Wikipedia talk:The future of NPP and AfC/Archive 7
![]() | This is an archive of past discussions. 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. |
Archive 1 | ← | Archive 5 | Archive 6 | Archive 7 |
Post ACTRIAL - next steps
- The conclusion to the report strongly reiterates our long-time on going requirements for the NPP and AfC processes to be improved. Within minutes of the trial being switched off, the feed was swamped with inappropriate creations and users are being blocked already. This is now the moment to continue to collaborate with DannyH (WMF) and his team to bring the entire Curation system up to date by making a firm commitment to addressing the list of requirements to the excellent suite of tools the WMF developed for Curation. The conclusions also make some strong recommendations for AfC. One place to discuss these issues initially is here where those interested in helping actively can also elect to join the task force. Kudpung กุดผึ้ง (talk) 18:03, 15 March 2018 (UTC)
- It looks as if the trial absolutely confirmed what you all have been saying for years: Wikipedia is better off if brand-new users are not writing articles as their first edit here. The brand-new editors themselves are also better off, since it is extremely frustrating to have their first attempts at editing unceremoniously rejected. The system worked EXACTLY as hoped for, preventing many inappropriate articles, saving untold hours of volunteer time, and avoiding burnout by the NPP volunteers since they don't have to spend so much time dealing with hopeless articles. The results are so clear and so positive that I hope the WMF will allow the system to be immediately switched back on while discussion about improvement goes on. --MelanieN (talk) 19:24, 15 March 2018 (UTC)
- It may have given NPP some relief, but only by shifting the backlog to AFC. It seems to have had little overall effect on reducing the GFOO. Roger (Dodger67) (talk) 19:36, 15 March 2018 (UTC)
- Dodger67, as I noted above and at VPP, while there has been a shift to AfC, the amount of work actually has gone down significantly. Article space deletions went down 227 a day during ACTRIAL, and deletions in draft space did not go up to match it (even in areas where CSD criteria exist, the rise was not proportionate to the decrease). Additionally, we cut down the NPP backlog by thousands while the AfC backlog has risen but remained relatively stable (and non-indexed by Google). The other thing is that even with the gross rise in AfC, the very old number of drafts is low, and for the first time since I’ve been around we’ve cleared the NPP backlog to be before the index point (90 days). There has been some shift in work, yes, but on the whole, things are much more manageable. TonyBallioni (talk) 19:58, 15 March 2018 (UTC)
- I've speedy deleted at least 25 articles today, a figure I cannot remember doing for some time - certainly not since we started the trial. Moreso, although this is not easily provable, the ease in which I can determine A7s has changed dramatically. During the trial, I'd see about 2-3 a day and think "well it might be A7, not sure, best leave it" - a lot of the ones coming in now are very obvious : things like Daniel Van Schie whose text was, in its entirety, "Daniel is vewy good at inglish and luvs to play vidio gayms". I've half a mind to undelete that just so I can redelete it as G3 as well as A7, as it sounds like somebody having a giraffe. Ritchie333 (talk) (cont) 19:37, 15 March 2018 (UTC)
- It may have given NPP some relief, but only by shifting the backlog to AFC. It seems to have had little overall effect on reducing the GFOO. Roger (Dodger67) (talk) 19:36, 15 March 2018 (UTC)
- Very well articulated. Be sure to check out the talk page and comment [1] Legacypac (talk) 19:30, 15 March 2018 (UTC)
- I would like to know what genius thought it was good idea to only be able to select one or all filters in NPP.
- Have no categories
- Are orphaned
- Were created by newcomers (non-autoconfirmed users)
- Were created by learners (newly autoconfirmed users)
- Were created by blocked users
- Were created by bots
- Were created by
- Show all
- Logic fail.- MrX 🖋 19:50, 15 March 2018 (UTC)
- I would like to know what genius thought it was good idea to only be able to select one or all filters in NPP.
