Misplaced Pages

Revision as of 22:33, 3 February 2020

• WT:TFA

Archives

1, 2, 3, 4, 5, 6, 7, 8, 9, 10 11, 12, 13, 14, 15, 16

This page has archives. Sections older than 21 days may be automatically archived by Lowercase sigmabot III when more than 3 sections are present.

To notify all the TFA coordinators, use {{@TFA}} in your message, and sign it.

the calais entry... um....

the entry on Battle of Calais seems a little, er, dark, doesn't it? can we please use some entries that are a little more festive? something about traditional celebrations, or maybe just some facts about faraway lands? or maybe just an article on champagne? please? just a thought. --Sm8900 (talk) 01:27, 1 January 2020 (UTC)

Sm8900, you get champagne to featured status, and almost guaranteed it will run TFA Jan 1, 2021. SandyGeorgia (Talk) 02:02, 1 January 2020 (UTC)

lol ok, thanks SandyGeorgia!!!! I don't know if I can quite pull that off, but it's sure a nice idea!!! happy new year!!!! --Sm8900 (talk) 02:04, 1 January 2020 (UTC)

It could be fun! I used to regularly suggest that we needed to bring Casu marzu to featured status for April Fools. Champagne has a very good start-- get a team together and do it! We had more fun putting together Ima Hogg for April Fools TFA than a bunch of pigs in mud ... SandyGeorgia (Talk) 02:11, 1 January 2020 (UTC)

SandyGeorgia, sounds good to me!!! hopefully, we can start a new tradition, to post items that are actually festive here, for TFA items for New Year's Day, and similar occasions! --Sm8900 (talk) 15:24, 6 January 2020 (UTC)

some options:

Party
Champagne
New Years Day
Holiday
etc etc. --Sm8900 (talk) 15:25, 6 January 2020 (UTC)

This'll be good news for the future. In the meantime, the battle took place on 1 January (—and you think your new year fell flat!), and for most historical events, if there's a known and relevant date, posting on any other than the anniversary would be frankly bizarre. ——SN54129 16:14, 6 January 2020 (UTC)

Disagree here. Having been part (from the beginning) of how we ended up with TFA selection so defined around dates, I think we need to drop this outdated selection method. Our readers don't give a whatever that this battle occurred on 1 January, in the vein of what Sm8900 points out. We originally had date relevance as part of a point system because people were clamoring for and competing to be on the main page, so one part of the selection gave them points for dates. We no longer have FA writers clamoring for a spot as we did a decade ago, and nonetheless, the date selection portion has become a defacto path to TFA. Sm8900's point is well taken. And some of our date-selected items leave a lot to be desired. We should drop this archaic practice in favor of something that furthers goals other than "what FA writers want". We should be giving our readers what they would like to see, or doing something useful to further FA involvement. (Hint, see section above, and decide what our priorities are.) SandyGeorgia (Talk) 16:45, 6 January 2020 (UTC)

Sometimes I wonder if making TFA display FAs at random order (possibly excluding FAs that already ran, "blacklisted" articles) might be an idea. Jo-Jo Eumerus (talk) 16:53, 6 January 2020 (UTC)

In times past, that idea was rejected for reasons ... related to times past :) I wouldn't so casually reject the idea today, all things we have been discussing at WP:FAC considered. That idea could kill many birds with one stone (and accomplish the same as ... see section above). Again, catering to what FA writers want for TFA is not advancing goals that need to be advanced at this juncture in the history of the FA process. IMO. SandyGeorgia (Talk) 17:12, 6 January 2020 (UTC)

Alternate TFA proposal

Or if you want to be even more radical, abolish TFA altogether and use {{random subpage}} to display a different randomly-chosen FA to every reader each time they visit the main page. (We have the technology to do this now and it wouldn't need any special coding; repeatedly purge my talkpage and watch the image at the top change for a relatively trivial example of random subpages in action.) It would be a pain as it would mean every FA needing a blurb, but it would put an immediate stop to disputes over scheduling, would abolish the need for WP:ERRORS to even exist, would end the unpleasantness of "main page day" since the vandalism and Randyism would be distributed over 5000+ articles instead of concentrated on just one, and would be a powerful tool for highlighting FAs that are no longer adequate since there would be no place to avoid scrutiny. For those who measure success in terms of pageviews it would also be a tangible reward for shepherding an article to FA status and maintaining it, since it would provide a measurable and consistent uptick in page views for every page with FA status. ‑ Iridescent 23:14, 11 January 2020 (UTC)

Iridescent, in the environment ten years ago, I would have so strenuously objected to this. And today, it seems perfectly reasonable! Every FA up to snuff, or to FAR. But ... the idea of generating a blurb for over 5,000 FAs just isn't doable. Is it? SandyGeorgia (Talk) 23:19, 11 January 2020 (UTC)

Rotate those that have blurbs, and as blurbs are written for new and old FAs, add them on. Possibly some deficient FAs never have blurbs written.--Wehwalt (talk) 23:24, 11 January 2020 (UTC)

(edit conflict) Dank is better placed than me to answer in terms of how much work it would take as he's been working backwards through the FAs in order of promotion writing rough-draft blurbs based on the lead (this one of mine just popped up on my watchlist today). It would be easy enough to institute a "they don't go in the mix until the blurb is written" rule and just gradually work through WP:Featured Articles without a blurb as and when anyone felt like it. ‑ Iridescent 23:28, 11 January 2020 (UTC)

I won't throw a fit if the TFA selection job goes away... Ealdgyth - Talk 00:00, 12 January 2020 (UTC)

Thinking through the practicalities, really liking this. How about if we let it percolate here for a few days to a week, and then put together a formal proposal over at WT:FAC? Theoretically we already have blurbs for all old FAs, so it's not as insurmountable as we might think. Running ALL FAs with ready blurbs would so deal with the FAR problem that concerns me: "bad" and "old" FAs would be up there right along with the "new" and "good", and the "bad" would reinvigorate FAR, eliminating the need for my "sweeps" proposal. But we would still need TFA Coords. For example, what if we wanted to override the randomizer for VERY special occasions ... remember when we ran BOTH Obama and McCain? We should still have a place and people (TFAR) for ultra-special occasions. So, Ealdgyth could focus on FAC, while Wehwalt and Dank held down the fort at TFA. I am really warming to this idea. BUT ... would it discourage current FA writers ? SandyGeorgia (Talk) 00:35, 12 January 2020 (UTC)

This might be a spot to worry about performance. Given the quantity of users who view the main page, I expect that a randomizer (which would I am fairly certain would kill the cache), might make for some very sad engineers at the WMF. --Izno (talk) 01:41, 12 January 2020 (UTC)

We've used randomizer scripts in the TFA slot before (e.g. the Middle Ages TFA where nobody could agree on the best illustration, the 2008 US election where we rewrote the code to shuffle the order of the candidates, or the "triangular constellations" TFA which randomised which order they appeared in). On each occasion the server has survived. As I understand it, they don't replace the cache for a truly random experience, but refresh what's displayed each time the cache is purged. ‑ Iridescent 07:11, 12 January 2020 (UTC)

That was 2/3 pages. This would be 5000 pages. --Izno (talk) 16:08, 12 January 2020 (UTC)

I'm not sure what I think of the proposal; I'd like to hear first from a bunch of featured article writers and people who work on the Main Page. - Dank (push to talk) 03:25, 12 January 2020 (UTC)

Regarding Izno's concern, let's see what Whatamidoing (WMF) has to say about this. I do plan to write more FAs in the future and already have some under my belt, I don't think it's a problem to have random FAs being run at TFA. Of course we'd need blurbs ready for all, and some might get picked twice in a short timespan, but the latter isn't a "problem" really. Jo-Jo Eumerus (talk) 10:13, 12 January 2020 (UTC)

Let's ask User:Krinkle about the performance question. It would presumably matter whether it's actually random-per-page view, or just changes several times a day (like DYKs). Whatamidoing (WMF) (talk) 23:51, 13 January 2020 (UTC)

Hey, good to see you around these parts, WAID. Guys, John and I are going to resume doing older blurbs (but that has nothing to do with this "randomization" idea). Note that comments are welcome from anyone for any of these blurbs. You can easily find all the blurb reviews for, say, 2017 by looking at the talk pages of the FACs that don't have a TFA date listed at WP:FA2017. - Dank (push to talk) 14:50, 14 January 2020 (UTC)

A few concerns that I see:

1. Is the featured article literally supposed to change by just refreshing the page, or is a few minutes of caching OK? I don't think it is wise to disable Main Page caching completely.
2. Is the featured article supposed to be the same random page for everyone every X minutes or can it vary depending on the user? Users with different parser options might see a different featured article. Logged-out users might see a different featured article too. This is due to fragmentation of the cache and unaligned cache TTLs. If a deterministic hashing of (year/month/day/hour => featured article) was made using a seed stored on a page somewhere (updated now-and-then manually), then use of CURRENTHOUR/CURRENTDAY/CURRENTYEAR could avoid this problem. I would have to change Parser::expandMagicVariable() to pick smarter TTLs than hard-coded MagicWordFactory value used now (1 hour). Ideally, it would be a TTL such that the content would expire "a few" seconds after the CURRENT* magic word output will change. The delay between the frozen "parse time" and parser cache save should accounted for too (since the clock doesn't stop ticking while parsing continues).
3. I assume cascading protection would still be enabled, right? If there was no caching at all, a lot of jobs would fight over each other adding and removing different blurb pages from the outgoing links from the Main Page. With the caching idea above, this would be less problematic.

