Research talk:Patroller work load

From Meta, a Wikimedia project coordination wiki

Related and unrelated editing[edit]

In my experience vandalfighters and New page patrollers have broadly three types of other edits:

  1. Directly related edits such as warnings to the vandals they've reverted and speedy deletion notices to editors whose pages they've tagged for deletion. These edits are not usually obligatory, but they are strongly recommended and a high proportion of patrollers will do them (especially the ones who are aiming for RFA).
  2. Indirectly related, such as welcoming new editors whose articles they've marked as patrolled (as opposed to the welcomes included in tagging newbies articles for deletion or the "welcome warning" option for level 1 vandalism by newbies), and making minor wikification, copy edit and categorisation edits to the pages they are patrolling
  3. Unrelated: such as editing older articles that haven't just been vandalised.

If editors are feeling stressed and are struggling to keep up with the pace at recent changes or NPP then I would expect to see the 2nd and 3rd decline. If they are being crowded out, for example by finding that they can't revert vandalism before Cluebot or a huggler does then I would expect to see them drift to another area of editing such as the back of the queue at NPP. WereSpielChequers 13:12, 8 June 2011 (UTC)[reply]

Thanks for the notes! I'll be trying to find ways of incorporating the direct and indirect activities into my dataset. --EpochFail 16:07, 8 June 2011 (UTC)[reply]

Impact of RFA[edit]

One of the things that might have altered this pattern is the consensus or at least blocking minority at RFA for editors who are not pure vandalfighters but have also made content contributions. Another is the deprecated and distinctly minority view that a high percentage of automated edits is a reason to oppose at RFA (I know one editor who isn't running at RFA because he would need to make tens of thousands more manual edits even to get his automated count down to 50%). WereSpielChequers 13:23, 8 June 2011 (UTC)[reply]

Problem with the data[edit]

I noticed that I'm 17th in the 2010 data, however, I never patrol new pages -- and I'm not the only one like me being included in the data. The problem is that due to either sysop rights or the autopatrolled flag, every new page generated by these users is generating an entry in the patrol log, no matter of namespace. If you want the data to be more useful, you need to rerun it, and then time remove all log entries that refer to the patrol action being done "automatically", as this is not indicative of any work actually being done by the user who appears in the log. Courcelles 00:21, 10 June 2011 (UTC)[reply]

Yeah, we're actually still working on it as we speak, and when we did QA on the list of names we noticed we forgot to exclude admins/autopatrolled from the list -- there are some people who created thousands of autopatrolled pages who were outliers. It will be fixed shortly, because thankfully the log data marks whether it was autopatrolled or not. Thanks for paying attention and for the helpful comment Courcelles. :) Steven Walling at work 00:29, 10 June 2011 (UTC)[reply]

Back of the queue[edit]

One of the features of Newpage patrolling is that unlike recent changes you can see whether other editors have checked an article, that is provided they have marked it as patrolled (sometimes you see an unpatrolled article surrounded by patrolled ones, very often these will be something like a Roman consul that most New page patrollers feel uncomfortable classifying). You can also go to the back of the queue and patrol the oldest articles not yet marked as patrolled, the back of the queue oscillates with thirty days as an absolute maximum after which all surviving articles are in effect marked as patrolled.

Almost all patrollers work at the ends of the queue, and most of them at the front. Obvious vandalism pages and attack pages are often deleted within minutes and rarely survive even for a few hours. Typically 200-300 articles a day are left unpatrolled by the front of queue patrollers and left to the back of the queue patrol, many of these are borderline articles that don't quite meet a speedy deletion criteria, and I suspect many have been looked at by at least one patroller.

The stress factor works very differently at the two ends of the queue. At the front of the queue it is very immediate, trying to keep up with the full flow of all new articles is like supping from a firehose, and anyone who is the lone patroller at a given moment of time will see far more articles in yellow (unpatrolled) than they can individually keep up with. The back of the queue works on a completely different timescale and the psychological pressure there revolves around the distance between the end of the queue and the thirty day cutoff. One measure of that stress would be the number of articles that are automatically patrolled at the end of the thirty days, another measure would be to see how the length of the queue oscillates. There have been at least a couple of occasions in the last 18 months when an individual editor has cleared the queue, patrolling several thousand articles in a catch up session that will take some days of focussed activity. WereSpielChequers 17:14, 13 June 2011 (UTC)[reply]

Do you think it would be interesting to have a tool on toolserver which reports how many editors are actively patrolling new articles, and which side of the queue they're working from? —SW— spill the beans 14:10, 4 July 2011 (UTC)[reply]
Yes, but it depends on the definition of active. For the back of the queue patrol it doesn't matter if no-one is active in a particular day, the interesting thing is where the back of the queue has been over the last 7 days. So a report that said that "the back of the queue is currently 29 days 12 hours and has fluctuated between 26 days 2 hours and 30 days in the last week with 98 articles unpatrolled articles reaching the end of the queue." Would be useful and if necessary could be used to call for volunteers. But the front of the queue is very much focussed on the first minute or two of an article's existence, if we have a gap there then it would be useful to know in realtime, and also to know where are gaps are around the clock. So I'd like to see two reports re the front of the queue, one an NPP equivalent of the Wdefcon userbox to tell those interested whether newpage patrol is currently operational - I'd suggest that it just show the proportion of the unpatrolled new articles created between 3 and 15 minutes ago that have been deleted or marked as patrolled, 0% implying that no-one is currently on patrol and anything over 60% being in various shades of green. The other front of the queue report that would I think be useful would be a chronological one that looked at each hour in the last 7 days and gave the ratio between patrolled and unpatrolled new articles (ignoring pre-patrolled ones). I think this would tell us when our gaps were at NPP and which batches of recent new articles might still contain G10s and G3s. I could certainly see myself using that report for the latter purpose, and possibly the former. WereSpielChequers 11:17, 5 July 2011 (UTC)[reply]
Well I've gone ahead and implemented my first idea, just a real-time report of who's been patrolling recently, the average age of the articles they're patrolling (i.e. which side of the queue they're working from), and how fast they are patrolling each article (see here. At the very least, patrollers could use this to better coordinate which articles they're patrolling. If 5 people are already falling over each other to patrol from the front of the queue, there's no sense in me trying to do the same and cause more edit conflicts. I think some of your ideas described above are quite good as well, and while some of them don't suit themselves to being run on toolserver, they are pretty easily implementable nonetheless. I'll take a look into it further and see what I can come up with. —SW— squeal 22:41, 5 July 2011 (UTC)[reply]
Thanks SW, that report certainly fits in with the theory that we have two distinct teams working at each end of the queue. Time between articles doesn't tell us whether the the slower editors have slow connections or are taking the time to fix a few typos and add a few categories rather than just patrol. But I suppose we should look at the very fast taggers as they are the ones who are likely to be cutting corners. WereSpielChequers 00:18, 6 July 2011 (UTC)[reply]