- It looks as if the trial absolutely confirmed what you all have been saying for years: Wikipedia is better off if brand-new users are not writing articles as their first edit here. The brand-new editors themselves are also better off, since it is extremely frustrating to have their first attempts at editing unceremoniously rejected. The system worked EXACTLY as hoped for, preventing many inappropriate articles, saving untold hours of volunteer time, and avoiding burnout by the NPP volunteers since they don't have to spend so much time dealing with hopeless articles. The results are so clear and so positive that I hope the WMF will allow the system to be immediately switched back on while discussion about improvement goes on. --MelanieN (talk) 19:24, 15 March 2018 (UTC)
- I am not sure what you are trying to say X. —usernamekiran(talk) 20:42, 15 March 2018 (UTC)
- "I'm saying that if I want to review orphaned articles from new users, I can do it. And why is there even a filter for bots? - MrX 🖋 21:05, 15 March 2018 (UTC)
- Yeah, that one or all is confusing. And for bot creations, there have been a lot of bots in past who created articles. Currently, there is one ongoing bot, Qbugbot; he is planning to create around 17,000 articles on bugs. So when he goes operational, this filter will be handy. There was a lengthy conversation at village pump. Although; last time I checked, it was going in the favour of bot publishing articles in draftspace, and then a human moving it in mainspace. —usernamekiran(talk) 00:08, 16 March 2018 (UTC)
- "I'm saying that if I want to review orphaned articles from new users, I can do it. And why is there even a filter for bots? - MrX 🖋 21:05, 15 March 2018 (UTC)
I have read the report and identify two key and potentially opposing concerns:
- Sustaining the manpower required for maintenance activities
- Maintaining the collaborative Wiki Way for new articles
The report tries to claim that maintenance load is unchanged because NPP work just moved to AfC. That is not actually well supported in the report and is disputed by TonyBallioni above. I am not aware of any technology-based proposals that will make these maintenance activities significantly easier so we either need more manpower or less work to keep things in balance. Given our long-term trends in active editors and administrators, the more-manpower thing seems unlikely to happen.
The Wiki Way thing is up to the community to decide. Sentiment on this has clearly shifted over the years I've worked on the project and now favors and has adopted more restrictions on creating new articles and more enthusiasm for deleting marginal material. The WMF and I may not be happy with this shift but this is a volunteer, consensus-based project and so it needs to be respected.
In this context, the path I would choose going forward is to restore the ACTRIAL-style restrictions on new page creation and focus on improving things at AfC. If we can get promising AfC submissions to mainspace more quickly and efficiently, we can still have the Wiki Way but avoid the unnecessary churn and maintenance overhead associated with unrestricted creation. ~Kvng (talk) 14:09, 16 March 2018 (UTC)
- Even if the absolute raw numbers are transferred directly to AfC (surely a correlation, but not a one to one relationship), CSD and PROD requires two people minimum, AfD often many more, and the only workload transfer that actually increases overall people involved is accepted articles at AfC by non-autopatrolled users, and repeatedly rejected articles, many of which would likely be submitted to AfC and repeatedly rejected anyway (at least I would argue).
- As to the wiki-way, I'd say it's difficult to argue in any way that a rejection at AfC and an invite to the Teahouse is ever going to be more BITEY than a big red template saying "You as an individual and/or your company or band is worthless to us, and we're deleting your work post haste". GMGtalk 14:17, 16 March 2018 (UTC)
- Yes, there is certainly a correlation (the landing page was designed to help people make AfC submissions amongst other things). It isn't 1:1 though, and the fact that we've seen an approximately 10,000 page decrease in the new pages feed while only seeing less than 1000 increase at AfC I think speaks volumes (and that backlog has been going up and down, while the NPP has been heading steadily down).The other thing to point out, as I did above, is that for the first time in at least two years both AfC and NPP are at their sustainable points: the number of AfC submissions that is over 60 days in only 4. The NPP feed is below the 90 day no index point. The actual gross numbers themselves don't matter as much so long as these two statistics remain low or stable: I would be fine with a new pages feed of 100,00 articles if we were able to keep it under the no index point. TonyBallioni (talk) 14:58, 16 March 2018 (UTC)
- Also, as an aside about manpower since it was raised, the idea that our manpower is shrinking is a myth. The English Wikipedia reached a low point of highly active editors in November 2014 at 2971. Since then we've never had a month below 3,000 highly active editors. If anything, we're moving slightly up with a trend towards a median of 3,500 HAE each month. Our active editor pool has also consistently stayed in the ~30,000 realm. Not relevant to the main point of ACTRIAL, but it is indeed possible to recruit more manpower to our individual projects, and the project as a whole is staying stable with a slight trend up. TonyBallioni (talk) 15:18, 16 March 2018 (UTC)
- We know that ACTRIAL worked, and we already know about the inadequacies of NPP and AfC, but some users above may have not have noticed whaI was inferring with my original post:
The conclusion to the report strongly reiterates our long-time on going requirements for the NPP and AfC processes to be improved. (...) This is now the moment to continue to collaborate with DannyH (WMF) and his team to bring the entire Curation system up to date by making a firm commitment to addressing the list of requirements to the excellent suite of tools the WMF developed for Curation. The conclusions also make some strong recommendations for AfC.