Another option is a bot that randomly picks FA blurb pages and chooses one to manually transclude onto a "current blurb" template, itself transcluded onto the main page. This would require Admin rights and trust that the randomization is not post-hoc manipulated to justify the bot owner's favorite blurb. Maybe Lua could be used to pick the FA blurb page using the list of blurbs and a seed (build from three pages with partial seed numbers updated by bots with different owners). From the server-side, it's easier to handling caching when soft state is avoided, with changing happening explicitly by users/bots (even if they are just doing semi-periodic stuff randomly). For example, if page Q makes use of transcluded seed page P via Lua, and someone edits P, then Q simply gets updated like when any other template is edited. The updates are queued in the background and wait on database replication to reduce glitches (e.g. RefreshLinksJob::run). On the other hand, when rendering depends on what minute of the day it is (throw in clock skew and time-of-check to time-of-use hazards), it's a different kind of trickiness.

I also wonder if a new TTL-optimized magic word would be useful, e.g. using an HMAC of (current UTC year/day/hour, stable server side-secret).

Aaron Schulz 04:24, 24 January 2020 (UTC)

I may have missed something here, but what's the provision for date-linked FAs, linked to specific events (eg moon landing) or anniversaries of people? Many FA writers specifically request that their article runs on eg the 100th anniversary of a birth. In January, 8 of 31 TFAs were linked to a specific date, starting with the Battle of Calais on Jan 1, and I'd guess that's below the average per month. Jimfbleak - talk to me? 06:36, 24 January 2020 (UTC)

My comment which prompted this particular sidetrack was in response to SandyGeorgia's point above that we should be looking for a way to discourage the kind of abuse of "date relevance" that currently takes place. Looking over the eight "date significance" TFAs for January, we have:

• the 670th anniversary of a battle;
• the 50th anniversary of the formation of a weather pattern that was later to develop into a cyclone (a date so irrelevant to the topic that it's not even mentioned in the blurb);
• the 71st anniversary of the indictment of a Nazi (not his conviction);
• A 220th birthday;
• A 75th anniversary;
• A 135th birthday (possibly—the date in the article and the year of birth in the blurb don't match, and there's no mention of the birthday in the blurb so even this tenuous connection isn't apparent to readers);
• A 90th birthday;
• A 139th birthday.

So of those eight, we have only one that's a legitimate use of "significant anniversary" (the 75th anniversary of a battle). TFA was never intended to be a bolt-on to On This Day, and the overwhelming majority of these "date significance" requests fail the basic "would any reasonable reader understand why the article is being run on this day?" test. IIRC, this kind of abuse was why we got rid of the points system for TFA selection in the first place. ‑ Iridescent 07:14, 24 January 2020 (UTC)

Iridescent Point taken, but what about genuinely significant anniversaries like 50th ann of moon landing? Jimfbleak - talk to me? 07:48, 24 January 2020 (UTC)

I suggested somewhere that the Coords should be able to override the random selector for truly significant dates (eg, when we ran Mccain, Obama). SandyGeorgia (Talk) 08:08, 24 January 2020 (UTC)

(e.c. but this took ages to type so posting anyway) That's already covered by Sandy up above as well—the TFA delegates would still have an override power for genuinely significant occasions (all it would take would be to temporarily overwrite the randomization script with a single blurb for the day), but it would only be for genuine landmarks with no more "187th anniversary of his election as the first superior of Jesuit Maryland province" (to take a genuine current example from TFAR) pseudo-significance.

There are obvious drawbacks to the idea. I don't believe "performance hit" for a second. (The main page gets a lot of hits but the number is still trivial in terms of the WMF's traffic as a whole, and even deleting the cache altogether—which nobody is suggesting—wouldn't cause a huge impact. Only refreshing the TFA when the cache is being purged anyway shouldn't cause any performance hit at all.) That said, there would be a genuine drawback in the need to keep 5000+ blurbs protected (presumably we'd redefine them as templates and restrict them to editors with the templateeditor right). The WMF would probably also scream blue murder as they'd no longer be able to send out their "Our featured article today is…" press release, and until readers got used to it it would be potentially confusing, as it would mean conversations like "hey, check out this interesting article on Robert Sterling Yard on the front page of Misplaced Pages!" "What do you mean, I just looked and the TFA is Catherine de' Medici's building projects". (Paging Mvolz, who pointed out that particular issue to me.)

I still think the benefit of spreading the view-spike that currently affects each TFA would outweigh the downsides; at the moment we have the unhealthy situation where people actively avoid taking articles to FAC because they don't want to risk them ending up on the main page, and where the authors of FAs regularly end up getting themselves in trouble trying to hold the line against 30,000 crazies and vandals when a page is at TFA. It would also go some way to solving the problem of FA-rot where articles gradually deteriorate after passing at FAC, as even the most obscure topic would get a steady couple of hundred views every day making it much more likely that issues would be pointed out. ‑ Iridescent 08:19, 24 January 2020 (UTC)

• This is a fantastic idea. I am calling this brilliant, and invoking genius :) See what I did there ;) ——SN54129 09:03, 24 January 2020 (UTC)
• In case I didn't support already, +1 from me. As for protecting blurbs, one could use a bot or a centralized cascade-protected list or both. Or if we are OK with template editors editing blurbs, the Titleblacklist and a common title pattern. Jo-Jo Eumerus (talk) 09:18, 24 January 2020 (UTC)

What about image protection? If we have several thousand blurbs, most with associated image, do we have to protect them to prevent someone substituting something in that they think is funny?--Wehwalt (talk) 09:45, 24 January 2020 (UTC)

Yes. I am not sure if protection on Commons would scale in such an instance; perhaps having upload+move protected local copies of Commons images would be necessary. Jo-Jo Eumerus (talk) 09:54, 24 January 2020 (UTC)

Just a few thoughts on the potential performance issues. When Iridescent suggested this to me at last week's Oxford meetup, my initial reaction was that any performance hit would be small, but the more I look at it, the more I'm convinced that the best way to achieve what Iri wants to achieve would be slice up the day into time-slots, similarly to DYK, but having a much smaller time interval, a pick a random FA as the FA for that slot for everybody. I suspect from what I've gleaned about caching on the main page, any slice greater than a few minutes would reduce concerns about the caching, clock skew, etc. Considering we only have around 5,000 candidates for the pick, the randomisation at that rate should be simple to achieve, as Lua would do the job in milliseconds or less. If you had a protected file containing the candidates for TFA, then adding newly promoted and removing demoted potential TFAs would be simple job - and having a "special" day with just one FA run all day would require nothing more than temporarily removing all the other names and leaving just one (or two, or whatever), as long as someone remembers to revert the edit for the next day. With around 5,000 potential TFAs and a time slice of 10 minutes, an article would appear on average every 35 days for ten minutes; with a time slot of 30 minutes, every 3 months, and scale to taste. Does that sound like what folks would want? --RexxS (talk) 15:07, 24 January 2020 (UTC)

Viewed that way, I'm thinking something like a one-hour time slot would do the job nicely. We should get more feedback from the technical people, and then run an RFC to decide the best time allotment. Also, thanking Jo-Jo Eumerus for the musings which kicked this off. I am going to re-flag this discussion over at WT:FAC to make sure people are paying attention. SandyGeorgia (Talk) 15:13, 24 January 2020 (UTC)

Just saw the notification about this at WT:FAC. No strong opinion about this, but I've certainly no objections to it, and I speak as a fairly prolific FA creator who would probably get a one hour slot a couple of times a week. That would be fine with me. I've never sought TFA for the articles I work on, but I don't mind it either. Mike Christie (talk - contribs - library) 15:29, 24 January 2020 (UTC)

Mike Christie the length of the time slot would need to be decided via RFC, depending on feedback from technical people. Do you know anyone who is good at phrasing RFCs? <hint, hint> Not yet, but when the time is ripe? For example, I'd rather see a bit longer time slot used (thinking an hour). SandyGeorgia (Talk) 15:34, 24 January 2020 (UTC)

Sorry! I appreciate the compliment but I am swamped at work and my spare time is being eaten by up by a couple of different multi-month projects, so I'm trying to edit WP as little as possible at the moment. Mike Christie (talk - contribs - library) 18:57, 24 January 2020 (UTC)

My sense is that there's always been at least some level of support for three things inherent in this proposal: more variety in what shows up at TFA, more chances to get on the Main Page than just the typical one-and-done, and better handling of blurbs. OTOH, it doesn't take a lot of effort to imagine various ways this could all go wrong. It's not my call, and if there's an RfC, I'll support whatever the result is to the best of my ability. But here's the thing: I never signed up to be on call 24/7 to deal with potential problems with thousands of blurbs ... my job so far has been dealing with one blurb at a time, and when we're done with the blurb, we're done ... and sometimes even that feels like too much. So I need to think about setting appropriate limits, starting now. Doing blurbs back through 2016 has been going smoothly, so I'm guessing I can stop doing blurb reviews as soon as FACs are promoted ... it will probably work fine to go back and cover them after 6 months or a year. I'd like to see where this proposed RfC is going before I wind up being overextended. (Except: I reserve the right to keep working with taxonomy articles and blurbs, and Gog the Mild has been working on MilHist blurbs ... much appreciated ... and I'll continue to support whatever direction that initiative takes.) That's my position, but thoughts are welcome. - Dank (push to talk) 15:13, 25 January 2020 (UTC) Well ... I'll support any reasonable result. - Dank (push to talk) 18:34, 27 January 2020 (UTC)

We might want to "pre-fill" the list of blurbs with existing TFA blurbs. And let editors carry out/request fixes and addition (and removal) of new blurbs as appropriate. I don't think it needs to be anyone's obligation to write a new blurb for each FAC that is successful. Jo-Jo Eumerus (talk) 16:10, 25 January 2020 (UTC)

I can't say that I support this concept. When the time comes for an RfC, I will strongly oppose it.

I've written and nominated dozens of FAs, over half of which have yet to appear as TFA. I find Main Page day to be a stressful endeavor, but one I'll take on when there is an anniversary connection. But the idea that now the increased attention and stress from a Main Page appearance could happen at any time, without warning is just too much. I see this proposal as full of downsides that will push away FA writers, not attract them.