Improving the back of the queue process[edit]

I've filed a report on the EN wiki village pump suggesting that we show the patrolled option to wikignomes who happen to come across these articles anyway WereSpielChequers 11:17, 5 July 2011 (UTC)[reply]

Now moved to https://bugzilla.wikimedia.org/show_bug.cgi?id=15936 apparently this was raised a couple of years ago but the fix had performance problems. WereSpielChequers 13:58, 5 July 2011 (UTC)[reply]

Stats[edit]

It's possible some the participants here are not aware of the research that was done by Snottywong and myself since around October last year. (I know WereSpielChequers has been following what we have been doing). The background is here:

and the recent table we made: This is a list of users who have recently patrolled articles. The bot which generated this table looked at the most recent patrols from mid-October 2010 to mid-March 2011 that were performed in the Main namespace (413,902 patrols total). It counted how many articles each user patrolled, as well as each user's total edit count. It then generated this table, sorted by "Edits per patrol". The users at the top of this list are those who have both a relatively low total edit count and a relatively high recent patrol count. It also lists the date of each user's first edit. The bot initially found a total of 3,310 unique users who have patrolled articles in this time frame, but skipped any users who have patrolled less than 40 articles. This resulted in 882 unique users. The table was generated at 19:51, 22 March 2011 (UTC).

I have done a lot of patrolling, including of pages already patrolled. My own subjective empirical findings were:

  • I'm not convinced it causes stress - claims of 'burn out' are exaggerated. However, new pages arrive - and get left as unpatrolled - much faster than I can work. Nevertheless, NPP is far from my main work here, but I have an 'NPP' window open while I'm working on other stuff.
  • Large backlogs are cleared by experienced users and admins who occasionally have a stab at it.
  • Patrolling is done mainly by new and/or inexperienced users. Some of the reasons are touched on at Fetchcomms' essay
  • Haste, is probably one of the reasons for poor patrolling, and not using patrolling as a means to catch persistent vandals and socks. This means a little more research is needed to be done by patrollers than just clicking for a Twinkle template. The blatant COPYVIO, attack pages, and hoaxes, need high priority and especially extra checks made on the creators.
  • Patrollers do not check page logs for deleted pages/recreations.
  • The pages in the 30-day backlog are due to being passed over by new and/or inexperienced users who generally work from the live feed. They are not too difficult for experienced users to resolve.
  • Many patrollers forget to click the 'mark this page as patrolled' link (which I feel should be graphically represented by a proper button).
  • Some patrollers are using AWB and I don't think this is a good idea.
  • There have recently been software glitches due to system and Twinkle upgrades.

I'm not entirely convinced that the stats shown in the graphs reflect the true situation. If fewer pages appear to be being patrolled it's probably not due to a drop in creations. Kudpung 07:34, 26 June 2011 (UTC)[reply]

I have to agree with Kudpung. The statistics you show right now are only showing that New Page Patrollers do less work, not that there is less work to be done. I think we are looking at a Patrolling quality issue. More pages are getting scarily close to the end of the NPP log and are being rushed through the process, because the NPP'ers want to make sure that the page gets patrolled. In the end, this means that the New Page Patroller, when patrolling pages, feel pressure not to spend more one on one time with new users that create good content, or investigating the work of users with fairly major problems. Also, many of the users in the top fifty are working on other backlogs at the same time, coming to New Page Patrol as a secondary thing, so the regular patrol focused individuals are surely feeling pressure, because they aren't getting consistent support and community. I think the concern ought to be: how can New Page Patrol create even less work for the patrollers that way if they do run across something good, bad or worth improving, they have the opportunity to interact with the user and content more fully and that won't contribute to an ever-longer feed, Sadads 09:07, 26 June 2011 (UTC)[reply]
That seems to bear out what I was saying. Nevertheless, at the risk of over emphasising the issue of new/inexperienced editors, this is indeed an area where many choose to work who are not ready to create articles, add content, or do other clean ups such as copyediting and adding cats, stubs, or add project banners to the article talk pages. They are probably unaware that all these templates add categories which more experienced editors work through. Nevertheless, our Snotbot now puts a magnetic collar on the articles that slip off the 30-day special:page, so we have this extra category to keep trace of them, and this should reduce the pressure. Nevertheless, it's more grist for the backlog mills. Kudpung 09:36, 26 June 2011 (UTC)[reply]
Thanks for the extensive feedback, both of you. =) Snottywong's count of patrollers from a span of 2010-2011 was partly the reason we were interested in this. It was great. There's a lot to respond to, but I wanted to say that I agree with Kudpung about the theory of overwork or burnout -- it's not conclusive but so far the study suggests that the relative amount of work done by each individual patroller hasn't been increasing over the years. (And Sadads: yes, the study doesn't show the rate of new pages overall, just the amount of patrolling actions done by each patroller.) As for the reason why or other factors I'm not sure, though it is really interesting to hear that a lot of patrollers simply forget to use the button. That may actually skew the numbers some depending on how often it happens. Thanks again, Steven Walling at work 05:41, 27 June 2011 (UTC)[reply]
I'm not sure it is so much that editors don't remember to mark edits as patrolled, I think the problem is that you only get the opportunity to do so if you encounter the article from special newpages. So if I come across a newish article whilst categorising or typo fixing I don't get the oportunity to see that it is not yet marked as patrolled. Fixing that might go some way to resolving the frequent backlogs at NPP. WereSpielChequers 22:43, 30 June 2011 (UTC)[reply]

Some of the major problems, if I remember rightly, are that the yellow page function is now full of bugs since the attempts at upgrading the Wiki software, the yellow page programer has gone AWOL, Snottywong who had a go at putting things right is now extremely busy in RL, Twinkle cannot decide whether it should mark new pages as patrolled or not (and has enough bugs of its own since it was upgraded), and even I am not sure now which pages have been patrolled or not. When I work through the backlog, I get the impression that nothing is working as it should. I have some suggestions to make:

  • All new pages, except those created by autopatroller should have a template placed on the top by the site software: "Unreviewed new article. Some content may not be factually accurate until any issues have been addressed."
  • Pages should be marked as patrolled if:
    • They are clearly acceptable without any further intervention, stub tags are added, and at least one cat is added, and any blatant format errors are addressed such as reference tag missing, reflist tpl missing, indents leaving ugly grey text boxes, vestiges of the Wizard's template page ('page name is', demo image, etc, - for all this the Wizard needs to be rethought).
    • Acceptable after being appropriately tagged for maintenance tasks.
    • New page patroller tasks at WP:NPP should either be taken into policy or become an official guideleine.
    • Convert the now largely redundant 'reviewer' user right to 'patroller', and require new reviewers and/or patrollers to have reached at least Novice Editor. If they get three reminders in any four week period for any incorrect tagging or failure to mark as patrolled, they temporarily lose the right until they have been mentored and tested on their knowledge of NPP and WP:DELETION..
  • A proper button on the page instead of the 'Mark this page as patrolled' link: "Press to pass this page as OK".

