This is an archive of past discussions about Wikipedia:New pages patrol. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Going through the posts above, I am restating my opinion again: we need more reviewers.
Currently, a publicly visible message is sent out only if the reviewer has something to tell to the creator. I think a message should be delivered to the creator whenever the article is reviewed. Also, I have created {{NPR invite}} which seems to be used almost never. —usernamekiran(talk)03:44, 21 November 2017 (UTC)
Insertcleverphrasehere. I don't believe for a moment that we need more reviewers. If all those who asked for the user right (which is abolutely not difficult to get for anyone with sufficient clue) would some reviewing we would not need more. What we need is fewer hat collectors. Kudpung กุดผึ้ง (talk) 20:26, 26 November 2017 (UTC)
That being said. I think we need to request a change to the measured statistics to see who isn't using the user right at all and who is just reviewing a few pages a month. We need a full list of all of the admins/patroller user right holders and how many reviews they have made in the last year (as well as an 'all time' longlist). I personally don't have a problem with someone who only plans to review articles that they randomly come across as part of their other activities (as long as that person has the knowledge and expertise necessary to do a good job). This might only be a drop in the bucket, and might not make much of a difference, but they still deserve the user right and are not necessarily a 'hat collector' (although there are undoubtedly some of those). More reviewers overall will increase the number of those that are dedicated solely or mainly to the project (rather than it being a side thing, or a tangential thing that they might do if they randomly come across a new article. Realistically, most of the people with the user right have no interest in reviewing regularly, just like most admins have no interest in reviewing regularly (very few of the regular reviewers are admins, despite there being 1700 of them with the user right). Personally I can't exactly remember how I got the new page reviewer user-right, I don't remember applying for it, so maybe I was one of those grandfathered in, but I definitely have not been consistent with it. I am pretty sure there have been many months with no reviews at all when I was working on other stuff (GA reviews etc). Others may think that NPP is something they will do, apply, and then just not get into it for whatever and never come back. Still others may have once been very active patrollers, but given it up for some reason and never come back (resulting in their review count for the year being zero, but their all time count would be much higher). My point is that there are many reasons why there are lots of users with the NPR user-right who do not patrol many/any articles, not just hat-collecting. We need to stop thinking that there are 500 patrollers. It isn't productive at all. I will soon be requesting a change in statistics so that we can see how many active reviewers that we actually have, so that we can get a good idea of what our community of user-right holders actually looks like. — Insertcleverphrasehere(or here)06:34, 27 November 2017 (UTC)
Some feedback from a new user. I was alerted to this permissions group when a reviewer reviewed my page.
I received a (notice) message informing me that it had been reviewed by user Fyddlestix.
A message informing me of a pass/fail was conspicuously absent - I had to message the reviewer and ask if there was anything that needed doing
I found it relatively easy to become a page reviewer
Now I am one, my biggest issue is knowing what to do with obvious passes (I haven't actually made use of the permission yet - I plan to soon, and I have been looking a lot at other people's usage of the tool). My instinct in the absence of a button which leaves a default message on the author's talk page is to message them manually to let them know that I think its a pass, with no formal issues, which need tagging, perhaps suggesting some ways of expansion etc.
This would essentially address the confusion I myself felt after having had a page reviewed. I get confused very easily, so I may not be a paradigm case worth addressing
By just clicking the review button in the page curation tools and not leaving a message, it doesn't usually leave a message at their talk page at all (though it does give them a notification). If there are issues it is handy to leave a message to the user using the page curation tools. However, most "obvious passes" are done by experienced editors that have written quite a few articles, and realise that if the article is reviewed and no tags etc are placed on it, then it is ok. I don't always see a need to send a message to these sorts of users just to tell them that their submission passed review with no problems, but with interesting submissions of sufficient length, I often send a message encouraging the user to take the article to WP:DYK. If you want to message every user, that is ok too. — Insertcleverphrasehere(or here)20:22, 28 November 2017 (UTC)
Backlog 13,500
Hi all, the backlog has started rising again to 13,500. I didn't send out a newsletter this month, because there isn't really much new going on, but would everyone please consider reviewing some pages if you haven't recently (and I count myself in this number as I've been pulled in a lot of different directions recently). As I've mentioned above, I think a lot of this has to do with some a few particular users who do a lot of the heavy lifting dropping their activity a bit or leaving Wikipedia, which I certainly can't blame them for, we are all volunteers. If there are any users who you think would be good at this or enjoy it, please encourage them to apply. Any user who has been editing for at least 90 days and has a clean block log is a good candidate. Thanks everyone. TonyBallioni (talk) 02:24, 20 November 2017 (UTC)
I know that there are some regulars here that are a bit opposed to backlog drives, but I think if we are clever about how we organize it, we can ensure that quality reviewing is occurring. Can I suggest that we do a backlog drive for the Month of December and call it something hypeable such as: "The Great December New Page Backlog Drive". I propose that we set up a page dedicate to it where we can collect statistics on the # of reviewed articles for each patroller, and tiered awards for participation (awards specific to the backlog drive).
Most importantly, I suggest that on the statistics page we also have an additional column "review the reviewer" where other reviewers can put a dated check mark saying that they have reviewed that reviewer's reviews (no comments on the statistics page, any issues should be brought up to the reviewer directly). There should also be tiered awards for users who 'review the reviewers'.
I think if we can make up some good advertisements and post them around on various noticeboards, we might be able to get some experienced editors that might not otherwise participate in New Page Patrol to join us for a while and crush the backlog down (and hopefully some that will become regulars). Happy to do most of the work organizing it and make the awards if we can get some technical help about how to set up the statistics page.