The "Tidy Up Tuesday" concept played much better. Run older FAs regularly, give them some attention, and if they don't meet expectations, send them off to FAR. Even some advance notice that something has a TUT appearance coming could be enough to spur some improvements before TFA day, or have the article sent to FAR and withdrawn from a TFA slot. Imzadi 1979 → 01:54, 27 January 2020 (UTC)

Imzadi1979, I think you're completely misunderstanding the proposal. We're not considering having a randomly-selected article be TFA each day; we're considering randomising what's displayed, so any given reader at any given time has a 1⁄5711 chance of seeing any given article in the slot. Thus, instead of every FA having a single spike of 50,000 additional pageviews and then dropping back into obscurity, every FA would have a consistent 10 or so extra pageviews every day. (It would probably be slightly higher in practice, since at least some readers would click-through to multiple FAs.) The chance that increased attention and stress from a Main Page appearance could happen at any time happens all the damn time under the current system—as I know from bitter experience—but under this proposal it literally couldn't happen. When there's a genuine date relevance—the vast majority of purported "significant dates" at TFA at the moment are nothing of the kind—then the randomisation script would be shut off for the day and TFAR would operate as normal, so things like the moon landings wouldn't be affected. There are legitimate concerns that can be raised about it, such as the need to pre-emptively protect blurbs, but neither "it will make things more difficult for the writers of FAs" nor "it will mean we can't run TFAs to coincide with significant dates" is one of them.

I will say that regardless of whether this particular proposal—or anything like it—is accepted, your counter-proposal, of weaponising the TFA queue to get articles you think don't meet your standards delisted, absolutely stinks and I'll oppose it with everything I can throw at it. You're proposing deliberately subjecting editors to a double stress, first by scheduling articles they've written as TFA against their wishes and consequently forcing them to engage in cleanup, blurb writing, source re-checking etc, and then subjecting them to the additional stress of FAR which is a soul-destroying timesink even when everything runs smoothly, and it hasn't run smoothly for a decade. ‑ Iridescent 07:44, 27 January 2020 (UTC)

I thought the entire purpose of this exercise was to identify old FAs that are no longer up to scratch, and use time on the main page as an opportunity to either improve them, or decide they're unsalvagable and send them to FAR. If writers of old now-substandard FAs are allowed to veto their article appearing in the slot, what are we gaining by it? Personally I could possibly support a version of this, reposting old FAs is a good idea, but I don't think completely real-time is the answer. It would be like what happens sometimes on a Facebook feed, where you're in the middle of reading something only to find it's replaced when you refresh the page. Perhaps scheduling them once an hour could work. That way you'd cycle through all 5,711 FAs every 238 days,but you still know in advance when yours will be shown, and the protection can be arranged. Or do this on some days but run new TFAs as usual on other days so people still get a full day in the sun as reward for a new FA...  — Amakuru (talk) 09:42, 27 January 2020 (UTC)

That's part of the purpose, but there are others such as reducing the amount of hassle that FA maintainers receive from TFA day, increasing the visibility of older TFA'd FAs and reducing the impact of pseudo-anniversaries. Jo-Jo Eumerus (talk) 12:14, 27 January 2020 (UTC)

What is the frequency of new FAs? I tend to agree that running a new FA as a TFA for a set time (12 or 24 hours) is a good incentive for creating more FAs. But when there is no new FA to run, and perhaps no VERY SPECIAL occasion FA either, then going with a random article to every reader each time they visit the main page is a great idea. MB 14:39, 30 January 2020 (UTC)

WP:FAS: less than the needed 365 FAs per year. SandyGeorgia (Talk) 14:54, 30 January 2020 (UTC)

Looks like around 300/year for the last few years. So If a new FA ran for 12 hours, then roughly 50% of the time a user would see a new FA and 50% they would see a random FA from the pool of old FAs. MB 15:17, 30 January 2020 (UTC)

Ok FWIW I support the idea of a random FA being spat out on the main page either 100% of the time of 50% of the time. I'll leave it to others to work out the details. Cas Liber (talk ·' contribs) 19:32, 3 February 2020 (UTC)

Date issues

Good idea. That means we will no longer be subjected to inanities such as The City of New York vs. Homer Simpson which has a photo of the World Trade Center and was the TFA on the 18th anniversary of its destruction, and (one hopes) we will not see Super Columbine Massacre RPG!, a videogame, run, as is now proposed, on the 21st anniversary of that tragedy. Kablammo (talk) 17:15, 24 January 2020 (UTC)

Without commenting at all on the appropriateness of it, I assume the TFA mentioned above was scheduled intentionally as a date specific entry, and with the consensus of those involved in the process. If you wanted to object to it you could have contributed to the scheduling discussions up front. This doesn't seem in itself like a strong reason to move wholesale away from all date-specific TFA scheduling.  — Amakuru (talk) 10:48, 27 January 2020 (UTC)

Summary of questions

In the interest of moving this Misplaced Pages:Random featured article proposal along, I've compiled some questions:

1. Is there a consensus for implementing this change?
2. How random? Random as in random FAs for each user (as could be accomplished by using randomizer scripts) or random as in the same FA for each user (essentially a random selection of TFAs, say by bot)? That will condition the technical implementation.
3. How high is the frequency? Daily (essentially randomly selecting TFAs), hourly (as could be done by bot) or by minute (cribbing Aaron Schulz's comment above that we might want to keep some caching of the Main Page) or something in between these?
4. What are the technical implications? We'll need new templates and scripts or bots. If it's scripts, how do they interact with caching and the other things Aaron Schulz mentioned above? If it's a bot, who maintains it and what is done if it breaks?
5. Where are the blurbs to be used stored (subpages of Misplaced Pages:Random featured article or Misplaced Pages:Today's featured article, one giant page or something else?)? How are they created and maintained?
6. Can we override the random selection for specific dates? If so, what are the criteria? I figure we want to make exceptions only for truly important dates.
7. Which processes need to be updated? TFA obviously, but also FARC as the blurb needs to be removed from rotation if the corresponding FA is delisted. Are there more?
8. What happens to existing processes, such as the TFA coordinators?

These might become part of a RfC or something. Jo-Jo Eumerus (talk) 16:08, 25 January 2020 (UTC)

I have some ideas, but ... Jo-Jo Eumerus would you mind converting your points to numbered bullet points so we can add comments below, referencing each numbered point above? SandyGeorgia (Talk) 16:13, 25 January 2020 (UTC)

Done and added a bullet point. Jo-Jo Eumerus (talk) 16:15, 25 January 2020 (UTC)

On points 1, 2 and 3, I suggest we run an RFC to answer those questions, if this proposal gains (technical and other) traction. On 5, I suggest a change we should implement independently, and soon. Blurbs should be stored/linked in the {{articlehistory}} template. Then, the randomizer would not pick up blurbs whose articlehistory does not indicate they are FAs. That is, FACBot would not need to do anything new (other than delist FAs). Storing blurbs at the talk page of the FAC makes no sense, since there can be multiple FACs and FARs; they should be an AH entry anyway. On point 6, I suggest we include that in the RFC, but that we focus on examples. For example, it seems that when we ran McCain-Obama is a no-brainer (both candidates were FAs in a national election, which is pretty cool for FACcers). How do we feel about something like Early life of Samuel Johnson, which ran on the 300th anniversary of his birthday, which was a big fat deal in the UK? We can construct the RFC in a way that we put forward enough examples to figure out what we actually mean by truly relevant date connections. On point 7, if we store blurbs in articlehistory and the randomizer only picks up FAs that are current per articlehistory, no changes would be needed at FARC. On point 8, FAC and FAR= no changes. Or, we could make blurb approval part of FAC if the RFC determined we should do that. TFA gets re-worked completely, to account for when we override, and how we administer blurbs; there will still be a process, and plenty of need for the Coords-- they will just be doing different things than they do now. SandyGeorgia (Talk) 16:25, 25 January 2020 (UTC)

@SandyGeorgia, storing the blurbs within the {{articlehistory}} template wouldn't be a trivial change. At the moment, the history is a part of the talkpage, and except in a few exceptional cases the article talk page has to be fully unprotected so unregistered readers who spot potential problems can discuss them, and TFA blurbs are one area where we explicitly don't want "anyone can edit" to apply. It would be feasible to move {{articlehistory}} to protected, transcluded subpages of each main talk page, but {{articlehistory}} is used on tens of thousands of articles, so we're talking about a significant change that will affect thousands of editors (and flood every watchlist on the wiki if/when it happens).