--Kudpung 01:53, 1 July 2011 (UTC)[reply]

The three reminders within a four week period part could backfire. Some people (myself included) have done enormous amounts of NPP at a time; in Snottywong's stats above, it'll show I did over 10,000 new pages in 7 months, and I can say at least a thousand of them were between February 15 and March 15. I think I got 3 reminders in that time frame, and I probably had around 5-6 misfires (not counting times when someone improved an article after I tagged it), but that was around .01% of my patrolling during that 4+ weeks; yanking that proposed right in a similar situation would seem unproductive. It's a good concept, but it should be a guideline as opposed to a bright-line rule. Other than that, if it were up to me I'd copy them over to en.wiki and make them policy now. If we're going to force users to have some experience before creating new articles, it should also follow that they have experience before patrolling them. The Blade of the Northern Lights (話して下さい) 02:28, 1 July 2011 (UTC)[reply]
  • I agree that the ability to patrol articles should be controllable in some form or another. My opinion (for en-wiki, at least) is that the user right should only be granted if requested, like rollbacker or reviewer, and there should be some minimum edit count requirements for it. There should also be a patroller noticeboard where serially poor patrollers can be reported and eventually have the user right revoked if they fail to improve. Until policies like this are put in place, there will be no way to have control over the quality and effectiveness of new page patrolling. The downside to policies like this is that it might diminish the quantity of patrollers. However, some editors (including Kudpung and Blade) have recently gotten consensus to conduct a trial whereby article creation would require autoconfirmed status. According to some stats I compiled, this would theoretically reduce the number of new articles created each day by about 20%, and reduce the number of new articles deleted each day by over 50%. We're on the right path, but some more work is still required. —SW— spill the beans 04:22, 1 July 2011 (UTC)[reply]
I agree that we need to do something about poor quality patrolling, but that mistakes need to be put in context. Somebody who gets their first three tags wrong needs help, someone making 3 mistakes in a thousand might deserve a friendly tease whilst a consistent 3 in 10 is troubling. I think one problem we have there is that admins are usually judging individual nominations, and that makes it easy to miss patterns. But we do have patrollers who will tag as a speedy if they think an article wouldn't survive AFD, and that way lies lost articles and bitten newbies. The difficulty about looking in to the subject of poor patrolling and having a noticeboard to discuss egregious examples is that as we saw from EN:EP:NEWT this is a very contentious area, and a noticeboard for naming and shaming those who make mistakes is bound to cause Dwamah. Even vandals only get reported to AIV after four previous warnings.:
  1. We are all human[citation needed], when we make mistakes we prefer a quiet word on our talkpage rather than a public pillorying.
  2. Those who think of this as a space invaders game find it very disconcerting when the space invaders start shooting back.
  3. Some deletionists use speedy deletion to operate a de-facto much more deletionist policy than the community would accept by consensus. Enforcing existing policy at speedy deletion thus precipitates a confrontation between inclusionists and deletionists. Personally I suppose I'm more of a Proceduralist, I don't greatly care whether we relax notability to allow semiprofessional footballers in or tighten it to exclude people who are only known for their role in sport or "popular culture". Either way the articles that I really care about will still be here. I'm more concerned that the rules that we set out to our new editors should be the ones that we follow. If we told the article creators that all new articles must have a reliable source, but we actually tagged for deletion on a lower standard then I think we would have a workable system. But tagging to a stricter standard than we tell the article creators they have to work to causes much of the carnage that we have at NPP. WereSpielChequers 07:42, 1 July 2011 (UTC)[reply]
(edidt conflict) SN's recent stats cover a period of 6 months up to the end of this month. The trial should also run or 6 months in order to cover any seasonal variations (e.g. a lot of the crap comes from schoolies). I am confident that the results will prove positive towards making the new rule a firm policy. It will therefore reduce the load on patrollers, who will/should then have time to complete the tasks listed at NPP. It's hard to convince that user rights (such as sysop too) are no big deal, but for most of those who do NPP and RCP it will nevertheless very much be considered as such, and may be a greater incentive for them. Kudpung 07:54, 1 July 2011 (UTC)[reply]
I concur with WSC - the articles we really care about will still be there. There is no need to change existing policy on deletions, we only need to get the 'unlicenced' patrollers to read NPP and WP:DELETION before they start work. Giving them a 'rights flag' may solve the problem. Putting friendly remarks on their talk pages if they get it wrong is of course the best solution, and it's what I do, even if it takes up time. I once redesigned a couple of Twinkle templates, but in hindsight I dropped he idea because I find my personal messages work better - to the extent that I now get asked for advice by a lot of NPPers. Kudpung 08:00, 1 July 2011 (UTC)[reply]
I used to do personalised messages, and switched to twinkle becasue, well it is too easy. But using it for declining speedies has a number of drawbacks, not least that it doesn't watchlist the talkpage that you've just left a message on. When I was having conversations with patrollers I was able to persuade some to install hotcat and above all to stop trying to keep up with the whole flow themselves. I think that is the route to stress, if you happen to be the only active patroller for a session and each time you refresh there are another batch of articles then it is way too much like Tetris or Space invaders at the point where you are being overwhelmed. Better by far to think of this as a conveyor belt, and if we each pick off what we can comfortably handle eventually it gets done. I agree that we need to be able to bar some patrollers from new page patrol - especially the ones who tag for speedy anything where they would vote delete at AFD. But I'd treat that as a last resort - what i think would really help would be some sort of report tool that let the taggers measure their accuarcy level. WereSpielChequers 15:29, 1 July 2011 (UTC)[reply]
I haven't seen any evidence that overzealous speedy taggers are a big problem. They can tag as many articles as they want, but an admin needs to make the final decision on whether that was an accurate tag. If they're wrong, they'll be notified about it, and maybe the only collateral damage will be a confused/irritated newbie. Also, to be clear, my noticeboard idea above was only intended to be a place to report patrollers who are constantly doing a poor job (i.e. the "3 in 10" editor), not as a place to point out ever editor's shortcomings. If the new userright is granted, then there needs to be a place for discussions to take away that right. It's possible that ANI would serve this purpose just as well. —SW— converse 19:22, 1 July 2011 (UTC)[reply]