I'd be fine with something like this as it seemed to work pretty well in September to get the non-autoconfirmed backlog eliminated. I'd had concerns before, but given the increase in the quality of pages and the previous good work done, I think it could work. TonyBallioni (talk) 21:50, 20 November 2017 (UTC)
I retired in February after years of coordinating NPP. I personally don't approve of drives because as clearly demonstrated at AfC several times, they lead to sloppy reviewing. We have an excellent tool that already provides an accurate overview of who is doing the most (and the least) of the work, and I would advise against checking on the the work of the the reviewers - sounds too much like a witch hunt. What does need to be done however, is keep the non accredited patrollers in check. They are still causing as much concern as they ever did without a user right - NPP is definitely not for beginners. On the whole, however, Tony is our de facto coord and understands the issues and his input to the idea would be most welcome. Certainly something needs to be done because since ACTRIAL was introduced, despite the marked drop in the number of pages to patrol, the backlog is actually growing again. Don't count on me though for doing much patrolling any time soon - I've had enough of it, really.Kudpung กุดผึ้ง (talk) 22:28, 20 November 2017 (UTC)
Reviews of the reviewers is a good idea, just as all reviewing is good, in the right (small) proportion. Clever's an additional column "review the reviewer" sounds like encouraging way too much. A process page to receive complaints, like done at DRV and MR, has proven to be a good way to work reviews. Some important points: (1) Reviews should be giving the reviewer considerable leeway to use their own judgement, and perfection is not the goal sought by the reviewer; (2) Criticism should be reserved for actual mistakes that cross a defined line, not nitpicking; (3) criticism should always be constructive, what should they do differently next time; (4) The reviewed should always resist any temptation to argue a point of the reviewer, instead, they should adopt the mindset: the reviewer is working for me, trying to offer constructive feedback, and may be mistaken. At worst, say "thanks, In understand you, I don't agree", but preferably just the first clause. If a review is harsh, refer to Meatball:DefendEachOther.
NB. While I have heard (and repeated) a fair bit of historic criticism of AfC reviewing, I've not heard the same about NPP reviewers, except a couple of very specific things by Kudpung himself. --SmokeyJoe (talk) 23:07, 20 November 2017 (UTC)
Being the devil's advocate: with AfC declined drafts stick around and are re-submitted, and accepted drafts have to be patrolled, so there's always at least one other person looking over your shoulder. A badly-reviewed new page will either have a CSD/PROD/AfD declined or be ticked and likely never checked again. So all other things being equal you're probably going to hear about more problems at AfC. – Joe (talk) 23:30, 20 November 2017 (UTC)
A couple of months ago I suggested an alternative backlog drive. Instead of encouraging editors to review lots of articles, we'd encourage them to maintain a "streak" of reviewing at least one article per day. I think that would have the same effect of upping the volume of reviews overall, while avoiding the problem of people rushing through them to rack up their score. – Joe (talk) 23:24, 20 November 2017 (UTC)
That is a good idea, I'll emphasize this as the primary goal, and will craft the Award tiers accordingly (maintained a streak of # reviews per day for X days) in addition to overall awards. The problem that I have is that the Top reviewers page does not record any but the top 50 per day, making streaks difficult to record. Is there any way of getting the Community Tech bot to output a full list of reviewers (and # of reviewed articles) each day at UTC 0:00? — Insertcleverphrasehere(or here)23:52, 20 November 2017 (UTC)
The most important data at Wikipedia:Database reports/Top new article reviewers is displayed for 100 users. What happens over he most recent tranche of 24 hours is less important - there is no kudos to be gained in being the day's best reviewer. In a ny case, as I mentioned before, I'm adverse to turning Wikipedia processes into a MMORPG - it's bad enough when admons get accused of having too much power even if they don't abuse it. What people don't realise is - because they are not admins and don't work at PERM - is that a lot of hat collecting goes on there, and I mean a lot. I'm not pointing the finger at any individuals, but research will show that of the 450 or so authorised reviewers, the number who have never used, or rarely used the tools is staggering. Admins need to be reminded that it's not only meeting the numericial threshold for the right that counts, but also demonstrating sufficient experience and convincing of a genuine need and desire to help out. Kudpung กุดผึ้ง (talk) 00:18, 21 November 2017 (UTC)
Well, I agree that caution is necessary. Backlog drives and editathons (in this case a "patrollathon") are meant to employ elements of gamification to improve user engagement. I think that this is particularly advantageous when discussing something like NPP, which has a high burnout rate. It is essential to ensure quality reviewing however, and this is where 'reviewing the reviewers' is important. What form that should take is up in the air for me, but at the very least, I would consider it essential for my own peace of mind to re-review a half-dozen or so random articles from each of the top 100 reviewers during a drive that I organised, especially for users that aren't regulars at NPP. This should help identify reviewers that need help or forget to complete reviews fully. A healthy balance of good reviewing and good productivity will be the aim here. — Insertcleverphrasehere(or here)00:57, 21 November 2017 (UTC)
The difference has to do with the tool used to review. Both Page curation and Twinkle use the Patrol log, so if you tag an article with twinkle as patrolled, or review it in it will be marked as 'Patrolled' in the patrolled log. The Page curation log keeps track of all actions made with the page curation toolset, and in that log pages that are ticked off with the page curation tools are marked as 'reviewed' (but will also show as 'patrolled' in the Patrol log). Basically, Twinkle only logs as 'patrolled', page curation logs as both 'patrolled' and as 'reviewed'. Either one should record statistics at Top reviewers. As far as I can see, your reviewing levels would put you outside of the top 100 year/3month/month/week rankings most of the time. The reason you arent on the top 50 for today is because the list is only updated every 12 hours, and your two reviews today were both after the last update. You should show up there in a couple hours. — Insertcleverphrasehere(or here)01:24, 21 November 2017 (UTC)
Thanks. So I can treat them as synonyms? I am not an enthusiastic NPP reviewer hoping to win awards, I like the goal someone suggested, maybe it was DGG, of aiming to do a couple most days. I do find it much more rewarding that spending time at AfD, which I came to find depressing. I find the page curation tools make the job reasonably easy to do, and that I basically can't do it on a mobile device. --SmokeyJoe (talk) 01:47, 21 November 2017 (UTC)
Sorry to be a worry wort, but I'm not sure it will be a good idea/effective for a few reasons: (lameness alert)
