Archival categories for featured picture nominations

@Alvesgaspar, Giles Laurent, Kiril Simeonovski, KTC, Julesvernex2, Rhododendrites, Yann, and everybody who is interested in statistics for featured picture candidates (FPC) and/or the maintenance of featured picture (FP) logs, galleries, and categories:

– TL;DR: Passages in boldface provide an executive summary ;–). –

Right now all FPC statistics must be compiled from the FPC log pages, and as recent discussions show, this is a laborious task that can lead to further discussions about details of the counting process. The chrological archives and the categories for featured pictures can help us with the successful candidates, but we are entirely dependent on the log pages to find the unsuccessful ones. Rhododendrites wrote ():

“No, we don’t put failed candidates in any particular category, and we don’t use categories on the nomination pages. IMO doing so would help these kinds of projects moving forward.”

So it would be very useful to have consistent archival categories for all featured picture nominations, including declined, withdrawn, and FPXed/FPDed nominations:

1. Categories would make it much easier to compile statistics, and they could also help to make the discussion about their evaluation more objective, because the figures underlying the statistics would be easily understandable for everyone.
2. Discussions about the meaning and implications of the statistical results would be facilitated if everyone could, for example, browse the unsuccessful nominations of certain years by calling up simple categories. And one can imagine other reasons why someone might want to review past nominations and would be much better supported by detailed categories than by the linear and somewhat clumsy log pages.
3. Such categories would also be very helpful for maintaining the featured pictures. I constantly scan the featured pictures for incorrect entries like fake featured pictures, using (among other things) the nomination pages as proof of authenticity. However, I have also found missing featured pictures that were simply forgotten or removed from the gallery pages due to errors. Categories for the nomination pages would allow me to analyze them much more effectively.
4. Sometimes log pages are damaged and therefore hard to understand (e.g., CommonsDelinker used to remove the inclusion links of FPC nominations; sometimes people rename nomination subpages without leaving a redirect or updating the inclusion link in the log page). Such damages can easily go unnoticed. Having archival categories in the FPC nomination pages itself would add an additional layer of safety through which nominations can be found, even if the log page gets damaged or if the nomination page was renamed.

Since it’s easiest to start with the new nominations, and to ensure we properly record all new nominations in the future, I have taught FPCBot to add two categories to every nomination page that gets closed:

(A) A category for the year and month in which the nomination was archived, like Category:January 2026 featured picture candidates.

(B) A category for the type and the final status of the nomination, like Category:2026 successful candidates for featured picture status or Category:2026 successful candidates for delisting from featured picture status.

The one kind of categories (A) corresponds to the traditional log pages. It complements the log pages and allows to search for nominations more easily by time frame (e.g., using a category search or PetScan). You can combine the category search with other criteria. These categories are especially useful for maintenance tasks (item 3 and 4 on the list above).

The other kind of categories (B) addresses the needs of statistical investigations (item 1 and 2 on the list above). They are organized by type and final status:

Type: We must distinguish clearly between new nominations for featured picture status and nominations for delisting from featured picture status, because any valid statistic must treat these two types separately. (E.g., when we discuss whether there are too many successful nominations or not, we refer only to the new nominations, so we must exclude the delisting nominations from our numbers.)

Final status: The categories distinguish between successful, unsuccessful, withdrawn, FPXed and FPDed nominations. I have used the neutral term “unsuccessful” to avoid the derogatory connotations of words like “failed”. We must also consider that some “unsuccessful” nominations didn’t get any oppose votes, but just not enough votes at all (they did not reach the quorum of 7 support votes). I have separated unsuccessful nominations from withdrawn, FPXed and FPDed nominations for three reasons:

(a) All of these candidates were not promoted, but for very different reasons – withdrawn nominations are cancelled by the nominators themselves, sometimes just because they are peeved, FPXed and FPDed are cancelled very early by other users (for formal reasons: FPD, for quality reasons: FPX); some of the withdrawn, FPXed and FPDed nominations could have been successful if they would not have been cancelled.

(b) The share of withdrawn nominations varies greatly and therefore represents a worthwhile subject for statistics.