Experience tells if a page name is suspicious, particularly such as those that include a domain name, a proper name with an uncapitalised surname, an obvious band name, Hispanophone names (most likely South American footballers) etc., etc. When I patrol from the live feed I cherry pick both those that are highlighted and those that have been patrolled. I don't actually come across many CSD that need to be declined. What I do find however, is a disturbing number of CSD that use the wrong criterion. Some people just don't know the difference between gibberish, unsourced negative BLP, vandalism, hoax, and blatant advertising. G10 attack pages tagged as A7 are the most frequent offenders. These are just the basics - one can only wonder if the patrollers have really read NPP or CSD. When I find wrong criteria, I add 'AXX, not GXX as tagged' to the deletion summary, and if the patroller is not making many errors I leave it at that. If I find the patroller is making a lot of similar mistakes I leave a personalised 'Just letting you know...' message. (BTW: It might be an idea if Twinkle could make provision for articles that fit multiple criteria, eg: A7 + G11 ). Kudpung 04:36, 2 July 2011 (UTC)[reply]
It does have one, but it's not particularly intuitive. It has a "Tag with multiple criteria" button, but it's very easy to miss, and once you add the necessary criteria there's another button you have to press instead of the usual "Apply tag" button. The Blade of the Northern Lights (話して下さい) 20:20, 2 July 2011 (UTC)[reply]
Tanks for the heads up Blade. I'll look out for it. twinkle dropsdowns are now getting excessively long. The lists should be grouped into categories of similar criteria, divided by a horizontal rule. Kudpung 07:27, 3 July 2011 (UTC)[reply]
(outdent) re SnottyWongs's comment that overzealous speedy deletion taging isn't such a big problem as "maybe the only collateral damage will be a confused/irritated newbie." Firstly confused and irritated newbies are a big problem, we lose an awfully high proportion of new editors and something that frequently results in newbies being confused and irritated is a problem. But this isn't the only damage coming from overzealous tagging, it wastes time for admins and others who have to remove those tags, it does sometimes lead to mistakes where overzealous admins delete articles that almost met a CSD criteria, and if frequently results in unpleasant trainwrecks at RFA. I've been watching RFA for a couple of years now and overzealous or frankly incorrect speedy deletion tagging is one of the most common causes of the failures of experienced editors at RFA, it'd also much harsher for all concerned to have RFAs where the canddiate's mistakes become the focus rather than RFAs where everyone accepts that the candidate is on the right path, but there is disagreement as to whether are sufficiently experienced. So please don't underestimate either the scale or the seriousness of this problem. WereSpielChequers 08:35, 4 July 2011 (UTC)[reply]
I agree with most of that except for the loss of new editors. We lose a lot and I sincerely hope we do - as long as they are nn companies, SEO agents, and incorrigible vandals who post attack pages. However, there are not, and never can be, any metrics to demonstrate how many future Wikipedians we lose who might have stayed if they hadn't been bitten, or their clearly inappropriate articles was rapidly deleted. Let's also not forget that SPA very rarely, if ever, come back to see what might have happened to their article (PROD is generally the proof of that). I know of only two newbies in the last two or so months who have given up due to the way they were treated, and two established editors who have left in disgust. I do WP:BEFORE I tag or physically delete because the policies/guidelines say we should, but it's the most reluctant part of all the tasks I do at Wikipedia; it does have some small compensation though in occasionally discovering a sock or a long-term vandal. Kudpung 12:32, 4 July 2011 (UTC)[reply]
Zealous speedy deletion tagging should suffice for getting rid of the people who aren't here to help, my concern is about the overzealous taggers. As for the numbers, we need stats - we know that the community is gently declining and that our already poor ability to retain useful newbies is getting worse. What we don't know is how often we incorrectly tag and even delete at newpage patrol, and we don't know the proportion of goodfaith and useful editors who we lose in the process. My impression is that it is a more significant problem than you think, that could be because one or both of us is misreading the situation, or it could be because I'm more optimistic than you about the potential of some of our goodfaith but not yet entirely clueful newbies. WereSpielChequers 12:54, 4 July 2011 (UTC)[reply]
My personal points of view, albeit subjective, are based on my own new page patrolling, and patrolling the patrollers, so I admit I could be misreading the true situation. Nevertheless, there is no way one can gather and extrapolate stats for the reasons for the decline, nor to count the people out there who might become Wikipedians. It's my view that it's because most of the traditional encyclopedic articles have already been written, and are now maintained by regular editors, or the new, sensible articles are created by autopatrolled users.
However, a further complex set of stats has now been prepared, again by Snottywong, and will be used as the basis for a trial that is shortly to be implemented. The trial should show to some extent, if our assumptions are right or wrong. The immediate issue remains, as I said earlier, to educate the patrollers - it's too much to expect the admins to constantly duplicate their efforts and clean up after them. Kudpung 13:24, 4 July 2011 (UTC)[reply]
I agree with the approach of educating the patrollers, though I think some admins also need a bit of guidance as well. I suspect we need to identify and resolve some of the issues where custom and practice diverge strongly from policy, as these in my experience are the most confusing for patroller and newbie alike. As for how complete the wiki is, that varies sharply by topic, there are vast gaps where indeed a traditional encyclopaedia wouldn't have an article but where the subject meets our notability requirements. Though I'd concede that a high proportion of the new articles that we get are at or beyond the threshold of notability in contemporary western popular culture, rather than addressing our gaps elsewhere. WereSpielChequers 09:05, 5 July 2011 (UTC)[reply]
Again, there is no way of knowing exactly what the new pages are that we get now because they mostly get deleted before they get a cat. We can only go by subjective reports like mine; perhaps we should poll some of the more prolific and reliable patrollers and get their views. However, I would say that most of them are nn companies who (deliberately?) mistake us for a B2B directory, children who think it's a new site to replace the ailing MySpace, barely notable soccer players (there's enough there to make a stand-alone WikiFooty), nn SEO companies from the Indian sub-continent, autobios from wannabe book writers, and campaign managers for political candidates - and probably in that order. There are occasionally batches of stubs from entomologists, perhaps we should check on them and make them autopatrolled. Kudpung 10:18, 5 July 2011 (UTC)[reply]