Which means in fact that while they have correctly identified a need, the WMF hasn't told us how to improve these systems. It's up to us to tell ourselves what we need to do. Which means coming up with some concree suggestions we can work on, such as for example, addressing that curation to-do list, and figuring out how AfC can raise its performance. Kudpung กุดผึ้ง (talk) 15:39, 16 March 2018 (UTC)
- The immediate next step for my team is to learn more about AfC, and see if there's something we can help with there. I feel like we understand NPP a lot better than we did last year, but we don't know the AfC process very well, and I think they've received even less software support than NPP has. If ACTRIAL is implemented permanently, then that's going to add more burden to that backlog, so we want to make sure we're not crushing people. A product manager from my team is going to be reading and talking to folks in that area over the next couple weeks. Is there a place that's similar to the suggested improvements page where people are talking about the pain points for AfC? -- DannyH (WMF) (talk) 17:19, 16 March 2018 (UTC)
- Thank you for your input Danny. Because AfC and NPP go hand-in-hand, I guess this project is as good a place as any. Primefac who is the de facto coord of AfC watches this page and can draw his team into the discussions. My personal view is that the best options are either to merge NPP with AfC so the work with their respective templates can be handled from one GUI, or to elevate AfC to an official process as I did for NPP. Previous to that I had introduced some better vetting of candidates for AfC reviewing. It's not perfect, but Primefac continues to take care of it very well. Kudpung กุดผึ้ง (talk) 17:35, 16 March 2018 (UTC)
- The immediate next step for my team is to learn more about AfC, and see if there's something we can help with there. I feel like we understand NPP a lot better than we did last year, but we don't know the AfC process very well, and I think they've received even less software support than NPP has. If ACTRIAL is implemented permanently, then that's going to add more burden to that backlog, so we want to make sure we're not crushing people. A product manager from my team is going to be reading and talking to folks in that area over the next couple weeks. Is there a place that's similar to the suggested improvements page where people are talking about the pain points for AfC? -- DannyH (WMF) (talk) 17:19, 16 March 2018 (UTC)
- See WP:AFC talkpages. The templates need work. We need to discourage resubmission of crap and/or seek deletion sooner. We also should be encouraging promotion of notable topics even if not perfect. These issues can be solve with discussion and better instructions for reviewers. The backlog is less than 2500 pages and could be cleared with some additional attention. I'd encourage WMF people to add the AFCH script and do 50 submissions to really understand the process before trying to improve it. Legacypac (talk) 17:46, 16 March 2018 (UTC)
- I think combining the two is the way to go. It may even prove helpful in recruiting article creators to grow into new editors by offering them guidance and working with them from creation right thru the review process. I'm also of the mind that a hands-on, combined approach will serve as a deterrent to spammers and COI editors. It will certainly help maintain the quality of the encyclopedia. Atsme📞📧 18:33, 16 March 2018 (UTC)
- @DannyH (WMF): There are always areas for improvement, but in my opinion the AfC tools (i.e. WP:AFCH and various bots) are actually better than Page Curation, because they are community-developed and therefore have been kept up-to-date. The problem with AfC isn't the tools, but the process itself. You made what I thought was a very perceptive comment in your report:
The Wiki Way is to publish instantly, but make it easy to undo [... not] review-then-publish
. AfC has been the one major exception to that model for years. I think for that reason the community has never been happy with it. Reviewing is a thankless task even by WP standards and every now and again there is a discussion where the wider community jumps on AfC for having too high standards, for alienating new editors with repetitive templated messages, for allowing single reviewers to determine the fate of hundreds of articles, for being inconsistent and lacking quality control, etc etc. And that's all valid criticism, I think even most reviewers would agree – but nobody seems to know how to fix it.