(c) People doing statistics will probably hold different opinions about the question whether to include withdrawn nominations in the number of unsuccessful nominations or not. Therefore, the categories must record these nominations separately so that everyone can decide for themselves how to include their number in the statistics.

The long category names may seem a bit cumbersome at first glance, but they were intentionally designed this way to make them clear and 100% unambiguous and to eliminate any misunderstandings.

So how does this work? Head over to the root categories Category:Featured picture candidates by year and Category:Featured picture candidates by type. If you click on all the little collapse triangle icons in order show the subcategories, too, you can directly read the count of nominations in all the subcategories. This means: In the future we can compile most statistics just from the numbers on a single page! Both root categories give access to the type/status categories for each year, they are just organized differently. So depending on the way you want to compile your statistics, you can either start with Category:Featured picture candidates by year or with Category:Featured picture candidates by type.

There is one point which may raise objections. I have organized the type/status categories by year, not by month; i.e., there is no category for “Successful candidates for featured picture status from January 2026”, just for the whole year 2026. The reasons are simple:

1. People create FPC statistics mostly by year and rarely need to distiguish between the months of each year. Therefore it is much more practical to organize the type/status categories only by year: we can read the number of nominations for each type/status directly for the entire year, instead of having to add up the numbers for twelfe months first.
2. If we were to divide the type/status categories according to each month, we would get significantly more categories, but many of them would be almost empty, e.g. the categories for FPXed and FPDed nominations. That’s not exactly economic.
3. In order to use the archival categories for maintenance purposes, we will often need the monthly categories and they should include all nominations at once. So we would need even more categories.

Of course you can still get type/status numbers for each single month. Just combine the two kinds of categories in a single search, e.g.:

incategory:"January 2026 featured picture candidates" incategory:"2026 successful candidates for featured picture status"

Enter this search string on Special:Search, click on “Search in:” below of the search box, click on “Add namespaces…” and check the “Commons” item. Then you get this result, and in the upper right corner of the page you can immediately read the count of nominations: 166 results. You can just exchange the name of the month (and, optionally, the year) to get the number for any other month. – If you need the numbers for several months at once, even a very simple scipt can help you (contact me if you need some Python code to start with).

Please take a look at the two root categories (as described above) and check if the new archival categories are suitable for your use cases. If you would strongly prefer type/status categories per month, or if you see other serious flaws in the category scheme, please write a comment.

Of course, the new categories will only reach their full potential when we also apply them to the nominations from previous years. I have written a special bot program which reads the log pages, parses and analyses all nomination pages, and adds matching archival categories to them. (Unclear or problematic nominations are added to separate maintenance categories, so that we can examine and categorize them individually.) If we agree that the new archival categories are useful and that the category scheme works sufficiently well, I will create a bot request for the program and use it to categorize the old nominations.

Hope it helps, – Aristeas (talk) 14:48, 6 February 2026 (UTC)

• Thank you very much Aristeas, for the impressive job! I didn't even know a search tool was available in Commons... At first, I struggle with the search, because the numbers did not match my own, for January 2026. Then, I realized that there are no monthly categories for the successful nominations. If I understood well, you intend to put your "bot slaves" to work, in order to fill up those categories for all the years, right? When do you estimate the job will be done? -- Alvesgaspar (talk) 15:55, 6 February 2026 (UTC)
  • First I have to request the permission to use a bot program for this task and to wait for an approval; seems this can take a long time. Then I can start to process the FPC log pages, beginning with the most recent ones (2025) because they include mostly neat and tidy nominations, unlike the older ones. For every month, I must first check the log page for broken links and other obvious errors, and fix them – this must be done manually, doing this with a bot program would be too risky, and partly impossible. Then I can tell the bot program to process all nominations from that month, and check any nominations which were flagged by the bot as broken or unclear. Only then I can continue with the next month. As you see, this is an extensive project, and it may take several weeks or months until I reach the oldest nominations. I have tested my bot program with nominations back to 2010, but older nominations are often quite peculiar, so I will have to extend my program from time to time to cope with unexpected peculiarities. – Aristeas (talk) 20:26, 7 February 2026 (UTC)
• Thank you very much, Aristeas, for this impressive and carefully thought-out work. It clearly adds long-term value to the project by providing a solid, transparent basis for statistics and maintenance, and by making it much easier to distinguish assumptions from verifiable facts. This looks like a very reliable tool for the future. Best regards, -- Radomianin (talk) 20:37, 6 February 2026 (UTC)
  • Thank you! – Aristeas (talk) 21:04, 7 February 2026 (UTC)
• Fantastic stuff, thank you! No objections to having type/status categories by year and not by month, I think it makes sense. Once you're ready to move to the bot request, can you post the link here so we can follow along? --Julesvernex2 (talk) 09:59, 7 February 2026 (UTC)
  • I will do that when I’m ready, thank you! But I will wait a few days to see if other users raise objections. – Aristeas (talk) 20:41, 7 February 2026 (UTC)
    • @Julesvernex2 and whoever is interested: I have added a request for approval to the bot requests page. However please note that a bot request is “not a vote. It is a discussion”. So while you can state whether you consider the bot’s task useful or not, and can comment on details of the implementation etc., we can’t vote on the request; it will be discussed and then closed by a bureaucrat. – Aristeas (talk) 17:17, 17 February 2026 (UTC)