Conclusion[edit]

Currently this sprinte concludes that pressure at NPP has not greatly increased as the total amount of patrolling is down, as is the amount of patrolling per patroller. But looking at it from a quite different perspective I come to the opposite conclusion. Despite a huge increase in the number of autopatrollers, the back of the queue at NPP has increasingly often been at thirty days, and that means that articles are being released into mainspace witout having been marked as patrolled. There are several drawbacks to measuring the busyness of patrollers in terms of the number of articles they patrol, not least that this makes no allowance for their other wiki activity. By contrast measuring the number of articles which leak to mainspace without having been patrolled, and how close the back of the queue is to the thirty day cutoff gives two clear and measurable indicators as to how overloaded the patrollers are. WereSpielChequers 13:43, 7 July 2011 (UTC)[reply]

The back of the queue has spilled over since as long as I can remember, although recently it has only been hovering at around 28/29 days. There was a short respite over the Xmas/New Year period which also coincided with Blade's massive clean up. SW has the stats on the patrollers and could probably run the script again to update it, together with some added criteria if required. We already have a bot that is keeping tracks on the articles that slip off the edge. Kudpung 15:49, 7 July 2011 (UTC)[reply]
Yes, I'm pretty sure that hitting the 30 day limit has gone from a rare event to near normal in the last two to three years. SnottyWong's new report shows one aspect of the problem quite neatly, though I appreciate only for the future, not for the period in question http://toolserver.org/~snottywong/cgi-bin/patrolgraph.cgi - of course most articles are patrolled or deleted at the front of the queue so it isn't as bad as it might seem at first glance - we also need figures for unpatrolled articles reaching the 30 day limit - but presumably Sw's bot or its tags would give that? WereSpielChequers 16:37, 7 July 2011 (UTC)[reply]
Snotbot puts them into Category:Unreviewed new articles. However, by the time we had got the bot authorised and up and running it was already late December and its start coincided with the massive sweep Blade and another editor made. There were then glitches earlier this year when the site software upgrade introduced a lot of bugs. We think we may have missed several thousand, maybe not. There were also the problems of Twinkle marking as patrolled or not marking as patrolled. We need a stable period of probably another 6 months to get a realistic picture - ironically, we need to stop Blade from patrolling to get that! The picture will change again dramatically when the other trial starts.
On a slightly different aspect, I've just been looking at the deletions of a few editors who asked me to review their tagging performance. The overall impression I get is that they tag the easily identifiable ones such as A7 and G10. They obviously leave the slightly harder criteria for other patrollers - who I assume to be from what I see, a handfull of admins. Kudpung 18:08, 7 July 2011 (UTC)[reply]
I expect that most of us are a little selective in what we patrol, for example I rarely bother to look at ones that look like company names as I try to rescue what I can and I don't feel motivated to help people writing about businesses. I think anyone who tries to methodically look at every new article is setting themselves up to fail and get stressed. Also I think we all rely on corensearchbot to check for copyvio. Aside from that A7 is I think the most common criteria - though stats on that would be illuminating. The other fairly common ones include A1 and A3, but anyone hovering at the very start of the queue is supposed to leave those two to later patrollers (which to be practical means that I think we need a Twinkle option "if I hit A1 or A3 then hold off till the article is at least 15 minutes old and do the action then if there hasn't been a subsequent edit". WereSpielChequers 20:21, 7 July 2011 (UTC)[reply]
I fully understand your point about 'business' editors. There should be a tightening up of policy regarding these. It's shameful that they expect to get free advertixing on the back of our voluntary work.
As you say, looking at every new article is impossible, especially during two major time zones - North America, and the Indian sub-continent - when new articles arrive on an average of every 20 seconds. With all but the most serious and blatant infringements (attack, vandalism, copyvio), which for example, a patrolling admin would delete instantly, it can take several minutes to properly research and tag a new page. In the case of instant deletion, the research for editor misbehaviour, recreations, warning, and blocking, etc., can be done afterwards. The frustration is that while one is doing the required research for a fairly straightforward deletion, someone else has already tagged it too quickly, and all too often, an admin has then already deleted it on face value. A large number of mis-tags are done by editors using Huggle or AWB. I still don't understand the logic in using AWB to systematically apply orphan, nocat, and Wikify tags (IMO, they are the least important) to one-line stubs and doing nothing else. It's not very helpful. Kudpung 04:00, 8 July 2011 (UTC)[reply]

Patrolling, deletion or other outcomes[edit]

The main problem I see in this work is well shown in the first graph, although meant as a simplified representation: the only two outcomes considered for pages are being patrolled or deleted (probably because they're logged and measurable). But being marked as patrolled is not an outcome in itself, it's a tool to find which articles need to be worked on and put in the appropriate process, each with their possible outcomes (one of them is deletion and that's quite simple; if the article survives, many things can happen). What I suggest is that in the last few years editors have found other ways to mark of find articles and edits to be checked: I don't know the details of en.wiki processes, but you can think of the many bots and of the tags put by the abuse filter; this means that perhaps more and more articles are being worked on without the patrolled/unpatrolled distinction: it's almost impossible to measure all the (un)related things which are the actual outcomes of patrolling (some of them mentioned above), but at least I agree that you should measure a) how many articles have not been deleted or patrolled in their first 30 days (you should check also deleted articles; I don't think there are logs to measure the Special:NewPages queue length as correctly suggested above because they're deleted after 30 days) and b) what proportion of new (non autopatrolled?) articles has been deleted, where articles to delete may have been found in other ways.
I'm eager to see the data for other wikis (please do all of them, at least at the level of extracting data and possibly creating some simple graphs automatically): it will be interesting to analyse them and a), b) will be useful to see how/whether the feature is actually used. There will be the additional difficulty that processes are different on every wiki depending on whether recent changes patrol, abuse filter, bots are used and what processes are in place to work on problematic articles/edits; actions have different meanings because marking a page as patrolled can just mean that it's not a vandalism and front-line recent changes patrollers don't need to look at it, or that it's been put in the appropriate process (marked with a warning or no warning needed, for instance), or that it respects all guidelines perfectly, and all possible variations (in general, it means that people who use the patrol log have done what the community has decided to use it for); permissions are given in very different ways and this can also give some clue about the underlying community/social processes. --Nemo 08:20, 14 September 2011 (UTC)[reply]