-Some of us are currently delayed due to participating in other things, WOMRED 60's November contest e.g., and may wish to take a break, also
-December is a month of holidays and I assume a large amount of patrollers* will either be taking wiki-breaks or won't be available as much due to family, feasting, fooling what have you
- a month is an awful long time keep at something. I'm not a learned psychologist and maybe I overdo Wikipedia, but i would feel much better about a drop-everything-and-patrol if it were only a couple weeks
Asterisk in point dos: if there were a couple thousand of us patrollers we might be better able to "survive" but as there are only 300 active patrollers we could play hit or miss
if there were a drive I would participate. I like NPP (even though there aren't any keyboard shortcuts ;-) and am trying to do it at least every other day. But new year new drive might possible be a better idea. I know we are running up against the end of ACTRiAL, but I have a reasonable expectation that someone will forget to remove the specs and it will become the new normal. As for awards, none were handed out in July so I never expect to see any recognition for holding at the top of lists. Again, I may be getting on about nothing in regards to the above 3 points, but its just something to think about. Thanks, L3X1(distænt write)03:06, 21 November 2017 (UTC)
There is a clear distinction between patrolling and reviewing. Patrollers are anyone and his dog, including absolute newbies whom the community insisted should still be allowed to tag new pages and bite new editors. This is the group that causes most problems. Reviewing is done by Reviewers who have had the right accorded to use the Page Curation tool and mark pages as patrolled thus releasing them into the encylopedia and free to be indexed by Google. 'Patrollers' can't do this, and their tags still have to be controlled and reviewed by proper reviewers. The end effect is that while we now have somewhat better quality control over what gets passed as an appropriate new page, it hasn't really reduced the work load for the real reviewers, out of the 450+ of whom, only about 10 - 15 are actually doing it. And the backlog is rising again.Kudpung กุดผึ้ง (talk) 06:16, 22 November 2017 (UTC)
@Kudpung: I've let things pass previously, but as someone who is not one of the 10-15 real reviewers (and props to those who are), I have no intention of reviewing many hundreds of pages a month, and don't appreciate feeling guilted into doing more than I'm doing. There's not a lot of fun at the back of the queue. Agree with comments elsewhere that what we need is (many) more reviewers. 120 reviewers doing 50 each a month would be far less fragile than another 6 doing 1000 each a month. (General call for volunteers at WP:VPM and WP:AN?) ~Hydronium~Hydroxide~(Talk)~07:16, 22 November 2017 (UTC)
Hydronium Hydroxide, I'm just throwing in my 2P. I retired from years of NPPing. I don't want to know about it any more, except to say that I'm very disappointed in the result of the efforts I went to over the past 7 years - 100s of hours of coord and coercion, and 1000s of patrols I've done myself. It all feels like a wasted effort. No one needs to feel guilty unless they are one of the hat collectors. They are generally user names I've never heard of. They know who they are.Kudpung กุดผึ้ง (talk) 09:43, 22 November 2017 (UTC)
Hydronium Hydroxide, I agree that we need a lot more reviewers, and I would say anyone reviewing even a page a week on average would be a 'real' reviewer. It would take a lot of pressure off us all. General calls for volunteers at certain pages may well work, as those pages are likely to be watchlisted by people who already have the skills we're looking for. We may also want to consider more of the method that led me to become a reviewer - I was simply given the right unasked, and it was then up to me whether I used it or not. For a long time, I didn't, until the newsletters eventually persuaded me to check out what NPP actually was. Asking people to nominate themselves and have their voluntary work picked over will put off a lot; what about giving the right to people who have earnt it, editing well over time? Boleyn (talk) 11:06, 26 November 2017 (UTC)
Just pointing out that I did just send an advert out over at WP:VPM and at WP:AN. Also, I think that this is a good idea if any admins are up for it (give the right to users that havent requested it, but that they know have the skills. In the meantime I'll keep inviting peeps. Looking like around a 20-25% invite success rate (to successfully given reviewer user right). — Insertcleverphrasehere(or here)18:23, 26 November 2017 (UTC)
Boleyn: Agree having to request it is a barrier and that where clearly appropriate it should be handed out with only the hope that it will get used. I would have done some reviewing back in December if I didn't have to actually ask for the right to do so. Insertcleverphrasehere: thanks too; looks like your efforts are bearing fruit. ~Hydronium~Hydroxide~(Talk)~12:38, 28 November 2017 (UTC)
Kudpung, actually it was, which is why my RFR is so ambivalently worded, and why it took me until January to request the bit. My expectation at the time was that I might do 20-30 on average a month. Posts complaining about reviewer output would have been incentive to do something else. ~Hydronium~Hydroxide~(Talk)~12:51, 29 November 2017 (UTC)
That's a lot of people - roughly half the average number who attend Wikimania each year. 503 is a lot of New Page Reviewers. Obviously they've not all been around since 5 November last year. The disappointing thing however, is that despite this large number of reviewers (which is more than the number of truly active admins), and despite significantly few new pages to review since ACTRIAL, the backlog is again seriously on the increase. If during the next 12 months if they would all do just 10 reviews a week there would be no backlog. Perhaps the coordinators could highlight these issues in their next newsletter. Kudpung กุดผึ้ง (talk) 20:18, 26 November 2017 (UTC)
See my reply in the section above. We don't have 503 reviewers. Using your criteria of 10 per week as a baseline, have about 90 over a year, 70 fully active reviewers over three months, 55 over a month, and 60 for the last week. I think it is fair to say that there is barely more than 100 active reviewers at any given time, and the trail off from there is people that are busy with other stuff or people that are only dabblers (and a few hat-collectors no doubt). We would know more about the distribution of reviewers if we had lists that were longer than 50/100, but it is fair to say that there are about 5 times fewer reviewers than everyone keeps claiming. At best there are 100 reviewers at any given time and by that rationale we need to be reviewing 50 pages per week per active reviewer to keep up, and currently the difference is made up from very dedicated users that do way more than their fair share. Propagating this myth of "10 reviews a week" just isn't going to help much. Realistically the distribution of editors will always be a decreasing exponential distribution with a long tail. By increasing the total number of users we can increase the area under the curve of the distribution. We can do this by inviting promising users with the invitation template ({{subst:NPR invite}}) whenever possible. Not to say that encouraging reviewers to review more is a bad thing, but I think this logic of "if we all did X, then Y" isn't productive because it is never going to happen and most reviewer right holders are not active at any given time.