      Ups, Aristeas, sorry for voting instead of merely stating my opinion on the usefulness of the bot request, hopefully my position is clear nonetheless. Pinging @Alvesgaspar, Giles Laurent, Yann, Rhododendrites, Kiril Simeonovski and anyone else that may be following this discussion: if you have a chance, please leave your opinion on the usefulness of this bot (together with any feedback, if you have it), in the Commons:Bots/Requests/FPCArchivistBot page. My understanding is that this information will be useful for those that decide wether or not the bot can run. Julesvernex2 (talk) 17:25, 22 February 2026 (UTC)

• Amazing, many thanks for your hard work Aristeas! Since we are now making statistics, do you think you could also do a bot that could refresh the statistics from this page (and post it in a new official page)? I find the numbers there interesting -- Giles Laurent (talk) 18:00, 7 February 2026 (UTC)
  • Thank you! Well, that page is certainly very interesting, but honestly I do not sympathize with it – it emphasizes the personal success/failure ratios of nominators, and IIRC the discussions correctly this table was used to blame users for making stupid nominations etc. To be honest, this is not my style, so I leave this task to people who like ad hominem discussions. ;–) Sorry, – Aristeas (talk) 20:41, 7 February 2026 (UTC)

    Sorry, I didn't knew it was used for such purpose. I only thought it would have been interesting to see the evolution of ratios over time as people got better gear and experience over time (in line with the above discussions). But if the stats from the page are used for a bad purpose then it's probably better to not have a new one -- Giles Laurent (talk) 21:57, 7 February 2026 (UTC)

    No problem, Giles Laurent! I agree that the table is very interesting, but I suspect that updating it would not help to bring peace to our discussions, and at the same time I have to set a limit on what can do here without neglecting too many other duties, therefore I draw the line here. Best, – Aristeas (talk) 16:16, 8 February 2026 (UTC)

    I totally understand and also agree it is best to not update it -- Giles Laurent (talk) 16:45, 8 February 2026 (UTC)

• Thank you very much for the great work Aristeas. I think we are on a very good track to get useful data for better understanding the FPC process. I agree that generating those statistics on a monthly basis would be too arduous, so having them on annual basis is perfectly fine. If we want to be more thorough in our analyses, my suggestion would be to get data across different FP categories. Since pictures with different motifs are not comparable one to another, it would be highly useful to know what FP categories had the most nominations and what is the promotion rate in each of them (as a start, we could go with the general categories, such as "Animals", "Nature", "Places" or "People" from the template). --Kiril Simeonovski (talk) 07:51, 10 February 2026 (UTC)
  • Thank you, Kiril! Your idea to consider also the subject of the FPs is very good, and would certainly help us to gain additional insights. I will try if I can implement this without making things too complicated. Best, – Aristeas (talk)

    Kiril, I have implemented your idea for FPCBot. If the tests go well, I can deploy it soon. Thank you again for this excellent idea! – Aristeas (talk) 20:25, 10 February 2026 (UTC)

    Hint: This subject is continued below. – Aristeas (talk) 09:06, 18 February 2026 (UTC)