Thanks for the comments Nemo. To do German Wikipedia for example, the first step is to confirm that new page patrolling generally works in the same way. I know that DE has "Spezial:Neue Seiten" but is there a project/documentation and a community around it? Or does DE use another mechanism primarily? The same goes for any other language. We could run the same charts for other languages quite quickly I think, but without knowing whether there is a dedicated New Page Patrol I'm not sure it will mean the same thing. Steven Walling (WMF) • talk 17:53, 14 September 2011 (UTC)[reply]
We constantly seem to be ignoring the vast amount of reserch and evaluation of stats on New page patrol that has been being carried out by en.Wik since September last year. Among the major of the major concerns that investigated and confirmed are:
  • New Page Patrol is carried out mainly by the most inexperienced newcomers to the Wikipedia project.
  • A very large number of pages, especially COPYVIOS and attack pages, are patrolled as 'passed' by patrollers.
  • Patrollers will not read the advice at WP:NPP
  • The very few exprienced editors and admins who occasionally do a stint at NPP are unable to catch up with all the wrongly tagged pages and place warnings on the taggers talk pages.
  • One needs to be reminded that en.Wiki is not an American project - it is read and edited by n a pool of over one billion native and near native .
If stats are required, please don't hesitate to ask - there is already a vast catalogue of useful data. Kudpung 13:20, 15 September 2011 (UTC)[reply]
Was this comment really meant for this section? I don't see how it's relevant, can you move or explain it further? Thank you, --Nemo 00:56, 16 September 2011 (UTC)[reply]
It's basically saying that English Wikipedia is the 800 pound gorilla that the WMF wants to pretend doesn't need to be handled differently. We're in a vastly different phase of development from other wikis, but certain people (I think you can infer who I'm referring to) would rather forge ahead with more of the same ideas while blowing off the people currently doing NPP on en.wiki. Another issue we have is the WMF's recruitment of Indian editors; this comes at a price for en.wiki in many ways, one of which is truly horrific new pages because most Indian English speakers don't have nearly the skill required of someone writing an encyclopedia article (in fairness, it's not limited to Indian editors, as we get a large number of ESL speakers from all different languages who aren't much better; I can't speak to how much of a problem this is on other wikis, but given English is the language with the most second language speakers I imagine it's significantly less). I helped compile and interpret some of these stats (I am one of en.wiki's most prolific NPPers; Kudpung, Snottywong, and I are about the only three experienced users who will touch it), and I echo Kudpung above. The Blade of the Northern Lights (話して下さい) 03:25, 16 September 2011 (UTC)[reply]
Ok, so I see that there's a lot of confusion here. But we're entirely offtopic so I'm not going to reply to all your points in this talk page (I disagree with each and all of them). --Nemo 10:40, 16 September 2011 (UTC)[reply]
de.Wiki lists all new pages at Spezial:Neue_Seiten for 30 days. For the period 15:46, 16. Aug. through 16:30, 15. Sep.20, the number of pages crated was 11,578. The Spezial:Neue_Seiten highlights new pages '...that have not yet been viewed.' For this period, no pages were highlighted. Stats gathered at 16:30, 15. Sep.20, so who views them? Kudpung 16:44, 15 September 2011 (UTC)[reply]

Stress[edit]

I see there's been some activity on the front page here today. I would like to emphasise one point: I have never heard any reports directly from any NPPers about them getting stressed or burned out, and I don't know where this conjecture comes from.It's either an idle rumour that was invented as a plausible explanation for the breakdown of of NPP, or was just the experience of one in 800 NPPers who happened to make a joke on a thread that happened to get noticed by people who just love to pounce on that kind of feedback. Volunteers don't get stressed out, because there's no one standing behind them cracking a whip and complaining about the salary they have to pay them. The can get fed up or bored though, and when that happens in America they go to a ball game, in the UK they go down the pub for a game of darts, and in Germany they go to the Kneipe um eine Runde Skat zu klopfen, and if they are in a Thai village called Kudpung they post messages like this. --Kudpung 00:49, 27 September 2011 (UTC)[reply]

I think that "fed up" and "stressed" are sufficiently synonymous for our purposes: either one will result in sloppy work and biting newbies. WhatamIdoing 00:59, 27 September 2011 (UTC)[reply]
For my purposes of getting some objective research done, I would like to know where the official stats are that people get fed up or strssed with doing NPP. Please read the post again. --Kudpung 01:11, 27 September 2011 (UTC)[reply]
I think we fairly infer it from the aggressive behavior against newbies, from the unusually defensive and thin-skinned replies to questions and complaints, and from the fact that most people stop doing it after a while, or only do it occasionally.
We could alternately assume that the unfriendliness and defensiveness indicated bad things about the average (i.e., not the people on this page) NPPer's character and competence, but I think that the behavior of the average NPPer is, after a while, inconsistent with someone finding it a pleasant and relaxing activity. WhatamIdoing 01:25, 27 September 2011 (UTC)[reply]
That NPP is done extremely badly the the vast majority of patrollers is clear, as evidenced by the huge number of wrong tagging and bitey messages. There is no evidence whatsoever that this is in anyway due to burn-out, stress, boredom, or being fed up. Evidence, if any, points to the simple fact that the majority of patrollers are a transient army of adolescents who grasp at the fact that here they have a power, without needing any experience whatsoever, over the world's 5th biggest website, which they belive to be FaceBook or a forum, and where they would have to climb a greasy pole to get promoted to such a peak of moderator power on any run-of-the-mill Internet forum. They do NPP because their Nintendo BlakBox is broken and there is nowt interesting on the telly, they won't read the instructions, and the biteyness comes from their arrogant schoolyard teen-talk. When their game console is repaired, they're off to their shoot 'em up videos where they can experience some real virtual blood and stress. The nearest they get to stress on Wikipedia is when they organise their games to see who can tag the most new pages in half an hour and the winner get a 'Hey, cool man', a barnstar, and a candy bar from the coolest girl in Grade 6. Kudpung 01:48, 27 September 2011 (UTC)[reply]
Well, it appears that you'd rather assume bad character and incompetence in NPPers. I suppose that it could explain most of the data, but not the fact that people stop doing it after a while. WhatamIdoing 23:28, 27 September 2011 (UTC)[reply]
I was acually stating rather than just assuming; but bad character was not part of it. Nevertheless, you appear to have recognised that I'm only describing the average NPPer, and I do appreciate that we appear to be in agreement on that definition. I've light heartedly suggested why they stop patrolling, but what I have clearly outlined are the reasons why the average NPPer does the job at all, and the inescapable fact that they are not suited for it. But I'm not here to discuss these issues - I asked where the official stats are that clearly state that people get fed up or stressed with doing NPP. Those of us who have monitored NPP for the last 12 months have not heard any reports of stree or burn-out, nor experienced any ourselves in spite of sessions at NPP often lasting several hours through the night. Stats already support a lot of our findings, but suggestions of 'stress, and burn-out' are vague and are not documented. Please answer the question if you can. and let me know where they are posted. Thanks. Kudpung 02:42, 28 September 2011 (UTC)[reply]
Hi Kudpung, I see at least two potential causes of stress at NPP. At the back of the queue when people see the 30 day cutoff approaching, and at the front of the queue when people try to keep up with the flow of new articles. The back of the queue you and I have discussed before, and hopefully snotbot resolves something that I could see being stressfull. At the front of the queue there are those like me who see this as a team operation, I will pick off a few articles and handle them while assuming that others will handle the rest. But I suspect there are others who instead of sipping from the firehose put the end in their mouth and try to keep up with the whole flow. If so these could be the patrollers who tag articles as A7 because they haven't read the third paragraph that might include an assertion of importance, or even a reason for it to be a G10. They are probably also the taggers who never add a category but instead add category needed tags. WereSpielChequers 23:54, 1 October 2011 (UTC)[reply]

