would be easier for re-users to find "the best" images if they are QIM ("Quality imports"). So both contributors and re-users win. --Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 18:19, 2 September 2022 (UTC)

  • @GPSLeo: , stricter standards could be set for imported images. But currently the category "Self-published works" includes the data "This category contains 33,627,949 files, which is 38.9% of all 86,439,822 files in Wikimedia Commons." which indicates that the difference is between 40% (forty percent) of the files and 60% (sixty percent) of the files. We could simply raise the standards of the "Quality imports" system to make sure that the system doesn't get overwhelmed. How would you suggest raising the standards so the system doesn't get overwhelmed?
I would think that "Quality imports" probably has a larger diversity of types of content of high quality (professionally produced videos, top quality graphic works, Etc.) So it could probably also be more easily split. I don't expect the system to immediately get flooded as "Quality images" has existed since mid-2006 and as far as I know most users who currently import images from external sources do not use the current quality accreditation system so I don't expect a huge influx of submissions immediately. Perhaps it could go "under the radar" for a couple of months to set up the system, or limit daily submissions initially until more volunteers participate. Otherwise I don't really see how we can have a separate system for re-users to find high quality files that aren't Self-published at the Wikimedia Commons. --Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 08:48, 6 September 2022 (UTC)

I think that @GPSLeo: made a great point. FP, QI and VI procedures are lengthy and bureaucratic, as a result, we have 16K FP (0.02 % of the content), 2K VI (less than 0.01 % of the content) and 301K QI (0.35 % of the content). In total less than 0.5 % of quality assessed content, which is an extremely low number. The less visible the system is, the less it is practically used. For example at enwiki 95% of the articles are quality assessed. I'm afraid that means the whole quality assessment system is flawed and creating yet another category with different rules certainly won't help. Jklamo (talk) 09:38, 6 September 2022 (UTC)

@Jklamo: , those numbers have more to do with the culture around the quality assessment than anything else, the English-language Wikipedia's "quality assessment" for Wikipedia projects for example is completely random, basically any member of a WikiProject can assess any page and even after large updates assessments can remain unchanged. In contrast, the Dutch-language Wikipedia completely abolished this form of "quality assessment" because its community voted against having random WikiProjects assess quality almost randomly for within their scope. These systems aren't even comparable as the quality assessment of the English-language Wikipedia is only for a few million articles with many more volunteer contributors than we have here. I agree that the current system is too bureaucratic at the Wikimedia Commons, but all this proposal does is allow for another category of quality assessment to be created, it will help make high quality files more discoverable.
If the visibility of the system is an issue then why not promote the quality assessment more in visible places? What do you find "too bureaucratic" about the current FP, QI, and VI procedures that can be improved?
I don't see how an improvement of these other procedures has to be mutually exclusive with the creation of another one, this is simply another venue to assess the quality of images that usually aren't recognised for their quality and may be difficult to find, imported images make up more than 60% (sixty percent) of all images at the Wikimedia Commons and the only way an imported image currently can have its quality assessed is if it gets nominated for FA (which has very strict standards), all this proposal does is lower the requirements for imported images to be assessed for their quality, shouldn't this be preferred if you think that there aren't enough assessed images? --Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 11:30, 6 September 2022 (UTC)
@Jklamo: I am confused why you at the same time complain that less than 0.5 % of Commons content is considered as premium/high-quality AND protest against making more content eligible to be recognised as high quality. Also, " enwiki 95% of the articles are quality assessed" - this assessment is pointless and useless anyway (except maybe as motivation for editors), while on Commons it would make easier to find high-quality images among thousands. Mateusz Konieczny (talk) 18:22, 7 September 2022 (UTC)

How about "professional media imports"?