AI issues

Hi, everyone. I probably don't really owe anyone an explanation for why I have stopped judging photos, and I doubt most of you were waiting for one, but for whatever it's worth, here it is: I haven't stopped loving beautiful photos, but when even trusted regulars are using AI for editing, I feel like it's over. I can watch a video of a live performance and judge with a fairly high degree of confidence whether the personnel onstage really performed. I don't have the expertise in photography that might enable some of you to judge to what extent the person calling themselves the photographer is really the author of an image presented here. And this problem will get worse and worse in every field. I fully believe that not long in the future, no-one will be able to tell whether an image was genuinely photographed or not, and similarly, it will become impossible to determine whether music recorded beyond a certain date was really recorded by players and singers or fully AI. I have more thoughts about how AI will destroy human civilization if global warming or (less likely) nuclear war doesn't do so first, but that goes beyond the purview of this website. Anyway, it's been lovely being here and I wish all of you great photographers all the best as fellow artists. I at least am mostly a live musical performer, there's still a role for the human interaction a performer has with audiences, and as long as people actually have jobs and can pay for concerts, I'll be able to get some work. I fear that visual arts, especially digital ones, are facing a very troubled future. Sorry for the dark thoughts, and again, my very best wishes to you all. -- Ikan Kekek (talk) 08:17, 10 April 2026 (UTC)

Thank you for your long-standing and thoughtful contributions as a reviewer here, Ikan - they are deeply appreciated. Your judgments have consistently added real depth to the discussion, and for many of us - even those of us who are still relatively new - the critical perspective you brought has been an important reference point in evaluating images. I can only say that I find it genuinely regrettable that you are stepping back. Your voice has often been a decisive factor in how images are discussed and understood, and its absence will be felt, much like the departure of Daniel Case, which was also experienced by many here as a noticeable loss.

I understand and take seriously the concerns you raise about AI and its growing role in creative workflows. From my point of view, however, the responsible use of these tools is not necessarily the end of photographic authorship, but rather a continuation of long-standing practices of post-processing - now with more powerful instruments. Used transparently and with restraint, they can help reveal the potential already present in an image, for example in difficult low-light situations or high-ISO noise reduction, where the underlying photograph remains fundamentally authentic. For that reason, I strongly believe that transparency is essential. Any significant intervention - especially when AI-assisted tools are involved - should be clearly documented, ideally using a retouching template. This would not only maintain trust but also establish a shared standard of responsibility within the community.

Ultimately, what matters most is precisely the kind of careful, critical engagement you have provided. Whether one agrees with every individual judgment or not, such voices are essential for keeping a platform like this meaningful. I sincerely hope you may reconsider after some time away. Your contributions have been valued more than you perhaps realize. Kind regards, -- Radomianin (talk) 09:58, 10 April 2026 (UTC)

That's very kind of you. I may show up from time to time on COM:DR, where I might be able to help somewhat with the maintenance of this great repository of images. And perhaps I'll finally upload some of my cellphone photos of New Rochelle, which have mostly not been edited at all, and in some cases, just cropped. I just need the motivation. -- Ikan Kekek (talk) 10:42, 10 April 2026 (UTC)

• Thank you for your kind reply. It's good to hear you might still be around on COM:DR from time to time. I would also genuinely enjoy seeing your New Rochelle cellphone photos on Commons one day - sometimes unedited, straightforward images have a very special quality. -- Radomianin (talk) 11:28, 10 April 2026 (UTC)

Just popping in to say that I am finally getting close to finishing my images from Kenya last year, and when I do it is my intention to return. Daniel Case (talk) 15:25, 10 April 2026 (UTC)

• That's really wonderful to hear, Daniel - I'm genuinely very glad. Really looking forward to your Kenya images and your return. The "will" has been warmly noted :) -- Radomianin (talk) 17:20, 10 April 2026 (UTC)

Wait, why is there no requirement to declare image source (camera, scan, drawing, etc) and manipulation (like Photoshop, AI, etc) when nominating the work of an active user? - Alexis Jazz ^{ping plz} 13:03, 10 April 2026 (UTC)