I've not had time to comb through all the replies, but I just wanted to say: the stress/burnout thing was a hypothesis, not a statement. It was based on the fact that we do hear about burnout as a factor in experienced Wikipedians leaving all the time. However, as far as a clear conclusion can be derived from this research, it doesn't seem that ever-increasing workload (as measured by number of patrolling actions per user) is true at all, and thus we probably disproved the stress hypothesis. Steven Walling (WMF) • talk 19:12, 3 October 2011 (UTC)[reply]

There are several possible measures of wiki stress, ever increasing workload would be a pretty good one, but you need to measure it differently. Patrolling is just one partial facet of wikimedia activity, and there are many others. I would suspect that number of hours per week in which a Wikimedian made a logged activity would be a more meaningful measure. But you could at least expand the patrolling to also count summary deletions - if there has been a shift from patrolling articles to simply deleting them then a fall in the number of patrols per patroller could be a very misleading statistic.
For the back of the queue, patrollers distance from the 30 day cutoff is a fairly obvious stress factor (though how big a factor may vary dramatically between patrollers)
Turnover is a useful metric, high turnover could be an indication of stress (or of several other things). Low turnover would be a pretty good sign that patrollers as a group were not overstressed. The burnout may not be evenly spread. Aaron did some interesting work that identified stress amongst our most active vandalfighters, It seems the pattern there is that in each year about ten very active vandalfighters will emerge and then frequently they burn out and some more step forward.
Increased snarkyness would directly increase one possible cause of stress. Are we getting more brusque and snarky in our dealings with each other and with newbies? If so, is it that we are developing a less collegial atmosphere in which the snarkier editors are preferentially fostered, or are we seeing some editors get more snappy over time? If the latter it could well be a direct sign of stress.
Many new articles are frankly disappointing, some are complete dross and a few percent are frankly unpleasant. Does that stress patrollers? Logically one would expect it to.
Dissension amongst the NPP community and lack of respect from the wider community. If you look through RFAs by active patrollers, do you see more disquiet amongst patrollers and from the wider community about CSD tagging than about say AIV reports or page protection requests?
If two or three of the above are taking place then it would not be unreasonable to conclude that NPP is an area where stress and burnout are problems. I rather suspect it is somewhat worse than that. WereSpielChequers 08:53, 4 October 2011 (UTC)[reply]

Dividing up the workload[edit]

This is not exactly on the topic, but the comments above have reminded me of a feature I'd like to see:

I think there's something to the theory that people who have reject the same thin excuses a thousand times before might not be friendly to the person who proposes it for the 1,001st time. I think that less work by "NPP specialists" and more work by subject-matter specialists might reduce this problem by dividing the workload over a much larger number of editors. I believe that if you could get a Special:NewPages list of just medicine-related articles—without the endless string of entertainment- and sports-related articles that often seem to dominate the queue—that I could easily find half a dozen editors who would keep up with just that one subject area. None of them would have to do very much (the subject might average 10–20 articles per day), so I think they would feel like they had plenty of time to check for copyvios, fix problems, and coach new editors. Also, I think they'd do better with one aspect that NPP specialists do almost none of, which is knowing whether we already have a perfectly fine article on this or that subject.

Basically, what I want is yellow links on the list at en:User:AlexNewArtBot/MedicineSearchResult. WhatamIdoing 00:59, 27 September 2011 (UTC)[reply]

Actually, there are never more than 2 - 7 patrollers at work at any one time. I fail to see how we are going to find a medical specialist among them. Besides which, as you would know, most medical articles are created and maintained by established editors, medical articles comprise very very few of the new articles that arrive at a rate of one every 15 seconds, and of the new medical articles that do arrive from new editors, very few, even if unreferenced, are likely to be deleted. The worst thing that generally happens to them is that they get merged into an existing related article. Kudpung 08:11, 4 October 2011 (UTC)[reply]
Hi WhatamIdoing. I think this is an excellent idea, though we'd need to broaden it out to all 700 or so wikiprojects as was done during the uBLP cleanup. Nobody knows how many of the tens of thousands of BLPs that were referenced during that drive were done by the individual wikiprojects, some like Spain and Heavymetal did amazing work, others were less active. But even the ones who told us they look at them all and prod the hoaxes and non-notable were making a big contribution. Producing a specialpages version for each wikiproject would depend on getting new articles categorised or tagged for Wikiprojects, but I would hope that Dashbot or similar could replicate what was done at en:User:DASHBot/Wikiprojects/Templates and produce a report for each wikiproject of articles created in the last thirty days. Or maybe we could get a routine that let people view Newpages related to a particular wikiproject. WereSpielChequers 09:11, 4 October 2011 (UTC)[reply]
Kudpung, my point is that if I had such a focused list, I could get you some new patrollers, and all of them would be "medical specialists". I agree that the medicine-related list is likely to be small (I estimate ten to twenty per day, and most of those about people, organizations, or books rather than diseases), but if it works for that subject, then it's likely to work for other subjects, too.
WSC, the advantage of using something like AlexNewArtBot's algorithm is that it's based on keywords in the article, rather than on the presence of a WikiProject banner on the talk page, so it requires no human intervention. WhatamIdoing 17:58, 4 October 2011 (UTC)[reply]
I think the idea of a link on each project's main page, to new pages relevant to that project, would be seriously awesome. Not only because it would stop me ever having to see any one-line-and-infobox unreferenced (apart from a sporting-body profile in EL section) sports-related BLP stubs ever again ...... but because pages would be patrolled by people who had a real interest in that area. That would solve a lot of challenges. ThatPeskyCommoner 05:34, 28 October 2011 (UTC)[reply]
It might work to some extent. However, the Wikipedia Schools project which is probably one of the largest in the encyclopedia in terms of pages, has nearly 400 'members' . We use AlexNewArtBot and a special ToolServer watchlist for schools but only around five members are active. Even worse is NPP, which has nearly 4,000 patrollers, but there are rarely more than around 8 on duty at any one time to cope with 1,000 - 1,500 new articles every 24 hrs. --Kudpung 06:37, 28 October 2011 (UTC)[reply]
Yes, but there are more WikiProjects than there are new articles needing review each day, so you really only need one WikiProject-based NPPer at most of them. Especially if something like this got a couple of folks at the music and BLP-related projects to pitch in, the NPP regulars might see their workload decline substantially.
If this is a good idea, how do we make it actually happen? Do we need to find someone who could create a bot to make and update lists? WhatamIdoing 16:39, 28 October 2011 (UTC)[reply]
My recent checks through Wikipedia projects (because obtaining their cooperation on page patrolling issues is going to be essential in some way or another) is that the vast majority of subject projects either have the same problems as the schools project (on which I am one of the 3 or 4 most active), or they are dormant or even dead - for example, in spite of the monumental issues with the India Education Programme that has cost the Foundation a lot of money and has vexed and exasperated the experienced Western volunteers, the WikiProject India has not proven to be particularly helpful. Plenty of bots and ToolServer tools are already available for patrollers, but the problem of getting the patrollers to use them, and do their patrolling properly is still the main issue, and the reason why we are prioritising the development of a completely new control panel for them to use, a video tutorial for them, and the eventuality of making NPP a user right. You and I had discussions about this already a year ago, where you expressed some views that may have been misunderstood; nevertheless, your recent comments are most thoughtful and relevant, and most welcome. We need to look at these ideas together more closely when the results of my NPP survey have been analysed. Kudpung 03:37, 29 October 2011 (UTC)[reply]
Yes, I know the WikiProjects are often quiet. But think about how little we'd be asking: My quick scan of all new pages created yesterday showed about five articles about schools. Do you really think that five articles is too much work for the three or four active WPSCHOOLS folks in a whole day? WhatamIdoing 16:41, 29 October 2011 (UTC)[reply]
It has little to do with what I 'think'. It's based on what I 'know' from coordinating the school project, and virtually running it alone for a long time last year. No amount of motivation and incentive such as welcome templates, barnstars, bullying, or even creating a en:WP:WPSCH/H has helped. Kudpung 17:16, 29 October 2011 (UTC)[reply]
So you "know" that WikiProjects will not be able to review articles, and I equally know that WPMED in particular would be do this easily. Do you want to find out which of us is right, or shall we keep saying "tis—tisn't" like a couple of kids on a playground? WhatamIdoing 02:14, 31 October 2011 (UTC)[reply]
I base my facts on my empirical findings through my work on the schools project, and the apparent lack of cooperation on NPP from one or two other projects. I have no idea of the level of participation of your medical project. I would assume however, that medicine being an academic subject, that the people involved are more serious about their work on Wikipedia, and are more prepared to be of assistance. And that is your empirical experience, and I'm not interested in sparring with you over it - I value your comments and I'm more concerned with collaborating with you objectively. With the proposals for making selective queues for patrollers, project involvement is going to be essential. There is a 'New article' bot that can populate a project sub-page, and there is a relatively unknown/unused ToolServer tool that provides excellent dedicated watchlists for projects. Its only downside is that it does not support the hover AJAX. --Kudpung 03:03, 31 October 2011 (UTC)[reply]
So back to my earlier question: I'm certain some projects can do this. What would it take to make it happen? WhatamIdoing 15:21, 31 October 2011 (UTC)[reply]
I've no idea. Kudpung 16:01, 31 October 2011 (UTC)[reply]
It looks like "TedderBot" has taken over for AlexNewArtBot. I'll go ask its owner if s/he has ideas. WhatamIdoing 16:09, 1 November 2011 (UTC)[reply]