Some valid points have been made above about criteria and workload. How about we take a step back and think about alternatives. What kind of badge/filter would actually be useful for a re-user? We have FP for awesome images, VI for useful ones and QI for OK images taken by (mostly) amateurs. This may be a bit of a hot take, but I think it would be very useful to have a filter for professional shots too. "Professional" media could be defined strictly as the work of professionals, i.e. people who got paid to take the photograph, shoot the video, plot a figure (making the whole thing less subjective). There would be explixitely no assessment for quality (hence less workload), only the hope that professionals deliver professional work (well knowing that that is of course not always the case). There could be a process for batch-assessing entire categories (such as Category:Photographs by Ansel Adams) in one go. The would be made once per person or institution (NASA, Google Art project, ...) and then applied to existing files and later uploads. It wouldn't be a badge of honor, just a useful filter for re-users. If "professional" sounds like too much of an endorsement we can of course think about another name for the whole thing. Just an idea ... --El Grafo (talk) 07:49, 7 September 2022 (UTC)

Decentralized process using categories and reviewers

The QIC process is sort of functional, but I don't think many people would say it's ideal such that a new process should just duplicate it. I think King of Hearts was onto something in the section above with the idea of experienced reviewers able to browse for images and use their judgment to decide it's a "good image" (or whatever we want to call it). That may be reasonable, but I envision something a little different: a decentralized, categories and filepage-based process requiring two reviews that can happen at any point, rather than in a 7-day window, with a mechanism to allow non-reviewers to suggest images for review. So someone tags an image as a potential good image, placing it into a category for review, a reviewer sees it in that category and reviews it (or reviews any other image they see around the project), and if they approve it, it goes into a second category for images needing a second review. Since this is a process that's focused on results for end users rather than Commons contributors, a second review ensures it can be decentralized while maintaining quality control (to avoid friends reviewing friends' uploads, avoid potential damage done by an incompetent reviewer, erring on the side of higher quality, removing the need for elaborate consensual reviews, etc.). Worth considering? — Rhododendrites talk |  17:11, 7 September 2022 (UTC)

I don't see the need for a second review, which makes the process more work than QIC. Instead one review should be enough, and instead open up a page for challenging reviewed images after the fact. -- King of ♥ ♦ ♣ ♠ 17:38, 7 September 2022 (UTC)
@Rhododendrites: , alternatively we could create a system called "Endorsements" where if 3 (three) or more distinct users note an image as being "of good quality" that it gets upgraded to a quality image. This could be automatically counted on the file's talk page and to avoid vote spamming an account would have to at least be 6 (six) months old to vote. I don't think that counting a number of contributions would be relevant but I can see how such a system would be immediately more beneficial than creating yet another bureaucracy. --Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 21:44, 7 September 2022 (UTC)
In the proposal I voted  Support} but also said that we had an influx of 100k high quality scans from one bot alone in the last months. @Gzen92: did really good work with their Gallica bot. Many thanks for lots of great maps to play with! For review purposes, we can hardly process so many good images in the constraints of a workflow where a user suggests a candidate, then other users must vote in the same week for that same candidate, and then the candidate gets a quality seal. That's impractical for the most basic review where the question is just "is this a good upload". I'd suggest instead that we make use of talk/discussion pages for each picture: Eligible users would then be able to add "I think this is a quality picture/scan/upload" om the talk page. The file gets an approval once at least three people not including the original uploader have voted in favor; and the pro-side must outnumber the contra-side by at least 2:1. (So an image with 5 upvotes and 3 downvotes needs another upvote to be considered "quality"). No time limits involved. We have similar processes at work in WikiSource, where two independent users need to proofread the text scans; why would such a process not work here? We can and should still have the centralized, individual reviews to crown the very best files, of course. --Enyavar (talk) 06:46, 9 September 2022 (UTC)
Assuming that a system based on file talk pages / categories is implemented, sub-categories should be used for mass nominations, unless the number of nominated files per nominator and time period is limited to, e.g., 5 or 10 per day. Otherwise, manual assessment would be close to impossible because the respective categories would be flooded. Batch promotion of files looks rather meaningless to me. --Robert Flogaus-Faust (talk) 10:47, 17 September 2022 (UTC)