On "which anniversaries are significant enough to warrant an exception", my opinion in a putative New TFA Order is the same as it is in the current: if I saw a news photo headlined "World Leaders Gather to Celebrate the Footieth Anniversary of Bar", would my reaction be "why the hell is that happening?". So, centenary of the founding of even the most tinpot country, fine, since one presumes Trump, Putin etc all get invites to the ceremony even if they decline to attend; ditto the moon landings, genuinely major military events, genuinely transformative world figures etc. (FWIW as someone who was in the UK in 2009 I strongly dispute that Samuel Johnson's birthday is or ever has been a big fat deal in the UK. I doubt one person in twenty has ever heard of him, and of those that have it's generally going to be only as a character from Blackadder and they won't be aware he was actually a real person, let alone who he was. His 300th birthday didn't even get a commemorative set of stamps, and the Royal Mail will issue a stamp for pretty much anything they think is of any current public interest—we had "Elton John's current tour", "Final series of Game of Thrones" and "20th anniversary of The Gruffalo" in the last few months alone, not to mention the current "Darth Sidious’ apprentice Darth Maul" first class stamp.) ‑ Iridescent 13:59, 26 January 2020 (UTC)

@Iridescent: as always, I may have no idea what I'm talking about, but ... no, no, no! We don't store the actual blurb text in articlehistory; we create standardized templates that contain the blurb, and add a parameter to articlehistory making it easier for anyone to find them (exactly as we do, for example, for DYKs). On the significant anniversaries, yes, opinions will differ, so we make it part of the RFC with specific examples for clear guidance. (I think it was Ottava Riva who claimed that Johnson's 300th was a big deal; how would I know :) SandyGeorgia (Talk) 14:04, 26 January 2020 (UTC)

So for example one could store the blurb for Tarrare under Misplaced Pages:Random featured article/Tarrare? I don't think that there'd be a problem at linkint it from {{articlehistory}} but for the purposes of this discussion it's more important to know where the blurb will be stored. Jo-Jo Eumerus (talk) 17:25, 26 January 2020 (UTC)

As the blurbs are meant to be transcluded, my inclination would be to use subpages in Template space, so the blurb for Tarrare might be at Template:TRFA/Tarrare. It's not a big deal, of course, but seems a little more consistent to me. --RexxS (talk) 00:30, 27 January 2020 (UTC)

I am really hesitant to see prewritten blurbs and have been puzzled to see why this has become so much of a priority. For example, if Kobe Bryant was a FA today, we'd basically have to rewrite the blurb. But multiply that by how many FAs we have and consider that many of them are for lesser-known topics, we will be having a major sync problem on our hands. --Rschen7754 02:28, 27 January 2020 (UTC)

Indeed, and TBH I share Rschen's caution about the speed with which this is being rolled forward, for what would be arguably the biggest change to the main page since the cowboy days if the early 2000s... This entire proposal is predicated on a rather large amount of work being done up front to get the affected blurbs oven-ready. The current state of an article and the current important information on a subject may differ significantly from a blurb written several years ago, not to mention MOS changes etc, so each one will have to be looked at, sources checked etc. I think if we're to move forward it's worth having a clear idea of who is going to do this work (we can't expect Dank to do it all, as it's already a major job getting one blurb a day ready), and how many FAs will be included in the rotation on launch day.  — Amakuru (talk) 10:44, 27 January 2020 (UTC)

@Rschen7754 and Amakuru: No opinion on pre-written blurbs, but for what it's worth Dank has already written about 500 blurbs. Upon checking the first 10-20 in the search result it doesn't seem like any of them would need updating, and reading through the rest of the search results it doesn't seem like the number of blurbs in need of updating is large at all. That's not really surprising - most FAs are on topics whose key information does not require frequent updating.

Incidentally, don't think it's necessary to have a blurb ready for every FA right from the get-go. People who want to see a FA get on the mainpage could propose a blurb for a FA that doesn't have already one, or re-use a former TFA blurb if it doesn't need updating. I think that editors can take care of this over time. Jo-Jo Eumerus (talk) 11:14, 27 January 2020 (UTC)

Re: speed with which this is being rolled forward, what speed? Various ideas have been tossed about for about a month, and in clear evidence in this talk discussion is the intent to think this through before even approaching the idea of an RFC. Re: Dank has already written about 500 blurbs, those are only the blurbs on FAC talk pages. Every TFA that has ever run already has a blurb, linked from the {{articlehistory}} template, and that is the HUGE majority of all TFAs. (See WP:FA where those that have already run are bolded, or with a script you can see them in colors.) Most FAs already have blurbs. IF this ides moves forward, we could incrementally add only those that are ready; they don't have to all be ready at once. SandyGeorgia (Talk) 12:19, 27 January 2020 (UTC)

In case anyone is interested, I've written a sample list of past TFAs at User:Jo-Jo Eumerus/Sample of TFAs to check whether any of their blurbs can be reused. Note that "MIGHT" is not there for decoration; it marks blurbs which might need updating before reuse not these that must get updated. Jo-Jo Eumerus (talk) 15:30, 27 January 2020 (UTC)

This discussion is a bit all over the place right now, but I could swear I saw above that a randomization system could rely on new templates that only admins/template editors would have the ability to edit. Considering that a decent number of regular FA contributors don't have the tools, what happens when an update is inevitably needed in one of the 5,700 blurbs? Bothering a random admin doesn't seem like the best solution, as we can't even get errors reported on main page talk fixed all the time. You're going to have to create a new page to report proposed updates/corrections in the blurbs, which will add yet another layer of process while leaving the risk of blurbs becoming outdated. For me, this whole idea strikes me as unneeded process. If you want to run more old FAs, run more old FAs. You don't need to randomize the entire system to do that. Giants2008 (Talk) 02:54, 28 January 2020 (UTC)

Giants2008, I am not an admin, but I have template editor rights— they are unbundled. SandyGeorgia (Talk) 04:02, 28 January 2020 (UTC)

Giants2008 I don't think this would require a new process. We'd simply use the existing edit requests and WP:ERRORS processes for this; after all currently when someone sees a problem with an article featured on the mainpage they ask for a fix there.

Also, judging by User:Jo-Jo Eumerus/Sample of TFAs only about 7% of all blurbs actually might warrant updating ... and it's probably only half that which actually need an update.

Based on my sample User:Jo-Jo Eumerus/Sample of TFAs it seems like we might need about 38 - 76 blurb updates per year once we have started the new process. For the transition, about 504 blurbs would need analysis - and it'd be mostly the pre-2010 ones. Jo-Jo Eumerus (talk) 12:26, 28 January 2020 (UTC)

• Having random FAs displayed is something that Portals have done for more than ten years. Most portals used their own blurb pages, which very often went out of date. We'd need to make sure this doesn't happen with main page FA blurbs, especially for BLPs. Many portals no longer use specific blurbs for those, but just re-use parts of the article lead via Module:Excerpt. While not perfect, this avoids not noticing that somebody died. For the Main Page, the number of eyeballs will be a couple of orders of magnitude higher, so I hope we'd get error reports about outdated blurbs quickly, but we'll really need more admins (or relax editing rights for the Main Page) to make sure we fix problems right away. Portals could benefit from Main Page blurbs that are kept up-to-date and re-use them instead of using Excerpt tricks. Anyway, I believe the proposal is worth trying for a month or two, and perhaps going back to the old method if we notice serious problems. —Kusma (t·c) 13:22, 28 January 2020 (UTC)
  • There was a proposal being developed somewhere to unbundle the ability to edit through protection from the admin toolset, but I haven't seen anything on it in a while. If TFA is going to do this (and I'm still not sold on whether it should), I really think this would help. However, I'm unsure whether a consensus for unbundling exists among the larger community. Giants2008 (Talk) 22:46, 28 January 2020 (UTC)

    I would expect pushback from the people who object to non-admins doing things. Hawkeye7 (discuss) 23:26, 28 January 2020 (UTC)

• As far as I can tell, this idea removes the whole purpose of FAC. No errors will be reported or corrected, as the time is too short, but multiple pages will be vandalised, as there will be neither the time nor the mechanism to protect them. Page views will be less than DYK articles. Given the amount of work required to get an article through FAC, it will no longer be worth doing so. Hawkeye7 (discuss) 18:03, 28 January 2020 (UTC)

  Mmm? FAs will still be featured on the main page in a privileged position, just the turnaround time and selection method will change. Jo-Jo Eumerus (talk) 20:00, 28 January 2020 (UTC)

  An hour is worthless. Most people will not see them, and they will receive less page views than DYKs, which are there all day long. It's the end of FA. Hawkeye7 (discuss) 23:26, 28 January 2020 (UTC)

  Hawkeye, not to state the obvious but you are aware that FAC is an article assessment process, not a process to select material for the main page? The notion of "Today's Featured Article" was a later addition; before that we used to feature (blurbless) links to a random selection of multiple FAs on the main page. (This is what the main page looked like before TFA.) We're not proposing FAs no longer appearing on the main page, we're talking about all the FAs appearing on the main page, all the time. If your concern is the number of hits your articles get, this proposal would significantly increase the pageviews for every FA you've ever written—instead of articles having a single spike and then dropping into unread obscurity, every FA would have a consistent and permanent rise in its pageviews every day. ‑ Iridescent 08:04, 29 January 2020 (UTC)

  Not to state the obvious, but your math is faulty. A new article will take 15 years to accumulate 24 hours of screen time! You're proposing shutting up shop for good, replacing TFA with a showcase of old featured articles. Hawkeye7 (discuss) 10:08, 29 January 2020 (UTC)

  I don't think it will take that long. A given reader is unlikely to click through more than once if the main page features the same FA all day long; if they encounter a rotation chances are good they'll click through to more than one. Yes, the TFA "readership" will be split between many FAs every day instead of just one (and there are enough recent FAs that it wouldn't become a showcase of old featured articles) but the total readership of all FAs will likely increase. Jo-Jo Eumerus (talk) 13:03, 29 January 2020 (UTC)

• Three quarters of the featured articles date from 2015 or earlier. Only one in 20 dates from 2019. Hawkeye7 (discuss) 18:54, 29 January 2020 (UTC)

So, what are the next steps to get consensus for this change? One thing up front to clarify is #2 as there are two distinct approaches to randomization; it could become part of the main RfC question or of a follow up one. Jo-Jo Eumerus (talk) 21:19, 29 January 2020 (UTC)

• There needs to be a mechanism to stop certain FAs running on certain days and I would suggest it is probably better if the picker script rotates round the categories and years, to prevent it randomly choosing four Feb 29th 2012 hurricanes, four Anglo-saxon AEflebaerds, or four Upchuckina Bucketi fungi in a row (none of those probably exist, but you get the idea) Yomangani

@Jo-Jo Eumerus: it seems like we need more feedback from the technical people and the mainpage people before we are ready for an RFC. I'm not sure how to best go about soliciting that. SandyGeorgia (Talk) 01:01, 30 January 2020 (UTC)

@SandyGeorgia:I've ãsked on Talk:Main Page and on Misplaced Pages:Village pump (technical). Jo-Jo Eumerus (talk) 09:48, 30 January 2020 (UTC)