┌─────────────────────────────────┘
It looks like we're fairly well on our way to success here. Fred Gandt volunteered to write a script that filters the NewArtBot listings. The listings aren't real-time, so the filtered results aren't either, but they seem to work, and not being real-time should entirely prevent the one-minute tagging problem, so that's probably a feature, not a bug.

If you'd like to try it out, the instructions for installing it are at en:WT:MED#New_page_toy, and it should work on any of the listings. You might like to look it over in the context of en:User:AlexNewArtBot/MilitarySearchResult, which currently shows 79 pages (not all of which are actually about military history; you know how keyword scanning works) that need to be patrolled. The script takes about a minute or two to process a list of that length, but it's happy to run in another tab while you do other things.

This is still basically "beta testing", and Fred would like to hear about any problems. WT:MED or his talk page seem to be effective ways to reach him. WhatamIdoing 16:40, 21 November 2011 (UTC)[reply]

Articles for creation[edit]

It's heartbreaking sometimes what sorts of contributors we lose, or repel. w:Wikipedia_talk:Articles_for_creation/Nikky_Finney was written by someone who clearly knows how to write. Of course they haven't learned yet about WP:V, and about wiki formating, but that is something more easily learned than writing well.

The "review completed" template at the bottom is anything but appreciative. Its very colour and graphical design says, "You've failed". No personal contact was initiated with the author; just two templates on her talk page [1]. The user hasn't edited since [2].

We probably could do with a welcoming committee that looks at newbie contributors and helps those who clearly have the potential to become outstanding writers find their feet, rather than leaving them demoralised or, even worse, with the sense that we are actually below their intellectual horizon.

Welcoming could include one-to-one IRC chats to explain basic policy, etc. --JN466 14:09, 21 November 2011 (UTC)[reply]

Hi, this is MetaWiki at MediaWiki where new software developments and research projects are discussed that have a Wiki-wide impact. May I suggest that you refer this issue to the people at the English Wikipedia who take care of development of the Articles for Creation process at AfC Project. The Wikipedia also has an IRC help channel for new users. Kudpung 14:22, 21 November 2011 (UTC)[reply]
Hi Kudpung, thanks, I was wondering. :) Is there something here on Meta regarding en:WP template design, or is that handled on en:WP as well? --JN466 19:04, 21 November 2011 (UTC)[reply]
In general, everything to do with en.Wiki is discussed at en.Wiki, here at MetaWiki we discuss the Foundation's projects and related projects; from coordination and documentation to planning and analysis. To discuss specific templates, you will need to contact the people at the various Wikipedia projects that use them . For more technical information related to the physical use and functions of templates, you may wish to start here. Anyone can edit a template, but the source code of many templates includes embedded php routines. If these templates are used by transclusion, a single small modification could have far reaching consequences for tens of thousands of pages, so it is generally accepted that any changes are discussed by the template's parent Wikipedia project where anyone can make suggestions and join in the discussions. In many cases, thre are sub sections of those projects that are curently discussing the issue of making templates messages and warnings more friendly. If the WikiMedia Foundation is lending technical help for the development of templates, the discussion will also take place on the talk page of the related template project. This is an example of one such project. Kudpung 00:58, 22 November 2011 (UTC)[reply]

Status[edit]

JJMC89, Jtmorgan. This research is extremely old. There have been major software and policy changes since it started. It bears any relation at all to the current profile of today's patrollers or the new articles they review. I'm surprised anyone is still watching this page and editing it - perhaps it ought to be marked as 'historical'. Kudpung (talk) 05:39, 17 April 2020 (UTC)[reply]

I was just cleaning up mw:Extension:SyntaxHighlight deprecations. It certainly is historical. I would think all completed research projects are though. — JJMC89(T·C) 06:00, 17 April 2020 (UTC)[reply]
Kudpung we don't mark research projects as historical. That's not how science works. Cheers, Jtmorgan (talk) 16:00, 17 April 2020 (UTC)[reply]