I suggest a simple process, as suggested by KoH above: one single vote limited to users with at least 3 months and 100 edits on Commons. Yann (talk) 19:20, 13 September 2022 (UTC)

Yeah simple is important and my own suggestion was more complicated, but I do think we should require four eyes to review pictures, and not allow self-nominations. With at least two people as reviewers, we have a better threshold to prevent abuse of the ability to singlehandedly say "my upload is quality". Best, Enyavar (talk) 20:39, 13 September 2022 (UTC)

Follow-up questions

It seems like there is a good deal of support to implement a very simple Quality Imports system using a template/category system. Under such a system, someone would nominate a file by applying a template, which would assign the file to Category:Quality imports candidates. The file would then need to be endorsed by a certain number of reviewers (probably by adding a comment and signature to a template parameter). There are a few unresolved questions that need to be answered before finalizing this proposal. Nosferattus (talk) 18:47, 11 October 2022 (UTC)

@Donald Trung, Enyavar, Yann, Robert Flogaus-Faust, King of Hearts, Rhododendrites, El Grafo, Mateusz Konieczny, Jklamo, Kritzolina, Infrogmation, Strakhov, Mike Peel, and Ikan Kekek: Pinging some folks involved in the discussion above. Nosferattus (talk) 18:48, 11 October 2022 (UTC)

Can someone nominate their own import/upload?

How many endorsements (not including the nominator) are required?

  • I would say only 1, but the file would need to have greater than 2/3rds support (including the nominator), so if the file had 1 counter-endorsement, it would need at least 2 endorsements (which would give it 75% support), etc. I think that would make for the simplest system that still adheres to a semi-consensus model. If the level of support for the file drops below 51%, it would be automatically moved to a different category like Category:Unsuccessful quality images candidates in order to keep the main candidates category uncluttered. Nosferattus (talk) 18:27, 11 October 2022 (UTC)
  • Nosferratus suggestion of a 2/3 support systems seems reasonable, but the suggested automatism needs some tweaking. "If the Level of support drops below 51" is something of a burden: Just one early negative review in the first minute is sufficient to immediately close the candidature as unsuccessful. As a result, I would suggest a system were unreviewed files are in a preliminary category. Once there are at least two supports it is moved to "QImp" until support drops below the threshold and it goes back to the preliminary category. The "Unsuccessful" status is for files that have been in the preliminary category for at least two weeks and that are below 50.1% support. --Enyavar (talk) 14:22, 12 October 2022 (UTC)

Should there be any time limit?

  • No, the beauty of a template/category system is that it can always reflect the current consensus without having to have cumbersome processes for nominating, promoting, and demoting (as long as the template supports both endorsements and counter-endorsements). Nosferattus (talk) 18:46, 11 October 2022 (UTC)
  • As above: the "unsuccessful" status should only get assigned after at least 1-2 week(s) passed by, not immediately after a first countervote. --Enyavar (talk) 14:22, 12 October 2022 (UTC)

Discussion: Quality imports follow-up

Thanks for the ping, Nosferattus. Why is the proposed setup so different from COM:QIC? QIC requires a simple majority for promotion and has some time limits. I don't even understand how a system without time limits would work: in the case of a contested nomination, when would a decision be arrived at if there were no time limit for the debate? -- Ikan Kekek (talk) 22:27, 11 October 2022 (UTC)