I'd like to have a clearer understanding of what the problems presented are that this discussion is seeking to cure. As I understand it, there is a desire to de-emphasize date relevance. If we go forward with this, I gather, coordinators will still have discretion to suspend the random rotation for a significant anniversary. We will need a page for the community to guide us in this discretion, probably it would look a lot like TFA/R, probably it would be TFA/R since there is no point in chucking a well known page when you really aren't changing its function. Most TFA/Rs sail through unopposed. If that continues, then we become gatekeepers against consensus, if there is opposition in nominations to uphold some new standard, you have friction. Either way, you have date relevance becoming the only way to get on the main page for a day in the sun, something that, judging by the continued flow of articles to TFA/R and the pending template, people still value. The importance of date relevance is obviously increased under such a scenario. There's also an issue that the problem of main-page-article vandalism would be spread out, including to articles that few people may be actually watching. I'd like to see some attention to these issues at some point in the discussion.--Wehwalt (talk) 10:21, 30 January 2020 (UTC)

There are several problems that this attempts to address:

1. A lot of supposedly "relevant" dates are - as illustrated by the examples Iridescent mentions above - not really particularly date relevant.
2. The one-TFA-per-day system creates a heavy unbalanced interest in FAs, with the TFA getting a lot of often bad edits and pressuring its maintainers into preparing it for TFA day, while non-TFA (in particular already-TFA) FAs languish in obscurity and potentially accumulate problems without anyone noticing.
3. There is only one TFA all day long. There is a good chance that having more than one per day will result in people going to read several of them instead of just that day's featured article.

Jo-Jo Eumerus (talk) 10:35, 30 January 2020 (UTC)

Another question

What if the randomizer were to somehow be adjusted to give extra weight to recent FAs? Or to somehow assure that all FAs promoted in the last 30 days got plenty of time? SandyGeorgia (Talk) 14:58, 30 January 2020 (UTC)

My preferred approach to that would be a hybrid one. Not-yet-featured FAs would still get their own dedicated time in the sun, either by alternating daily between conventional TFAs running new promotions and the new approach, or morning /afternoon or something. TFA is without doubt a massive motivator for FAC people and I think we should keep that incentive in place. Let's not forget this conversation started out as "tidy up Tuesday" and has morphed into "tidy up every day"  — Amakuru (talk) 23:43, 31 January 2020 (UTC)

TFA ready

I am starting to work through some of the FAs at User:Dweller/Featured Articles that haven't been on Main Page and am finding that some of them would be fine to run on the main page. The idea that FAs need to be perfect to run on the main page is not good for the process; running an FA with a few issues is a good way to get them corrected, and not all of these are so embarrassing that they couldn't run. Those that are, I will be leaving "FAR needed" notices on the article talk page, so that anyone can submit a FAR a week after notice is given, if no one engages to improve the issues. Anyone can give a notice and add it to this template; anyone can submit a FAR for any article in the template after seven days if no one engages.

We should be processing the truly bad FAs at FAR, and running the "not perfect, but not so bad they need a FAR" at TFA. For example, Battle of Tulagi and Gavutu–Tanambogo. SandyGeorgia (Talk) 17:13, 28 January 2020 (UTC)

I don't see anything much wrong with that article. It would be a shame to send it to FAR where its FA status will be stricken. Hawkeye7 (discuss) 18:08, 28 January 2020 (UTC)

Right ... I added a few cn tags, but it does not need a FAR, and would be perfectly fine for the main page (notice that I did not add it to the template of FAR notices). There are lots of problems with that list, and there are many there that could run. I could list many more (and still working), but Witchfinder General (1968 film) has some sourcing and citation issues, but would fine to run on the main page, where someone might pick it up and improve it. Brownsea Island Scout camp could run; it's missing a few citations and has a boatload of MOS issues, but that is just the sort of thing that could be fixed on TFA day. SandyGeorgia (Talk) 19:50, 28 January 2020 (UTC)

I think I'm largely in agreement with Sandy on this concept: articles don't have to be perfect to run. The extra attention will spot those imperfections and will hopefully bring some new editors into the mix. If the article is worse than suspected, well, maybe that attention will shine a light on things and we'll head down the FAR path with it at some point. However, I appreciate some notice that an article will appear; that gives me a chance to do some polishing before the spotlight comes around. Imzadi 1979 → 20:20, 28 January 2020 (UTC)

Look that-a-way ---------- >>>>> and start submitting to FAR after a week, if no one engages! This is barely scratching the surface. SandyGeorgia (Talk) 20:23, 28 January 2020 (UTC)

Another one that would be fine at TFA, where issues might be addressed: 1994 Black Hawk shootdown incident. SandyGeorgia (Talk) 20:29, 28 January 2020 (UTC)

5000+ blurbs

This was an objection above about just showing a random FA to each individual viewer. My question is, "why are we writing blurbs for them at all?" The idea is to get people to read the actual article, right? Not read just our blurb. Now, if I am logged out, placing my cursor on any link brings up a preview of that page, so why not just display that preview in the Featured Article slot? --Khajidha (talk) 16:24, 31 January 2020 (UTC)

That is essentially the approach of using Module:Excerpt, used by many portals. One downside is that vandalism to the article could percolate to the Main Page. Another disadvantage that the first paragraph of the article is usually not as comprehensive as a bespoke blurb. On the positive side, this approach makes the blurb always stay as up-to-date as the article; for example when a person dies the blurb would pick that up in the exact same moment as the article. —Kusma (t·c) 17:31, 31 January 2020 (UTC)

I suspect that the cursor trick also does not work well in mobile view... although I cannot swear to that fact. Ealdgyth - Talk 17:32, 31 January 2020 (UTC)

Kusma, for User:Johnbod and myself, and per Misplaced Pages:Articles for deletion/Bespoke, wikilinking bespoke. SandyGeorgia (Talk) 17:58, 31 January 2020 (UTC)

1) Ealdgyth, I was speaking of the outcome, not the mechanism. Module:Excerpt is just what I was thinking of. 2) Greater ease of vandalism is balanced by easier correction of vandalism. 3) Why do we WANT the blurb to be comprehensive? That's what the article is for. That's kind of my entire point. --Khajidha (talk) 18:17, 31 January 2020 (UTC)

Khajidha, you're absolutely right about "comprehensive", that was not a good argument from me. However, there is no guarantee that the first paragraph of any article is a good teaser for the article either, and especially we have no control on how long it will be, or whether the chosen image is suitable (and even if it is suitable today, that may change without warning tomorrow, and you won't find out until you see the excerpt on the Main Page, unless you regularly check pages containing all possible excerpts). —Kusma (t·c) 18:47, 31 January 2020 (UTC)

I'm pretty sure using Module:Excerpt would cause weirdness with the Main Page cascading protections, with articles rapidly gaining and losing cascading protection as the selected article changes (or otherwise fully protecting all FAs). Danski454 (talk) 16:35, 1 February 2020 (UTC)

• One problem is that the shortened blurbs would have to be kept up to date with developments in the article by hand. This is hard work, and it is not clear to me who would be responsible for doing it. Espresso Addict (talk) 03:25, 2 February 2020 (UTC)

There's also the main-page balance issue -- unless they are all identical in length *including any image*. Espresso Addict (talk) 03:38, 2 February 2020 (UTC)

I think it's implicit in Misplaced Pages's philosophy that people who spot a problem (e.g on the main page or because they are editing an article) request a fix (i.e an edit of the blurb). That could be done through the usual processes. Jo-Jo Eumerus (talk) 09:07, 2 February 2020 (UTC)

This is starting to seem like an awful lot of work and ongoing maintenance.--Wehwalt (talk) 09:18, 2 February 2020 (UTC)

If people agree that the main-space blurbs must be fully protected from editing permanently, like the rest of the main-page content, this would seem to place a severe strain on the admin pool. If not, then we need to accept the possibility of some people visiting the main page getting served overtly vandalised information, which would seem at very least to rule out including BLPs. Espresso Addict (talk) 00:07, 3 February 2020 (UTC)

@Espresso Addict:I think you are overestimating the work needed to maintain 7300 blurbs a little. Going by the analysis on User:Jo-Jo Eumerus/Sample of TFAs only 76 or so blurbs per year would need updating (requests would presumably come from editors seeing the blurb on the main page and the article maintainers) which is well within the capability of the admin pool. Jo-Jo Eumerus (talk) 09:24, 3 February 2020 (UTC)

I find "main page balance" issues to be a poor objection, as it is our own fault for deciding to have two columns in the first place. That decision seems to me to cause more problems than it solves. --Khajidha (talk) 12:35, 3 February 2020 (UTC)

• I have an overwhelming feeling of 'meh' over the proposals (as I do to most of the MP stuff, tbh), but the "blurb" question has one possible solution. It tends to be seen on various news feeds where a subscription is needed to access the full conent, but they give you the first paragraph or 100 words or so before the text fades into a "Subscribe" button. We don't need the arty fading out, but the first para/100 words followed by an ellipsis and a "Full article" link would suffice. If it's a set number of words then the problems on length would be omitted (depending on whether there is an image in place or not). The bigger problem to me is not the blurb, but the images - David Levy does stalwart (and thankless) work on the MP images and may have good input to include on any parts of this proposal.