Inviting people is a solution though: For example, I spent about 4 hours inviting users to NPP last week (invited about 100-ish people, with about 20 successful applicants, and 10 have begun reviewing). I have totalled up the count of reviews that have been done by the users that I invited last week (the ones that have requested the user right and have begun reviewing), and it is about 175 over the last week, more than 4 and a half times my own average weekly review count over the last year! (my average is ~38 per week) These users will continue to 'pay dividends' over the coming weeks and months, so realistically, with 4 hours work I have managed to more than quintuple my effective review count (my impact to New Page Patrol). If we all spent a little less time reviewing, and a little more time inviting other promising editors to NPP, the whole project would be MUCH better off (a bit counter-intuitive, but true). — Insertcleverphrasehere(or here)07:13, 27 November 2017 (UTC)
I agree that 503 sounds great, but doesn't indicate how many editors we have who are regularly part of NPP or not busy on other projects at this time, but may become more active at some point. The 503 we have doing starting to do more reviews would be great, but isn't going to happen and I don't want editors to feel that their occasional contributions to the project aren't valued - they are and every review helps. We also need to make sure we're coming across as a positive and nice project to get involved with and not leave volunteers feeling beholden. We can't make the people we have give more of their time to this, so getting more reviewers is the only option I can see working. Boleyn (talk) 07:29, 27 November 2017 (UTC)
I am going to add a note about the Invite Template to the next newsletter, so that people know that it exists and that using it is our best hope for a manageable NPP. Several newsletters in the past year have stressed "If we all did X then Y" and while we have had success in reducing the backlog, I think that you are right that we shouldn't be implying that patrollers are being lazy for only reviewing a few articles. Every review helps and we just need more patrollers overall and NPP would become very manageable all on its own without the need to squeeze the few people that we currently have. — Insertcleverphrasehere(or here)07:36, 27 November 2017 (UTC)
We do have 503 people who have the New Page Reviewer right. The reduction in the backlog over the past 12 months is not a success, and it's actually fast on the rise again. It won't be long it will have reached the staggering number it was at last year. What I can't figure out is why that is happening with so fewer pages to patrol since ACTRIAL was rolled out. Kudpung กุดผึ้ง (talk) 11:28, 27 November 2017 (UTC)
It is a partial success. The backlog is bouncing up and down a bit at the moment (it was down as low as 12900, but then as high as 13600 last week, currently it is around 13250); I wouldn't call that "fast on the rise". There are not actually that many fewer pages to review than before ACTRIAL, and realistically, the pages from non-autoconfirmed users were often the easiest to review (they were often rubbish and could be immediately CSD/BLPPRODed or quickly tagged or AfDed. What is happening is that we went through the backlog and scoured out all the non-autoconfirmed submissions, and what is left are articles that are longer, require reviewers to check quasi-bad sources, annoying POV or mildly promotional articles, etc (i.e. they take a bit more time). I think that the NPP browser is also partly to blame, as I think that some of the 'easy' reviews have been successfully targeted during the period when the backlog was shrinking fast (i.e. short articles on sportspeople). All of this adds up to the bottom of our backlog being a bit more concentrated with the so-called 'time consuming judgement calls' which just ends up slowing the reviewing process enough to halt our progress in reducing the backlog.