@Ikan Kekek: All the decisions would happen automatically as soon as anyone added an endorsement or counter-endorsement. Here's an example:
  1. Someone nominates a file for Quality Imports by adding the template to the file page. This automatically puts the image in Category:Quality imports candidates and the template says that the file is a Quality imports candidate.
  2. Someone sees the file in Category:Quality images candidates and adds their endorsement to the template on the file page (eventually this would probably be done by a gadget from the category, but initially it would be done manually). The file then automatically moves to Category:Quality imports and the template says that it's a Quality import.
  3. Someone sees the file in Category:Quality imports and thinks it isn't good enough to be a Quality Import. They add a counter-endorsement to the template on the file page. The file then has 66% support and automatically moves back to Category:Quality imports candidates and the template goes back to saying that it's a Quality imports candidate.
  4. Another person sees the file in Category:Quality images candidates and thinks it isn't high quality. They add another counter-endorsement to the template on the file page. The file then has 50% support and automatically moves to Category:Failed Quality imports candidates.
The file could continue moving between categories indefinitely as the consensus changes. Does that make sense? It's a very different system, but also very lightweight and flexible. It's more of a passive continuous assessment system than an active time-limited system. Nosferattus (talk) 03:32, 12 October 2022 (UTC)
I understand. -- Ikan Kekek (talk) 05:47, 12 October 2022 (UTC)

Thanks for keeping this going! I like this idea, it seems lightweight and flexible indeed. It might even be possible for this to work without relying on bots, which I find particularly enticing given the frequent problems with them elsewhere. Some points to consider, though:

  1. What does this mean for potential manipulation? At QI, VI, & FP, there are many eyes watching the votes on a central page, so faking votes requires elaborated sock-puppetry. De-centralized votes would make it much easier to fly under the radar. How would we even keep track of the 5 nominations per day rule? On the other hand, there would probably be much less incitement to fake votes in the first place, since by definition this would not be about people's own work - that should hopefully make it less attractive for badge collectors.
  2. I'm not sure file description pages are a good place to have a vote. 1) I would rather not spam the source code of file description pages with even more stuff. 2) The chances of newbies accidentally breaking the page when attempting to add a vote seems a bit high for me too. (changes may go unnoticed, see above) 3) File discussion pages seem like the logical place for this kind of thing. But that would raise new technical challenges and take away from the simplicity of the proposed approach.

On a more general note, isn't it a bit weird that all kinds of votes and polls are such an integral part of so many processes in the communities and we're still messing around with templates like it's 2008? --El Grafo (talk) 08:34, 12 October 2022 (UTC)

On manipulation: I can see how and why "Featured images" is viewed as a contest of skill in creation. By contrast, uploading a thousand QImp is a lot of work, but doesn't require any skill by the uploader, aside from steadiness and a good eye for selecting files. Anyway: Our whole worry should not be about the uploaders or creators, but about delivering quality content to the readers.
On the right place to vote: en-WP handles their whole quality-assessment scheme on the talk pages. Most file-talk pages are currently unused anyway, so let's use the talk pages. Not the description, for exactly the reasons El Grafo mentioned - unless we want to start a whole new graphical interface to handle it.
On the general note of using templates: I would love a simple graphical interface on each picture's presentation page: Open the "More" menu, select "assess quality", and assign 0-5 stars for the picture, maybe in the two different categories "technical aspects" and "informational content". (to clarify: no, I don't think this would be advisable to implement) --Enyavar (talk) 14:22, 12 October 2022 (UTC)

Glad to see this picked back up. A concern about implementation on the user-end: by the time someone gets to a specific file page, it isn't difficult for them to determine the quality of that image. The utility of such a process is in finding the best images from among a group. We would, of course, need to update Special:MediaSearch to include this new designation, but MediaSearch is insufficient. We primarily organize by category, and the tool we rely on to identify good images from category pages, Help:FastCCI, is broken more often than not (and has been, consistently, for several years now). I'm wary of adding any new technologies of evaluation (like a star system) that aren't built directly into the system (i.e. which don't rely on more not-yet-defunct tools that rely on a single volunteer), and have limited expectations for how useful any such system can be without effective discovery and sorting systems. That doesn't mean this idea isn't worth running with, but I suspect it will turn into another inward-facing process like QIC -- something more for our community than for readers. — Rhododendrites talk |  19:22, 13 October 2022 (UTC)

Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons:Village_pump/Proposals/Archive/2022/09&oldid=705473542"