I'm not in favour of cursor hovering - that's a pain in the arse when reading on an iPad or mobile (and may cause access problems for some - I think, although I'm not 100% sure). - SchroCat (talk) 09:50, 2 February 2020 (UTC)

SchroCat has it exactly right. There is no "placing my cursor on any link" for mobile users, and they now represent the majority of our page views. We can't simply remove their access to the blurb. --RexxS (talk) 01:35, 3 February 2020 (UTC)

Again, I NEVER said to use the "placing a cursor" thing, I said to just put whatever that would display up instead of going to the trouble of making a blurb. To put it more simply, we are trying to show off the article so why not JUST SHOW WHAT THE ARTICLE SAYS. --Khajidha (talk) 12:35, 3 February 2020 (UTC)

Randomisation

Speaking personally, I would like to steer away from the idea of using the randomiser script and caching to achieve this. I think that's fraught with problems for admins and article maintainers, including mot knowing who's seeing your article on the main page at any time, not knowing where to look for potential vandalism, how to protect the images, unintuitive UI which looks different every time you refresh the page etc. I would likely oppose such a proposal. Instead, if we go down the randomising route, it should be using an approach similar to DYK, where the TFA template is updated on a schedule by a bot. That could be hourly, half-hourly, whatever we we prefer. But crucially, although its order would be mostly random, and the bot could also handle that scheduling, the order would be well-defined and known at least a day or two in advance. This gives time for images to be protected, and admins and others to watch articles for vandalism. This would also allow tweaks if there were too many similar things scheduled in a row, last minute corrections to blurbs etc. Thoughts?  — Amakuru (talk) 00:03, 1 February 2020 (UTC)

I'd surely oppose aggressive randomization too (one that changes after every refresh). I am already fed up with sites like Quora that do that. Once it refreshes you'll likely not found something interesting your eyes saw on the page unless you are lucky to quickly take the complete and accurate mental note of the headline/title. In most cases you'll never find it. To me, that's one of the worst features any site can adopt. – Ammarpad (talk) 04:27, 1 February 2020 (UTC)

I've tried out some portals bearing Template:Random subpage - both while logged in and while logged out in Incognito mode - and they only refresh when you force-purge the cache. Assuming that using that template or similar won't cause performance issues, it doesn't seem that a proposal on that basis would be unduly aggressive. I am not sure how the randomization script that was used for Misplaced Pages:Today's featured article/November 4, 2008 would perform, however; Iridescent? Jo-Jo Eumerus (talk) 09:06, 1 February 2020 (UTC)

Yes, we also use the randomiser at Picture-of-the-day, for cases where there are multiple very similar related images, and it would be overkill to feature each one separately on its own day. Template:POTD protected/2020-01-24 is a recent case of that, and the randomiser is a great tool for that purpose. The set is small, we protect all of them upfront anyway, and it doesn't really matter which one is shown. As I say though, I will strongly oppose a proposal to use this approach for TFA. Cache purging is a poor way to pick between different articles, for the reasons I've outlined (not to mention that actually, the interval between purges is essentially unknown because anyone anywhere can purge at any time). Given that many people above have already agreed that something like a one-hour window is an appropriate one, there's no good reason IMHO not to make it psuedo-random in a deterministic way. It's fine for the end readers to get a "random" article, but for those of us monitoring things behind the scenes it doesn't make sense for us to have no idea what's coming next. Cheers  — Amakuru (talk) 13:32, 1 February 2020 (UTC)

I think that having a queue with an one hour delay (or even several hours delay, or any delay really) might create a pressure to "prepare" the article without the time for doing that. So that is not an advantage and might actually become a problem. I wasn't saying that cache-purging is the way to pick between different articles, only explaining why I don't think that overly-aggressive refreshing is going to be a problem. Beyond that I am relatively agnostic on whether to use a quick-turnover bot or something like {{Random subpage}}. Jo-Jo Eumerus (talk) 14:06, 1 February 2020 (UTC)

For jobs like that we use a FIFO buffer, which would be simple to implement. If you wanted a couple of days' notice, a page containing the list of the next 48 TFA names, with a bot popping the top one off to the main page and pushing a random one on to the bottom every hour, would do the job. You can have as much advance notice as you want by scaling a simple system like that. --RexxS (talk) 15:10, 1 February 2020 (UTC)

Incidentally, if you have a list of 72 articles (three days worth at hourly intervals) picked at random from a list of 5,000, the chance of having at least one duplicate is about 40%. That may seem a little unintuitive, but hopefully it won't cause problems. --RexxS (talk) 15:29, 1 February 2020 (UTC)

Just an idea for those worried about not knowing which set of articles might appear on any given day: WP:FA has 30 categories. Might we consider some sort of scheme to run a given group on a given day? That would keep all the mushrooms, hurricanes, and battles on one day, as an example. SandyGeorgia (Talk) 15:34, 1 February 2020 (UTC)

I am replying generally to the tenor of this thread. We had a longstanding random rotation of images on the India page. It lasted some eight or nine years. It was instituted because there were calls for diversity by region, ethnicity, religion, and what have you. Each image location had eight images appearing randomly, and differently, for each user. But, in the end, the system was abandoned when the article was getting ready to appear at TFA for a second time, last October! There had been calls to abandon it earlier. I concluded that people had had their fill of diversity, and were now preferring consistency. They didn't seem to be liking the changing images anymore. They especially didn't like that what they were seeing was different from what others (they knew) were seeing. So be warned. Fowler&fowler«Talk» 01:32, 2 February 2020 (UTC)

That's quite understandable. It seems to me, though, that there seems to be a convergence of opinion toward a system that rotates through a randomly selected featured article on around an hourly schedule – perhaps "This hour's Featured Article" (THFA)! That should almost completely remove the effect that concerns you. --RexxS (talk) 13:09, 2 February 2020 (UTC)

I think this entire idea is dead on arrival. This removes most of the incentive for editor to go through the arduous FAC process. Editors are highly motivated by 24-hour highly visible featuring on the Main Page. Take that out and FA production is significantly disadvantaged. --- C&C (Coffeeandcrumbs) 02:40, 3 February 2020 (UTC)

Easy there on "DOA". :) I think there is a set of editors who want the glory of 24 hours and another set who would prefer the lower-level change set of this rotation that is being entertained. I can think of no reason why we could not pursue both. Users could opt-in to the 24 hour display if they wished and otherwise it would default to a rotation. (I have general unease with the direction of the overall discussion but I haven't sorted why. I think that it is because the formula we have today is one we've had for the past 15ish years.) --Izno (talk) 02:46, 3 February 2020 (UTC)

There's no glory in FAC. You won't get to be Editor of the Week writing featured articles. And What is disturbing me is the continual chipping away at the value of the featured article. We lost the concept of them being the best of the best when we capped nominations, now we are talking about downgrading the appearance on the front page. We are drifting further and further from the idea of peer reviewed content. Hawkeye7 (discuss) 03:56, 3 February 2020 (UTC)

I don't understand this. The nomination cap isn't a problem as you can just get somebody else to nominate - unless the problem with nomination cap is that the nominator gets the "glory" (If that is the problem, I have an old guide to help with that). The main page appearance seems to have little to do with peer-review unless you count the (what has been, in my experience) "once in a blue moon" occurrence when somebody with some knowledge is interested enough to both read and edit or review the article. Yomangani 11:27, 3 February 2020 (UTC)

No, the coordinators are wise to this trick. I had it happen a few times. Either I agree to be co-nom or the article gets archived. Hawkeye7 (discuss) 19:49, 3 February 2020 (UTC)

I feel that there are a lot of consequences of "random" and "random most of the time" that need to be elaborated. ¶ One important consequence mentioned already by two people: true randomness leads to poor outcomes. So someone (or a bot) has to manage the randomness by some method. The design of the bot's algorithm and/or any human intervention in a constantly rotating queue are a lot of effort, at least up front, once the possibly extensive debate about same has completed. (And SandyGeorgia suggested an idea that is the opposite of what I would have suggested for the "managed random" queue: she would repeat a certain topic in successive queue spots, I guess creating a theme? And I imagined the opposite, constantly picking articles from different FA topics, never repeating the same topic until n queue-spots later. I mention this to say that there are going to be a bunch of ideas for the semi-automated queue management stuff, and that stuff is necessary to prevent the problems with randomness.) ¶ I think we should call it a day on "random for every user all the time". Makes for extremely confusing user interface, unable to point out what is on the page to another person, and apparently significant technical-scale considerations. ¶ There is a general sense that exceptions to "randomness" will continue to be made. To my mind, any article featured for a 24-hour period in a context where FAs are otherwise being rotated constantly creates a potentially serious point-of-view or "endorsement" issue. Maybe I'm being too abstract, but any kind of change in information presentation needs to consider how people who know nothing about Misplaced Pages would perceive it. Main page viewer: "Normally Misplaced Pages is showing me different 'featured articles' very often. Now it looks like it's promoting topic X by 'pinning' that article for a long time. What does that mean?". (I think it would be fine if articles at the level of importance of the Rwandan Civil War (an example given recently on WT:FAC) were 'pinned', and if the obvious anniversary date was noted explicitly; but when I think of the average (very specific) FA topic, I can't see it having any excuse to ever be 'pinned'. But we're gonna have different ideas of what the exceptions are.) ¶ The reason I would be a possible supporter of this suggestion despite its difficulties is simply the idea that the whole spread of featured articles gets shown to people much more—possibly, just possibly, increasing interest in the process or project. Cheers, Outriggr (talk) 08:20, 3 February 2020 (UTC)

I thought the TFA co-ords spread the subject matter around as much as is possible already? Randomisation won't make the "whole spread" look any wider - it will have the reverse effect as the topics which are heavy with FAs (Milhist, mushrooms, roads, etc) will have more time overall on the MP than the smaller pools of Chemistry, Computing, Heraldry, etc. That can only be reversed by more articles in those areas. One of the wrinkles I have against bot-randomised article choice is that we'll end up with more of the bigger topics than we do already. - SchroCat (talk) 09:06, 3 February 2020 (UTC)

I thought so as well, given that Misplaced Pages:Featured articles that haven't been on the Main Page is considerably more unbalanced than Misplaced Pages:Featured articles. OTOH this probably indicates that "spreading the subject matter" isn't sustainable... Jo-Jo Eumerus (talk) 09:24, 3 February 2020 (UTC)