- That is, of course, exactly why so many of us have come over to Kudpung's idea of merging with AfC with NPP, which is a process that does follow the "Wiki Way" and runs much better for it. Again, the question is how. – Joe (talk) 20:42, 16 March 2018 (UTC)
"The Wiki Way is to publish instantly, but make it easy to undo [... not] review-then-publish." is great if someone adds junk to an existing page because any editor can easily revert it. If someone adds a whole NEW PAGE of junk (very easy) it takes a whole process involving multiple people to remove the NEW PAGE (very hard). That is not the Wiki Way, it is unbalanced in favor if idiots and spammers who in seconds can waste an hour or more of good faith editor and Admin time. Legacypac (talk) 20:53, 16 March 2018 (UTC)
- Agreed. We need to look at article improvement and new page creation differently. We probably need to handle them differently. We already do handle them differently. Both AfC and NPP have editors who review stuff before it gets too much exposure. The Wiki Way for new pages does not exist now because community consensus does not support it. ~Kvng (talk) 22:37, 16 March 2018 (UTC)
- NPP is a triage, AfC is more of a field hospital. A lot depends now on how much effort (and money) the WMF is prepared to help with solutions. Many of the required changes are going to be beyond what we can reasonably expect communiy geeks to do for free. Danny has stated that the WMF has learned a lot through the ACTRIAL exercise, and I think the Foundation now understands that with 5.5 million articles, the focus on en.Wiki has now to be more on quality rather than quantity (of both new editors and new articles). Kudpung กุดผึ้ง (talk) 07:55, 17 March 2018 (UTC)
Different conclusion
I would have written a different conclusion based on the content of the report. I do agree that AfC deserves more attention but I don't think you're going to find major surprises there. The major takeaway missing from the conclusion is the demonstrated viability of making ACTRIAL a permanent improvement to page curation. This is supported by the following trial findings:
- No apparent effect on new user activity levels and retention.
- No apparent effect on article growth.
- There is a reduction in encyclopedic content being created in the article namespace.
- Reduction in NPP backlog
The only thing I find in the report speaking against a permanent ACTRIAL is concerns about AfC. It is possible, reading the report, that readers would come to the conclusion that AfC requires improvement before it is ready for a permanent ACTRIAL but these concerns are not actually substantiated in the report and the authors admit their understanding of AfC is limited.
Based on the well-substantiated positives of the trial, and no well-substantiated negatives can we please include instituting a permanent ACTRIAL as a recommendation to WMF? ~Kvng (talk) 15:46, 17 March 2018 (UTC)
- Part of the agreement with the Wikimedia Foundation when we launched this was that they would not take part in framing the discussion after ACTRIAL as to whether or not we should make it permanent. That would be entirely community driven. I agree with your conclusions here, and I think most people who follow this talk page do. As I've mentioned here and on other pages, Kudpung and myself are working on getting an RfC up very soon that will propose making the autoconfirmed requirement for article creation permanent. TonyBallioni (talk) 15:54, 17 March 2018 (UTC)
- By publishing this report I would say they have clearly taken part in framing the discussion. I am trying to address this by identifying and editing unsubstantiated and non-NPOV content in the report. So far I have not been particularly WP:BOLD about this but I am willing to take it up a notch if the authors (Morten Warncke-Wang, with Ryan Kaldari and Danny Horn) can deal with some WP:BRD. If we can get consensus on the report, I assume it will be easier to get consensus on actions based on the report's findings. ~Kvng (talk) 15:34, 18 March 2018 (UTC)
- Kvng: I don't think the report suggests that AfC requires improvement before there's a permanent ACTRIAL, and as TonyBallioni says, we're not steering the on-wiki discussion about it. The report offers suggestions for both the English WP community and the WMF to discuss, but doesn't say whether ACTRIAL should continue or not. That's up to the future discussions. -- DannyH (WMF) (talk) 18:57, 19 March 2018 (UTC)
Some stats on the AfC process compared to NPP and notability and some ideas to float for AfC reform
Note that I have posted a breakdown of some stats that Legacypac and I gathered, as well as some ideas for reform at AFC on the AfC talk page HERE. Please have a read through and comment there. Cheers, — Insertcleverphrasehere (or here) 05:31, 21 March 2018 (UTC)
Copied here:
Some stats on the AfC process compared to NPP and notabilty and some ideas to float for AfC reform.
There has been a lot of criticism directed at AfC of late, and I decided on a little research project to collect some data. Legacypac and I gathered up a random selection of AfC drafts and reviewed 100 of them, we recorded the expected result of the AfC review process (outlined HERE), the expected result of the NPP review process if that draft had been in the main space (per the process outlined HERE), as well as assessing the notability of the draft's topic by examining the sources in the article as well as performing searches (per WP:BEFORE). While another pair of reviewers might have reviewed these submissions slightly differently, we don't feel that the result would be so different that we cannot draw conclusions from the data. The results are summarised in the chart above.