All in all though, it isn't the end of the world, even if that means that we are treading water at the moment. I hope that some fresh hands are what we really need. Realistically if we have 500 reviewers now, and only 100 are active, then a couple hundred more reviewers should be able to get us at least 40 or so 'actually active reviewers'. This should tip the balance in our favour in the war against the backlog. I've already managed 20 more reviewers, and I sent out invites to 60 or so people on the WP:List of administrator hopefuls today, so I expect some more applications will be forthcoming. — Insertcleverphrasehere(or here)12:00, 27 November 2017 (UTC)
@L3X1. There is a confusing thing between the 'pending changes reviewer' user-right (which is called 'reviewer') and 'new page reviewer' user-right, (which is called 'patroller'). There are 4 instances of that on the page you linked, (two are self requested, but two others were indef blocked for sockpuppetry). Looks like we have at least two to take off the list ( Done). I suspect that we will be back up in no time though, because PERM is also a bit backlogged with requests at the moment; it feels strange to be legitimately proud of singlehandedly creating a backlog. — Insertcleverphrasehere(or here)17:24, 27 November 2017 (UTC)
You should be, recruiting Sgt, I swung by perm the other day and was a bit shocked as well to see it full. I'm going to go ask Jo-jo Emerus if they are able to do something about the patrol/review NPP/PCR mixup in their popups. L3X1(distænt write)17:35, 27 November 2017 (UTC)
The NPP browser has eliminated all the "easy ones", what remains now are the difficult ones, thats why the backlog is at current state. —usernamekiran(talk)22:42, 17 December 2017 (UTC)
Curation tool filters
A requested Curation tool filter has now been deployed ([1]). Please check for bugs. Kudpung กุดผึ้ง (talk)
EDIT: UH... I guess it wasn't finished updating? There are currently 2398 articles created by 'learners', we should target these as a priority when reviewing from the new page feed. — Insertcleverphrasehere(or here)21:41, 10 December 2017 (UTC)
Backlog
I expect everyone has received the December NPR newsletter. Eleven years ago Jimbo Wales suggested Wikipedia contributors should shift their focus from the number of articles and instead work on improving their quality.
By preventing through ACTRIAL the possibility for first-time users to create articles directly in mainspace, we have successfully reduced the workload at the New Pages Feed by about 80%. But that does not mean we can retire. For some unexplained reason the backlog has not reduced one iota, unreviewed articles - mainly junk - is being released to Google after remaining unpatrolled for 90 days. And we are now faced with a new challenge: detecting an increase in artspam and other paid editing by users who do their 10 edits and patiently wait 4 days.
In a volunteer collaborative environment, participants can make their own minds up as to where they want to work and the extent of the work they do. One of the biggest challenges to effective New Page Reviewing is incentive and reward, but it’s not like Today’s Article for Improvement or Collaboration of the Week or the WikiCup. Those are all informal projects. NPP is an essential, but dreary core function that simply has to be done by someone, and all the time 24/7. Kudpung กุดผึ้ง (talk) 23:41, 12 December 2017 (UTC)
Looking at the graphs showing page creation stats:
ACTRIAL has reduced the number of articles needing review from about 600 per day (total less autopatrolled), to about 350 (only about 40% reduction). This reduction is significant, but not as much as you pointed out above, especially because the 'articles' by new users were typically the least time consuming ones to deal with. Perhaps we also lost some reviewer workforce over the last month for not having sent out the newsletter last month; difficult to be sure but we might get some info on this given that we just sent out the December newsletter. As a side note, over the last few weeks we have had over 60 new applicants, mostly from invitations that I have made, who have reviewed, together, some 1400+ articles (or redirects) and these new editors have ramped up to reviewing 800+ per week (about 20-30% of the total weekly workload!). Despite this, the backlog has only remained stable at a little less than 13K. :(
I think we might be dealing with a psychological issue that our top reviewers feel compelled to review when the backlog is raising, and relax a bit once it starts to go down again (I have personally experienced this effect on myself while reviewing over the last few months). This results in the backlog raising a bit, then getting crushed back down by a few people working really hard, who then relax and it starts raising again). We also had a bit of an air of celebration right after ACTRIAL that might have resulted in increased reviewing activity, which has now abated. When working with so many editors, small cumulative psychological effects can have pretty big overall effects on the raise/fall of the backlog.
With the New Year backlog drive planned for January, I hope that we can get a positive atmosphere of excitement at watching the backlog be reduced by our combined efforts, which will be encouraging to further work chipping away at the backlog, rather than the current desperation by top reviewers to keep it stable. — Insertcleverphrasehere(or here)23:47, 12 December 2017 (UTC)
I can't find anywhere that has a reliable dataset for how large the NPP backlog was at a given date, or a way to check any date except what it is right now (perhaps one exists that I am unaware of), but using the past newsletters, previous plots, time-stamped messageboard and talk page comments, as well as some data of my own, I have built the following plot showing the NPP backlog. Based on this comment, the backlog was last at a manageable level sometime around June 2016. The backlog steadily raised until end of 2016 where it was held steady for a while but ultimately rose again in mid 2017. There seems to have been a big push after the 22,000 Newsletter was sent out, after that, it dropped quite a bit. There is a steep dip immediately after ACTRIAL when we cleared out all the non-autopatrolled submissions, but since then it has been relatively flat. — Insertcleverphrasehere(or here)01:26, 13 December 2017 (UTC)
What is really needed more so than a number is to cut the backlog down to the 90 day no index period. Currently we're about 4 months over that, give or take. If there were constantly 100,000 unreviewed pages and it stayed stable, it wouldn't matter much so long as we reviewed them within 90 days. The number is a good metric currently because we are far exceeding the 90 day period, but if at some point we got to a stable number but less than a 90 day backlog, we'd be fine. TonyBallioni (talk) 01:42, 13 December 2017 (UTC)
Yep, I agree with you there, the number of articles past the index point should be the metric by which we measure the backlog (ideally), which is currently 7802 articles. Or else the number of months past the index point. Realistically though, there really isn't any good reason for the backlog to last more than a couple of weeks (once we get it down to a reasonable number and have enough reviewers, it should be small enough that we can keep it very, very short with just upkeep). At that point I would recommend that we reduce the index point back to 30 days so that it keeps us on our toes and doesn't allow too much of a backlog to develop. — Insertcleverphrasehere(or here)01:55, 13 December 2017 (UTC)
In the 5 months from July - Dec last year, it more than tripled
In November when NPR was introduced it stabilised and slowly began to drop
What suddenly happened in mid Feb and caused a 50% increase in just 9 weeks?
Certainly not an equivalent growth in new articles.