It seems that full randomisation is impractical and it appears that pseudo-randomisation (with scheduling, exceptions, protections, notifications, explanations, etc., etc.) will be as much, or more, work as the status quo, so the question should be what benefits the reader more? (If possible, forget about servicing editors' egos. Hard, I know.) Somewhere there are probably some statistics on how many people regularly visit the main page on several different occasions over the day, but I guess it will not be many: most people that religiously visit the main page once a day will probably be going directly to an article on the other occasions they visit (and google, with its own "blurbs", seems quite keen on not having anybody visit any pages at all). In this case, rotating the articles several times a day will be serving a tiny audience (even assuming they prefer a different article every time they visit). If we rotate the blurbs several times a day the "Recently featured" articles will also quickly shuffle off the older items, decreasing their "main page exposure" from four days to a couple of days at most, so if you skipped visiting over the weekend, all mention of Saturday's article would have disappeared from the main page by your visit on Monday. Journalists and bloggers who sometimes highlight today's featured article would also probably give up, as by the time they'd written their piece the article could have disappeared. Yomangani 11:27, 3 February 2020 (UTC)

This all looks like a solution in search of a problem and I am creeping slowly towards a (very) weak oppose on the idea. There seem to be more problems with the change than it's worth. I can't see the need for changing to a random system, and too little benefit to editors or readers to justify it. What we have may not be perfect, but it does have some advantages. What positive benefits does randomisation have? I am not seeing anything that outweighs the positives and negatives of the current system. - SchroCat (talk) 11:39, 3 February 2020 (UTC)

I wonder how many people make use of the "featured article feed", which would be broken by most versions of this randomization idea. Anomie⚔ 13:44, 3 February 2020 (UTC)

@Anomie: Offtopic: Where does that feed come from? Just want to check that the work at Talk:Main Page#Main_Page_January_2020_technical_update (with possible renames of element IDs/classes) won't interrupt that. --Izno (talk) 15:16, 3 February 2020 (UTC)

"Journalists and bloggers who sometimes highlight today's featured article would also probably give up, as by the time they'd written their piece the article could have disappeared" So? That's THEIR problem, not ours. --Khajidha (talk) 14:44, 3 February 2020 (UTC)

I thought we were supposed to be considering these changes for the benefit of readers? Journalists and bloggers also count as readers, and they are doubly welcome much of the time as they essentially advertise us to their audiences. - SchroCat (talk) 14:48, 3 February 2020 (UTC)

Journalists and bloggers can still advertise featured articles to their readers whether they're on the Main Page or not. They just have to direct the readers to the actual article. Which is supposed to be the actual point. The ARTICLE, not the Main Page link to it or the TFA credit for whomever. --Khajidha (talk) 14:59, 3 February 2020 (UTC)

That's an argument for scrapping TFA, not for or against the randomisation idea. Yomangani 15:08, 3 February 2020 (UTC)

I'm begining to harden my opposition to this idea. If they are trying to deal with constantly shifting sands it loses focus. And, for what it's worth, I don't give a toss about "TFA credit" (whatever that is - I've never seen anything resembling "credit", just grief). You can't throw out consideration of some parts of our readership by saying "That's THEIR problem, not ours": it's not - it's our problem too. - SchroCat (talk) 15:06, 3 February 2020 (UTC)

To repeat: "Journalists and bloggers can still advertise featured articles to their readers whether they're on the Main Page or not. They just have to direct the readers to the actual article." That would seem to me to make it EASIER on them and clearer to THEIR readers. Instead of writing a piece about Article X which is the TFA and linking to the TFA section of the Main Page, they would write about Article X which is the TFA and link to Article X. This would also benefit their readers by meaning that they could read the blog days, weeks, months, or years later and still be able to read the article in question without lots of detective work. --Khajidha (talk) 15:17, 3 February 2020 (UTC)

Repeating the point doesn't make it more agreeable. They normally link to the article anyway (and I've seen some which link to both the MP and the article), but they do so once a day, as does the WMF feed that pops up on my Facebook profile every day. There is no "detective work" needed by anyone at any time - that's something of a straw man (although the reverse is certainly true - I have been asked if I have read the "main article" today - it will need detective work to track down future references like that). The rebroadcasting of our content is free advertising of what we do. One posting a day works well for people and it has the effect of having the same article being publicised on several news feeds/blogs/social media profiles to add to the impact. - SchroCat (talk) 15:32, 3 February 2020 (UTC)

Well, if the majority of these bloggers are linking to the actual article instead of the Main Page section, I don't see why having multiple articles would be a problem. If they are just mechanically blogging about whatever the TFA is, then they can go on doing so about whatever article they see when they go to the Main Page to get their inspiration. If they only do it when there is an article that interests them, then they have a greater chance of finding an article that interests them. And having a wider selection of articles would seem to mean that more people would be likely to publicize them and that more people, with more diverse interests would be drawn in. --Khajidha (talk) 15:42, 3 February 2020 (UTC)

Randomisation (arbitrary break)

I'll go back to this being a solution in search of a problem. I'm still unclear as to any tangible benefits for us or our readers in changing to a random selection. There seem to be more pitfalls than positives and it will likely annoy or confuse many readers. Can anyone give me some benefits that this will provide, or which problems will be solved if we go over to this system? - SchroCat (talk) 15:32, 3 February 2020 (UTC)

There are in principle three advantages:

• One does no longer need to rely on pseudo-significant anniversaries that are mostly meaningless.
• Increasing the cycle rate decreases the imbalance between the TFA that gets too much attention, leading to vandalism and eyes being diverted from other FAs, to non-TFA FAs which get too little attention and thus deteriorate over time.
• Cyling more than one article per day on the main page might increase the number of readers going through.

I think the confusion stems from the fact that there are really two proposals under discussion: Doing away with most of the date significance and increasing the number of FAs featured per day from 1 to several. Perhaps we need two separate discussions. Jo-Jo Eumerus (talk) 15:40, 3 February 2020 (UTC)

Thank you Jo-Jo for coming back on this.

• "pseudo-significant anniversaries": This doesn't need "randomisation" and most days this isn't a factor. We can get rid of the date connection, although I think keeping some semblance of it is useful
• Vandalism will happen to featured articles on the MP. The current benefit is that the current single TFA will have a large number of additional watchers to stop vandalism. For some of the older FAs that have very few watchers, or whose primary editors have left, there is a possibility that vandalism will sneak in without people realising it for a significant period. I'd prefer three-day semi-protection of TFAs instead. Having an article appear on the MP for 15 or minutes is not going to stop article deterioration. FAR is the best way of doing that (although I know that there are problems with low levels of assistance in that part of the project)
• I don't think this will happen. From among my friend group that read the TFA, they do so on the train to work, or over lunch, mixing that time with reading the paper and other activities. I don't see people rushing back to see what we've got coming up next. (I know that's very rough and ready and people will all differ in when they access the MP, but I don't see having rolling TFAs as a way of generating more hits). - SchroCat (talk) 15:49, 3 February 2020 (UTC)

• I think there are a few problems that the randomisation idea was trying to solve, and I liked it in theory, but it seems that a workable randomisation solution will be more work than what we have currently for little discernible benefit to the reader. The date-relation problem is easily solved - stop doing it. If there are genuine strong ties to a date it can still be run then, but there's no need to try and tie everything into a date (this never used to be a problem and probably arose when there started being points for requested TFAs being date-related). I'm not persuaded by the other theoretical advantages. If there's some FA-recruitment/production advantage that comes out of this then that would be an argument in favour, but I can't see one. Yomangani 16:31, 3 February 2020 (UTC)

I haven't read the entire discussion, but after Jo-Jo's post, I like the idea of having more than one TFA in a day. I like using randomization as a means of leveling the playing field, skewing it away a little from the newly minted or the celebration-worthy. Like others above, I have questions. How often will the TFAs change in a day? Every hour, a la RexxS above, is too often. People need time to focus. They might see something in the morning while they are busy, skim it, plan to come back to it at lunchtime, by when it would have gone. Every four hours is better—in a kind of NY Times format: the current-four-hour-TFA in a largish frame above, the five others in a row of small frames below. The important things are 1) More TFAs, 2) more egalitarian choices and 3) less load on the TFA coords and Dank. Also, the TFA-blurb discussions won't loom as large, or wax as long. Fowler&fowler«Talk» 17:42, 3 February 2020 (UTC)

PS, someone like me, who hardly ever skims the main page—maybe once a week, if that—much less reads its contents, is more likely to do both if six FAs were gracing the main page, and the luck of the draw, rather than human diligence, were determining those appearances. Fowler&fowler«Talk» 18:03, 3 February 2020 (UTC)

Fowler&fowler, the work would be increased for the coordinators who schedule because scheduling is a several-step process, see here for further detail. Scheduling 186 articles that way would be laborious and probably too much. Is there thoughts about whether it could be automated in some way? Wehwalt (talk) 18:29, 3 February 2020 (UTC)

PPS @Wehwalt: Maybe I don't understand the underlying computer science, but why 186? In my way of thinking, an ordered random choice of six FAs (of five, when including an anniversary) would be made at 23:59:59 GMT. During the next day, the current- and already-appeared TFAs would be clickable; the others would not. Fowler&fowler«Talk» 18:32, 3 February 2020 (UTC)

I just meant 31 x 6, the largest number of days in a month times six.--Wehwalt (talk) 19:23, 3 February 2020 (UTC)

PPPS Dank already has the 5,000 blurbs. Or the first three sentences of the lead could be the blurbs. The six articles could be locked down for 24 hours, barring glaring errors in the lead, which can be fixed via talk page intimations. If people think other changes are in order, and those changes are important to them, they will come back the next day to edit. The vandals will not. Fowler&fowler«Talk» 18:42, 3 February 2020 (UTC)