@Ikan Kekek, I am sorry to hear of your departure. On your concerns about authenticity, coalitions like C2PA have metadata reflect manipulations to an image and allow validation. Photoshop often has C2PA enabled, and even when not, usually lists edits made. I would advise playing around with AI image generation/editing to get a sense of what the changes look like. JayCubby (talk) 19:42, 13 April 2026 (UTC)

JayCubby, I tested C2PA. After generating an image with AI, I opened it and saved it as a JPEG. The C2PA was gone. Virtually useless. Google's approach with SynthID is more resilient but (so far, AFAIK) can't be verified outside of an online service which isn't acceptable either. - Alexis Jazz ^{ping plz} 21:49, 13 April 2026 (UTC)

AJ, to be fair it is only metadata and can be easily removed. When present, it is quite tamper-resistant though. Assuming you generated it in an Adobe product, you may need to opt-in to C2PA. This tool works well for detection (and has a generous API). JayCubby (talk) 21:57, 13 April 2026 (UTC)

@Ikan Kekek: I’m glad to have gotten to know you and your open-minded, friendly, and constructive nature, and I hope to continue reading your posts and seeing your contributions in the future. I can understand your reasoning regarding AI. I can see the value in minor corrections using AI techniques, but in my opinion, images generated almost entirely by AI have no place on FPC. The enormous amount of electricity consumed by all these AI techniques—even by some search engines—is unacceptable. Perhaps it comes as little surprise that I’m also significantly scaling back many of my activities, such as the often fruitless discussions here on Commons, and refocusing more strongly on photography. So you’re not alone on this path. I wish you good light and plenty of motivation in your photography at all times. -- XRay 💬 20:19, 13 April 2026 (UTC)

Thank you very much. I've loved your photos, and I'm glad you continue to derive pleasure from depicting the world around us. -- Ikan Kekek (talk) 07:25, 15 April 2026 (UTC)

One possible partial solution for FPC would be to require access to the original RAW files for nominated photographs. Since Commons is not ideal for storing RAWs, those files could be uploaded to the Internet Archive, and the nomination could include a simple descriptive template linking to the RAW there. Also a Routers solution of years ago: https://petapixel.com/2015/11/18/reuters-issues-a-worldwide-ban-on-raw-photos/ Wilfredor (talk) 15:40, 18 April 2026 (UTC)

Minimum requirement of resolution or file size for FP nomination?

As for QI we have a minimum of 2MP, do we have anything for FP? If not can we have a specified requirement? As recently I have faced objections in FP nominations regarding small file size or low resolution, where as the image has passed QI standards. It would be easier for the participants if we have a guideline. Tisha Mukherjee (talk) 08:32, 14 April 2026 (UTC)

Hello Tisha Mukherjee, our Guidelines for nominators mention the same minimum: “Raster images of lower resolution than 2 million pixels (pixels, not bytes) are typically rejected unless there are strong mitigating reasons.” So we already have a guideline. – However voters will often require higher resolutions in order support a FP nomination; this is OK, people are free to apply higher standards when voting for featured picture candidates as long as (1) they explain their standards and (2) these standards are reasonable. Especially when it comes to photos of static subjects like landscape or architecture photographs, people today often require a high resolution; IIRC even photos with 12 megapixels are sometimes considered as rather low resolution today. For wildlife photographs, especially of small animals, people normally understand and accept that you cannot get too close to the animal and therefore still accept the 2 megapixels minimum, sometimes making even exceptions for especially difficult shots, according to the clause “unless there are strong mitigating reasons” from the guideline cited above. Best, – Aristeas (talk) 09:23, 14 April 2026 (UTC)

The only time I've seen an image be recently promoted below that threshold was for a historical (early 90s) digital photograph. With that aside, I think detail matters a bit more than pixel count. JayCubby (talk) 14:54, 14 April 2026 (UTC)

Good hint, Jay, thank you! I remembered that some time ago we promoted a photo by JJ Harrison, even though it was smaller than 2 MP, because the depicted species was extremely rare; but I can’t find such a FP, so I must have been mistaken (this rare picture has over 6 MP, this one has even 21 MP despite the extreme rarity; wildlife FPs < 2 MP like this one are apparently all very old FPs). So it seems we are not making exceptions for rare species anymore. – Aristeas (talk) 16:35, 14 April 2026 (UTC)

Upscaling guideline (not FPC-specific)