IMO the backlog should be at the level it was in April last year. The dramatic rise has defied all logical explanation to this day. Landmarks on the range of mountains are December during the enthusiasm of the roll out of the New Page Reviewer group where it flattened out and started to drop, then, by some strange coincidence rose sharply from the day I announced my retirement from NPP. There is no reason why a 5,000 backlog isn't achievable (and it's still too much), and it shouldn't take another 18 months. Kudpung กุดผึ้ง (talk) 07:28, 13 December 2017 (UTC)
To be honest, I do not think we need more reviewers at this stage. What we need is to work with the authors. If there are users who systematically write articles which comply with the requirements, they should be given autopatrolled. If there are users who on a regular basis produce substandard articles they should be worked with and explained what the problems are, so that eventually they could create articles with satisfy the standards.--Ymblanter (talk) 08:46, 13 December 2017 (UTC)
I do think more reviewers, even if they are not very active, can help - e.g 100 more reviewers that do on average 2 reviews a week is 200 reviews or 8%ish of the total. Insertcleverphrase was making a good point above about having a long tail of low reviewing patrollers can also help keep down the backlog. Galobtter (pingó mió) 08:57, 13 December 2017 (UTC)
(edit conflict)It is actually very simple to explain Kudpung กุดผึ้ง. We rely way too much on the top reviewers. There were 222902 main space reviews in the last 365 days (based on the data in this query on quarry). I made the following graph to group editors by percentile groupings, which shows quite clearly that our top reviewers review a HUGE percentage of our articles (NOTE that I haven't even included the bottom ten percent of editors as the tail would be WAY too long). Just 7 editors represent 30% of all reviews done in the past year. When one of these editors retires, the backlog can start growing at an enormous rate.
The only solution to this is more reviewers. I have been slowly working my way down the WP:List of Wikipedians by number of edits, and have invited everyone in the top 8001000 that was not already an admin or a new page reviewer (after a bit of a background check of course, I actually only invited about 350 users). I'll get back to it sometime soon and work my way down the list some more, but realistically, I don't see myself getting all the way down it by myself. Even if we only consider editors with >10,000 edits, there are potentially 8,000 editors to consider here (most will be inactive or ineligible for some other reason, but many will be good candidates). Of the ones I have invited, the candidates that chose to apply and were granted the user-right can be found on this page, where I have been keeping a review count to see how much of an impact I am making via these invitations.
The results are staggering, after spending a couple hours a day for about a week inviting people, about 60 have been accepted for the user-right, and they are now together reviewing 800+ articles a week, nearly half my yearly average.
Consider the number of pending changes reviewers, over 6000! Pending changes don't come in any quicker than new articles, in fact, I think the rate might be slower, and the amount of time taken to review a pending change is usually much less than reviewing a new article. However, we have many fewer hands on deck to review new articles, and it shows. We only are treading water because we have a few very dedicated individuals that are willing to hold the tide at bay with brute force. We might be able to hold the narrow pass in the short term, we might even be able to beat the backlog into submission for a while, but if we want to win the war, we are going to need an army.
If anyone else wants to help out by sending out invitations, let me know on my talk page so that we can coordinate which blocks of editors from the top Wikipedians list we are each going to hit so that we don't start doubling up. If you know anyone personally that would be a good reviewer drop {{subst:NPR invite}} on their talk page, or write a personalised invitation. In my experience, the success rate has been about constant at ~14%. — Insertcleverphrasehere(or here)10:03, 13 December 2017 (UTC)
I assume the amount will drop after a while (i.e not be consistently 800 from that group), but still seems to have already given multifold benefit than the time you've spent on it. Spreading the workload would mean less burnout. Galobtter (pingó mió) 11:10, 13 December 2017 (UTC)
I'm going to keep track of it in any case to see what happens. The users in my list have pretty much the exact same distribution as the full reviewer list above, so even though I was targeting the most active wikipedians of all time, an exponential distribution is what you get no matter what (one reviewer is responsible for a third of the reviews for this week). If we could get another copy of those editors from the 30th to 60th percentile though, the top reviewers wouldn't have anything left to do. — Insertcleverphrasehere(or here)11:27, 13 December 2017 (UTC)
Your recruitment efforts are definitely a huge help, Insertcleverphrasehere. I think the thing with backlogs is that they create a sort of burnout feedback loop: seeing that number stay the same despite your effort—or worse yet keep climbing—saps most people's motivation to keep going. But seeing it go down (as it did just after ACTRIAL was brought in) inspires people to do more. So hopefully the extra recruitment will push us over that tipping point into a downward trend. And if our wildest dreams come true and it ever reaches zero, it will probably be much easier to keep it there than it is to stay on top of this backlog.
On that note we should also remember the progress we have made. As the graphs above show, the backlog is now at the lowest it has been in a year, and it has nearly halved since its peak six months ago. Reforms like the NPR right, ACTRIAL, and the newsletter, were never going to be magic bullets, but they are helping. – Joe (talk) 11:30, 13 December 2017 (UTC)
Insertcleverphrasehere, I am not in any way criticising your efforts to recruit more reviewers. Your graph is excellent, but you didn't answers my questions - because there appears to be no explanations for the strange shape of my graph anyway which is partly reflected in your own curve. I will however mention that it should be possible to reduce the backlog to almost zero. By working hard, I and three other users managed to do it several years ago from a similar backlog in less than a couple weeks - without any compromise on the quality of patroling. If a concentrated drive achieves anything substantial, the effort should be to maintain the backlog at its new level and to avoid it growing again. Kudpung กุดผึ้ง (talk) 05:55, 17 December 2017 (UTC)
The recruitment should help, and I agree with Kudpung about motivation - I find I am most motivated when it is going down, or up, but I find it being static really demoralising.