I see. I know there are ways to write automated randomization routines. Someone such as RexxS or Saravask (if he is around) or Begoon might have thoughts. Fowler&fowler«Talk» 19:01, 3 February 2020 (UTC)

I'm making a fool of myself, but the code would look like a more sophisticated version of:

(TFARand0, ..., TFARand5) = ordered list of six random FAs among the 5,000 (some version of {{random number|5000|6}} that avoids double counting)
{{#switch: {{#expr: {{CURRENTHOUR}} mod 4}}
|0=]
|1=]
|2=]
|3=]
|4=]
|5=]
}} It would change the picture and blurb every four hours. Ideally, you guys shouldn't have to lift a finger. Fowler&fowler«Talk» 19:32, 3 February 2020 (UTC)

• Concerning the "meaningless anniversaries". The last one I was involved with was a celebration of Apollo 11 on the 50th anniversary of the Moon landing. This was extremely successful, resulting in an enormous number of page views. Run as regular TFAs, the articles did not have the same impact. Hawkeye7 (discuss) 19:38, 3 February 2020 (UTC)
  • I certainly did not use the expression "meaningless anniversaries." In fact, as Wehwalt and others know, I was very much involved in the India page appearing as TFA on Gandhi's 150th birthday last October. Some judicious selection criteria will need to be formulated that makes an exception for Apollo 11 or Gandhi (perhaps by some combination of the page views statistics and their Vital Article status), and gives them the page for the whole day. For other days, and for less significant anniversaries, their TFAs would share the day's spotlight with five others. The point is that there won't be any one-month lead time, or month-long process, or TFA talk page arguments. The blurbs would either be Dank's pre-made ones, or the first three or four lines of the lead. The blurbs, their fine-tuning, their spit and polish, has much greater significance to the proposers than to the average reader. That is not to say that the blurb is not important, only that the article is more important. Fowler&fowler«Talk» 19:54, 3 February 2020 (UTC)
    • But why should "good" anniversaries have to share with others? This is one of the drawbacks to the process, as far as I am concerned, and if people are used to seeing five articles a day, they are going to be pointing out that the MP is broken because only one article is on there on that day. Determining what is or is not considered a 'valid' anniversary for an article is always going to piss someone off if an article isn't deemed to have sufficient connection to a date. I still see very little benefit in random selection, and more problems than it is worth. - SchroCat (talk) 20:24, 3 February 2020 (UTC)
      • There are all sorts of ways to make the process adaptive, that a computer scientist can code for. I'm at the outer limits of my CS knowledge. One of the people, I mentioned above needs to say something here about my idea or ideas. If there is something to be said, that is. Fowler&fowler«Talk» 20:56, 3 February 2020 (UTC)
        • This isn't about an "adaptive process", it's about the reasons behind why on earth we should do it. Missing an opportunity for running genuine anniversary articles isn't beneficial. The alternative of getting people used to having five articles a day then going back to one for an anniversary day will confuse people. This isn't about what a computer scientist or coding can do, it's about whether or not this is beneficial for both our readers or editors. - SchroCat (talk) 21:09, 3 February 2020 (UTC)
          • It is not five, but six. My reasoning is the following. We have a large readership. Let's say it is ten million people. An average reader is a random reader. His or her top six topics of interest is a random set of six topics in Misplaced Pages. When we chose six FAs randomly, we maximize the chances of presenting something that will intersect with his/her top six choices. (We could increase that by having the TFAs change every hour, but then we run into other issues you and I have both pointed out above. So let's say, for now, we have settled on six.) The average/random reader is more likely to pay attention to all the TFAs if there is something there that is of special interest to them. They will read their article of primary interest, and very likely peek at others. Otherwise, it is highly unlikely they will, unless they religiously read the main page and the TFA every day. But an average reader does not do that. Fowler&fowler«Talk» 22:00, 3 February 2020 (UTC)
            • Misplaced Pages needs to decide whether the TFA is for its editor(s), their cohorts, their larger milieu of editors, or for the random Misplaced Pages reader. Fowler&fowler«Talk» 22:07, 3 February 2020 (UTC)

Objectives

With mixed feedback, maybe more time discussing the objectives will be helpful. I won't put this in enumerated list form, because that would imply an order of priority, and each person may assign a different priority to the problems that we need to solve.

For context, Featured article statistics can be viewed.

Declining participation in both FAC and FAR, declining participation as well at WP:TFAR (who remembers when we had hundreds of articles/editors clamoring for TFA), increasing reliance on (dubious) date significance at TFAR, not enough FAs being generated to fill main page slots, need to repeat TFAs, diversity of TFA, ways to re-invigorate FAC and FAR, how to get large number of old FAs brought back up to snuff, TFA as a recruitment tool lost by scheduling so far in advance that we end up running only "semi-perfect" FAs, giving no incentive to others to join FAC and FAR ... And so on. Others may have other objectives.

But, if the Featured article statistics continue to be what they are, and the FA community doesn't do SOMETHING, we risk having the broader community make some sort of decision at some point down the road. There is a need to clean up. Perhaps an RFC is premature and more brainstorming is needed, but this is not a solution looking for a problem. There is a problem. The main page does not belong to the FA community; the spot was earned in the past, and it should continue to be earned. If there is limited interest in cleaning up old FAs, finding ways to increase participation, or finding ways to provide something useful to the mainpage TFA slot, we will eventually pay the piper. SandyGeorgia (Talk) 16:04, 3 February 2020 (UTC)

You may not think that describing "randomisation" as a solution in search of a problem is terribly good, but there is such a disconnect between the idea of randomly generated FAs and the perceived problems in FA-land that I was genuinely at a loss to come up with any connection between the two - which is why I had to ask at the end just what we were actually trying to solve. "Randomisation" may or may not have other benefits, but sorting out declining participation at FAR isn't one of them.

I am unconcerned with low turnout at TFAR: I used to use it regularly, but do so very rarely now because I know we have an excellent team of three overseeing the process. When it was just Bencherlite on his tod I would nominate to ensure he didn't have quite so much to do on his own. TFAR isn't a problem.

Low attendance at FAR is a problem. It is at FAC too, although the figures have ping-ponged up and down for several years. I suspect (though don't quote me) that most editors are completely daunted by the FAC process, the reliance on high-level academic sources and the need for high number of sources and citations, as well as the laundry lists of prose changes that appear in every nomination. To me having those requirements is a Good Thing. It sets FAs out from the rest, regardless of the dubious honour of a front page view. I do agree with something you've said several times, that the FAs are only as strong as the weakest ones, so FAR has to be more effective or populated or active than it is at the moment, but I am afraid I do not have any suggestions for how to improve it at the moment. I just know that TFA randomisation isn't the answer. - SchroCat (talk) 17:27, 3 February 2020 (UTC)

Just trying to encourage ideas. (I would personally be happy just to convince people we need to deal with the older, non-compliant FAs, but my efforts have not been fruitful; barring that, brainstorming is a good thing!) SandyGeorgia (Talk) 17:46, 3 February 2020 (UTC)

Here is another set of data. (I am traveling and not on a good connection or computer, so there may be errors in my data, in case anyone else has time to look; even if there are some errors, the general trend is true and obvious). Taking the last 12 years, and splitting them in two periods:

• Six years encompassing 2014 1,721 nominators for 4,133 FAs to 2019 2,095 nominators for 5,695 FAs
• Six years encompassing 2008 1,008 nominators for 1,789 FAs to 2013 1,721 nominators for 4,133 FAs

• 713 new nominators for six-year-period 2008 thru 2013
• 374 new nominators for six-year-period 2014 thru 2019

• 390 FAs per year in six-year period encompassing 2008 to 2013
• 260 FAs per year in six-year period encompassing 2014 to 2019

• 1.8 nominations per nominator at end of 2007
• 2.4 nominations per nominator at end of 2013
• 2.7 nominations per nominator at end of 2019

Summary: FAC is turning out increasingly fewer FAs, written by increasingly fewer editors. New FA writers are appearing less and less frequently (by a large margin). More and more, FA writing is the domain of fewer editors. BOTH FAC and FAR need to be re-invigorated. SandyGeorgia (Talk) 17:53, 3 February 2020 (UTC)

One more point: some subject areas are less popular than others, and this trend is also continuing. The attempt to vary the subjects appearing on the front page, resulted in WP:FANMP becoming even more unbalanced than the pool of featured articles, and the complete depletion of available articles in certain areas. Hawkeye7 (discuss) 19:33, 3 February 2020 (UTC)

Not having anything constructive to offer, I have merely been page-watching the discussion. However, it seems to me that SchroCat summarises the main points pretty well just above: randomisation will not increase FAR participation, and may well decrease FAC nominations; TFAR is fine; declining participation at FAR and FAC are issues; newer editors with FACable articles are daunted by the process and actively deterred by reading FAC discussions; "something" needs to be done to improve FAR. If the discussion could kinda reboot from there it seems to me that it would be moving in a profitable direction. Gog the Mild (talk) 20:21, 3 February 2020 (UTC)

Featured article appearing as 'Today's Featured Article' multiple times

@TFA coordinators

I note that some featured articles have appeared as 'Today's featured article' more than once. An example is Rosetta Stone which was 'Today's featured article' on September 14, 2010, and on March 18, 2017.

Please can you tell me why some articles appear as 'Today's featured article' more than once, when there are other featured articles which have never appeared as such. Thank you.Boleslaw (talk) 13:02, 24 January 2020 (UTC)

The discussion which led to the changes is here. There's a list at Category:Featured articles that have appeared on the main page twice. ‑ Iridescent 13:08, 24 January 2020 (UTC)

Thank you, I wasn't previously aware of this and now I am! Boleslaw (talk) 22:09, 26 January 2020 (UTC)

WT:FAC#TFA proposal

Just a pointer. - Dank (push to talk) 19:11, 31 January 2020 (UTC)