Hi all. There's a draft of a Commons guideline on upscaling at Commons:Upscaling. I'd like to propose promoting it to guideline status soon. Thought I'd drop a link here for feedback/edits, given upscaling has come up several times in this forum. Note that this guideline is applicable to Commons as a whole, and isn't specific to FPC. (I've already posted a link from the village pump FWIW). — Rhododendrites ^talk |  20:26, 22 April 2026 (UTC)

FPCBot asking for help

The bot has stopped because of an uncaught exception: 2026 featured picture candidates of <add the gallery' contains illegal char(s) '< (InvalidTitleError). Abbreviated stack trace:

Traceback (most recent call last):
  File "fpc.py", line 4712, in main
    _handle_task(task, which_types)
  File "fpc.py", line 4849, in _handle_task
    check_candidates(Candidate.park, CAND_LIST_PAGE_NAME, which_types)
  File "fpc.py", line 4072, in check_candidates
    check(candidate)
  File "fpc.py", line 2118, in park
    self.move_to_log(self._FAIL_KEYWORD)
  File "fpc.py", line 1742, in move_to_log
    self._add_archive_categories(status, now, gallery_link)
  File "fpc.py", line 1844, in _add_archive_categories
    self._create_subject_categories(year, subject_cat, subj_phrase, subj_key)
  File "fpc.py", line 1965, in _create_subject_categories
    if category_page.exists():
       ^^^^^^^^^^^^^^^^^^^^^^
  File "_basepage.py", line 798, in exists
    return self.pageid > 0
           ^^^^^^^^^^^
  File "_basepage.py", line 196, in pageid
    self.site.loadpageinfo(self)
    ^^^^^^^^^
  File "_basepage.py", line 141, in site
    return self._link.site
           ^^^^^^^^^^^^^^^
  File "_links.py", line 493, in site
    self.parse()
  File "_links.py", line 449, in parse
    raise InvalidTitleError(f'{t!r} contains illegal char(s) {m[0]!r}')
pywikibot.exceptions.InvalidTitleError: '2026 featured picture candidates of <add the gallery' contains illegal char(s) '<'

Developers, please look into this. Thank you! / FPCBot (talk) 05:07, 23 April 2026 (UTC)

•  Comment On review, I realized I had missed that the creator and uploader information was incomplete (see diff). I've since corrected this and added the missing gallery link, which should resolve the title parsing error. I suspect this nomination is the source of the issue. Could a developer or admin kindly restart the FPCBot to process the nomination? Many thanks in advance. Best regards, -- Radomianin (talk) 13:17, 23 April 2026 (UTC)
• Addendum: I am not sure whether the bot resumed automatically or was manually restarted. In any case, I am glad to see that it has now correctly processed the closed nomination after my correction :) Best, -- Radomianin (talk) 14:02, 23 April 2026 (UTC)

FPCBot asking for help

The gallery link in the nomination Commons:Featured picture candidates/File:Yellow mongoose - Kgalagadi National Park, South Africa.jpg points to the gallery page Commons:Featured pictures/Animals/Mammals/Carnivora, but does not specify the desired section on that page. Therefore one or more new featured pictures are added to the Unsorted section at the bottom of Commons:Featured pictures/Animals/Mammals/Carnivora. Please sort these images into the correct section. Thank you! / FPCBot (talk) 21:02, 23 April 2026 (UTC)

•  Done The image has been moved to the correct subsection …/Carnivora#Family : Herpestidae (Mongooses). Best, -- Radomianin (talk) 22:30, 23 April 2026 (UTC)

FPCBot asking for help

The gallery link in the nomination Commons:Featured picture candidates/File:Wisła wczesnym ranem poniżej zapory w Łączanach pod Krakowem, 20250105 0836 6943.jpg points to a heading on Commons:Featured pictures/Places/Natural, but that heading is not a valid target because it is not followed immediately by an associated <gallery> element. Perhaps this is a superordinate heading and the image should be added to one of its subsections; but to which one? Therefore one or more new featured pictures are added to the Unsorted section at the bottom of Commons:Featured pictures/Places/Natural. Please sort these images into the correct section. Thank you! / FPCBot (talk) 13:06, 25 April 2026 (UTC)