I don't know how many I review, but I spend half my NPP time tracking my list of editors I've contacted about persisting creating unreferenced articles, trying to get them to stop this and trying to get them to communicate. This isn't nice work, and I have taken several editors (after many, many attempts to communicate) to ANI. If when someone saves a new article, it flashed up with a message saying it didn't have references and so wouldn't save - a bit like when you fill in online forms and leave out essential information like your phone number - then I would be able to review twice the amount I do. Boleyn (talk) 07:22, 17 December 2017 (UTC)
What Boleyn is trying to say is: "Do not let the hamsters save the new article if it doesnt have a source". —usernamekiran(talk)08:32, 17 December 2017 (UTC)
Hamsters? :) Yes, essentially that's what I feel. No idea how to put a proposal like that forward though. Any ideas? Boleyn (talk) 09:13, 17 December 2017 (UTC)
Yes, hamsters. Why do you think servers are so big? They run because hamsters run inside it, that's why they need to be so big. Also, there should be exceptions of redirects, and disamb pages; or else the hamsters would get all worked up. —usernamekiran(talk)22:21, 17 December 2017 (UTC)
Unreviewed
I'm seeing pages listed as unreviewed that have been already. I used to see a button that would let me mark a page as "reviewed", but no longer. -- Dlohcierekim (talk) 04:24, 13 December 2017 (UTC)
Which pages specifically? Are you using twinkle or the page curation tools to mark a page as reviewed? Pages can be marked as 'unreviewed' using the page curation tools, but can also be will be marked as unreviewed automatically if they are converted to a redirect and back again. — Insertcleverphrasehere(or here)04:35, 13 December 2017 (UTC)
When I look at the Special:NewPagesFeed from the back and set it to show only unreviewed pages, I see a list of pages marked with an exclamation point. Some of these pages have clearly been reviewed in some fashion. Used to be there was a button to push saying I'd reviewed a page. In twinkle, I do not see way to mark a page as reviewed. -- Dlohcierekim (talk) 04:43, 13 December 2017 (UTC)
Those pages are just the back of the backlog. Most of the articles in the backlog have been at least partially reviewed, tagged by non-reviewers, or had some fixes applied by the time they get to the back of the backlog. Being unreviewed just means that nobody with the New Page Reviewer user-right (or admin) has felt confident enough to fully review the article and tick it off. With the page curation tools you should be able to just jam the big check mark, with twinkle there should be a tiny popup near the bottom of the page on the right that says "mark this page as patrolled", both should work. — Insertcleverphrasehere(or here)04:56, 13 December 2017 (UTC)
When I attempt to add a tag about notability with the page curation tool, the process fails and I get a message "An error occurred while marking the page as reviewed: Invalid CSRF token." Any comments? Cwmhiraeth (talk) 13:46, 17 December 2017 (UTC)
Thanks, I think L3X1 had the answer. I had left the computer on while I thought about sending the article to AfD, and came back to it later. I have added a notability tag to another article since then without problem. Cwmhiraeth (talk) 06:04, 18 December 2017 (UTC)
The right dog can be an adjunct to editing, but at what cost
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
I was about to get a St Bernard, but I didnt cuz I am very busy in real life now. St Bernards need a lot of emotional activity. Ironically, my workload is expected to decrease from mid January. Once that happens, I will be spending more time on wikipedia. That is, at least 10 pages per day. What breed do you have Dlohcierekim? I am thinking to get a lab, they require less time than St Bernard. But I really want a St Bernard :( —usernamekiran(talk)00:43, 21 December 2017 (UTC)
@Usernamekiran: a very happy emotionally needy pit bull. She began licking me before I brought her home. She has not stopped. They said she was moderate energy. If so, I'd hate to see "high". They did mention the tendency to pull on her leash. She does prevent me from becoming glued to my seat in front of my computer. So, among the many positives, I don't become overly fatigued/shaky/tremorous/contracted. -- Dlohcierekim (talk) 01:18, 21 December 2017 (UTC)
@Dlohcierekim:Libra looks happy in the photo. :) I like friendly, non aggressive large dogs. A st bernard is perfect fit for my choice. I would also love an Irish woulfhound, but I couldnt even find any breeder, or owner having that breed. It is a tad bit more aggressive for me though. If I cant decide which to choose, then I will go with Indian pariah dog. We had many dogs (one-by-one). Lab was a very "full of excitement" experience, pomeranian was a "short tempered" experince, and all the indian pariah dogs i ever had were extremely loyal, and friendly. But they are the same size of lab. We should stop our off the topic chat or BobTheRobot will block us with edit summary "bob the robot... can we do it? Yes we can!" —usernamekiran(talk)01:52, 21 December 2017 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
RfC at Village Pump (proposals) that concerns this project
There is a tip at the top of this page that states "When you see a page that appears to be obviously a commissioned work, take a moment to check the history." So, the article I am looking at is CIK Telecom. It is obviously commissioned and has been in existence before, but I only know this because I added WikiProjects to the talk page and saw a notice there. The history of the article gives no indication that a previous version was deleted, and I don't know why it was deleted, nor how many times it has been deleted. It may have been spammy before, but the present incarnation is quite modest. Cwmhiraeth (talk) 09:20, 24 December 2017 (UTC)
Deleted:
09:17, April 1, 2016 Seraphimblade deleted page CIK Telecom (G11: Unambiguous advertising or promotion)
03:56, March 25, 2016 Anthony Appleyard deleted page CIK Telecom (Multiple reasons: speedy deletion criteria G5, G11)
I also came many times across the fact that deletions and undeletions do not show up in the editing history (whereas protections do). Does anybody know about previous discussions of the issue, possibly a Phabricator ticket open?--Ymblanter (talk) 11:14, 25 December 2017 (UTC)
I believe we had already asked the WMF to include a code letter in the entries on the New Pages Feto denote articles that are a recreation. Technically it should be doable - but so are all the other 50 or so requests at Wikipedia:Page Curation/Suggested improvements. Over a year ago Noyster created the page which I largely populated, but todate, very little has been addressed. Another (or parallel) solution would be to automatically have a banner put on the talk page of new creations announcing that the article was previously deleted, and an appropriate category automatically populated: 'Previously deleted articles'. It might not help detecting pages that are recreated under a slightly different name in defiance of the deletion(s), but it would be a start in the right direction. Kudpung กุดผึ้ง (talk) 06:11, 26 December 2017 (UTC)
Hi everyone, I just wanted to note that we've almost decreased the backlog by 1000 pages in less than a week. Very exciting and shows that with a bit of effort from everyone, we can clear this thing while ACTRIAL is ongoing. Just my standard request to everyone to remember quality of reviews matter (and to always check for copyright), but I did want to draw attention to the accomplishment. Great job, everyone. We're now 10,000 down from the peak. TonyBallioni (talk) 23:28, 17 December 2017 (UTC)
Just want to say... with the backlog reducing, now is not the time to take a break. Lets all do a few more reviews each to reduce it further while we have some momentum! — Insertcleverphrasehere(or here)02:25, 18 December 2017 (UTC)
I think your inviting of reviewers is helping. Adding a few hundred more reviews a week is enough to make it start going down. Hopefully with the backlog reducing people are encouraged to help get it down to a few thousand :) Galobtter (pingó mió) 04:45, 18 December 2017 (UTC)
Typically we see a reduction around the newsletter and it plateaus out after that until the next one is sent. Call it a reminder/call to arms. Inviting more people definitely helped because there are more people getting the newsletter. We skipped November because there was nothing really new to tell people, and I certainly don't want to send out news for the sake of sending it out: that annoys people and causes them to opt out of the mailing. Anyway, very good news indeed. :) TonyBallioni (talk) 04:48, 18 December 2017 (UTC)
Regarding the regularity of newsletters, i totally agree with Tony. "Meh, they print anything there just for the sake of printing and making it monthly". With the other method, "oh. The newsletter is here, there must be something interesting in it. But the new recruites have proven very helpfull as well. If it keeps on increasing, we will very easily be able to hold backlog at 5000. Very easily. —usernamekiran(talk)22:30, 18 December 2017 (UTC)
Personally I would be aiming for a backlog of 1000 or less (three days or so worth of articles). Something that would be feasible for a single person to review in a day or two (if they were really, really motivated). That way, it feels like your contributions actually make a difference, instead of feeling like you are chipping away at a mountain. Note: once we work through the article backlog, there are another 10000 or so redirects to deal with (a few very dedicated users have managed to keep the redirect pile from blowing out of proportion, and though we don't talk about it much, those editors are greatly appreciated. — Insertcleverphrasehere(or here)20:33, 19 December 2017 (UTC)
Thats where Joe's idea comes in. I didnt like the "push" events much. But maintaining a streak of a nominal number would be very good. Assuming, there are 200 reviewers who are not much active. If they accept a "plan" of reviewing only 5 pages per day, for 30 days. It would result in 150 pages a month by one user; and 30000 pages by 200 reviewers. It will also effect positively on the quality of their reviewing. Once a batch is done with 30 day program, we can take in new batch of editors. I think we should have something similar to "AIV" programs. I dont know much about it. But they have "training units" and what not. I am thinking to go there undercover to learn about there methods. Once we have that, NPP/R will also be attractive to the newcomers. —usernamekiran(talk)03:06, 20 December 2017 (UTC)
I am planning weekly streaks rather than daily, due to the fact that some users are only active during the week, or only during the weekend (with tiers running from 10 per week up to 500 per week). There will also be 'total' awards, as a separate list of tiers. — Insertcleverphrasehere(or here)05:50, 20 December 2017 (UTC)
Clearly New Page Reviewers have no life or families. See the graph at the top of this page. We've seen a reduction of the backlog of about 2,500 in the last few weeks (down to nearly 11,000, which is pretty cool, though I am starting to get a bit worried that all the hard work I have put into planning for the January backlog drive will be wasted (I made special awards and everything). We might not have much of a backlog by the time the holidays end! What a shame that would be...(sarcastic drippings) — Insertcleverphrasehere(or here)21:04, 20 December 2017 (UTC)
It will soon be a 4 character number. And hopefully, the circumstances wiil soon come where it doesnt become a 5 character number again. —usernamekiran(talk)21:48, 21 December 2017 (UTC)
This thread was started six days ago, when the backlog was 11,800. It is now 10,798, more than a thousand less in 6 days! Well done, everyone :) Boleyn (talk) 13:30, 23 December 2017 (UTC)
Yup. But that wouldn't be stable for a few days I think. I mean, it would keep going up n down 10k by a few numbers. But after a few days, the "less than 10k" will become stable. From mid January I will be free from my work, so I will be able to review around 25 pages per day. —usernamekiran(talk)17:01, 27 December 2017 (UTC)
While reviewing HDF 4-473.0 and adding tags I tried to leave a comment on the creator's talk page using the review interface. The tl subst wound up munged.[2] Is this a known bug? --mikeutalk23:32, 30 December 2017 (UTC)
Done). I suspect that we will be back up in no time though, because PERM is also a bit backlogged with requests at the moment; it feels strange to be legitimately proud of singlehandedly creating a backlog. — Insertcleverphrasehere (or here) 17:24, 27 November 2017 (UTC)