- Results:
- The AfC process is considerably more strict than NPP when it comes to accepting submissions. In particular, the AfC review process regularly declines drafts on topics that are clearly notable (17 out of 100 drafts were declined despite being clearly notable). The proportion of notable topics that are currently being declined at AfC indicates that AfC is perhaps demanding too much of new editors.
- In the case of topics of 'borderline' notability, it seems that both processes are likely to proceed with caution, with the result that AfC will most often decline borderline topics, while NPP will often accept some borderline topics. This result is to be expected, and perhaps, desired.
Ideas for improvement of the AfC process
I'd just like to float some ideas that I've had with regards to potential improvements to the AfC process.
- Require that reviewers do a search for notability before declining based on notability (WP:BEFORE for AfC).
- The current system declines many clearly notable topics. With this change, reviewers would perform a search, and if good reference material is found, the reviewer could simply accept the article and add a tag (
{{sources exist}}
) indicating that they had performed a search and found the subject to be notable (to prevent others from having to search again and discouraging other editors nominating the article for deletion in ignorance). Alternatively, the AfC reviewer could simply drop a couple references from their search into the article.
- The current system declines many clearly notable topics. With this change, reviewers would perform a search, and if good reference material is found, the reviewer could simply accept the article and add a tag (
- Overturn WP:NMFD. At least with regards to multiple-declined drafts where the submitter has been told several times that the subject is not yet notable (not with regard to userspace content or unsubmitted drafts).
- This would allow us to deal with repeated submissions on non-notable topics that tend to clog up the AfC process, and this suggestion enables the next one:
- Discussion at MfD for repeatedly resubmitted drafts:
- On the third decline with unaddressed issues AfC reviewers are recommended to list the draft MfD for discussion, if notable it should be moved to main space, if not-notable it should be deleted. If the topic looks promising for future notability (i.e. a film in pre-production) it could be userified with a prohibition on submitting without new high-quality sources.
- Reviewers should not decline the same submission more than once (prohibition does not apply to commenting on or accepting a draft previously declined).
- Repeated declines from the same reviewer decrease collaboration in the AfC workspace, a frequent criticism of AfC. This change would fix an issue that currently exists where one user watchlists a draft, and will see when it is submitted again before other reviewers get a chance to have a look at it. Having a different set of eyes review a submission each time ensures that the maximum amount of collaboration possible occurs. While it may seem a bit like asking the other parent, that is intentional, and most reviewers should be reviewing to roughly the same standard anyway (if they are not, then it is a problem that this change will help identify).
I am keen to know what you guys think with regards to the data displayed above, and also how feasible/workable you think the ideas above might be. If any ideas get a predominantly positive reception we could propose one or more of these ideas as a formal change to the AfC process. I am also keen to hear any other ideas that you guys might have come up with for improvements to AfC. Cheers, — Insertcleverphrasehere (or here) 05:27, 21 March 2018 (UTC)
- That was an excellent study. Thank you cleverPhrase and Legacypac.
- Ideas:
- RE #1. Support, and support implication that notable topics should be mainspaced even in poor condition.
- RE #2. Suggest alternative, WP:Drafts for discussion. Warn that deletion for non-notability necessarily raises the bar, raises the difficulty of reviewing and making a decision. Urge that failure results in a blunt rejection, but not deletion, at least not immediate deletion. The author may need to review what was rejected.
- RE #3. Oppose. Repeated resubmission is a result of a combination of failure of AfC decline notices to adequately convey the intended message, and behavioural issue. The first should be remedied by better AfC responses, the second as a behavioural matter, warnings, blocking.
- RE #4 Suggest that all reviewer comments go on the talk page. On that talk page, each review comments in their own thread, and adds comments to their earlier thread if commenting a second time. However, it would be far preferable if post submission comments were to become more normal human-like conversation on the talk page, resembling the talk culture in other namespaces.
- I think the data is great to have. I feel the data rings true. Something has to be tried.
- RE Wikipedia:WikiProject Articles for creation/Decline comments workshop. It is hard to say this there, I think there are too many templated options. As per my notes at User:SmokeyJoe/AfC_wishlist#AfC_acceptance_/_rejection_process_improvement, I think five templates is the right number, and that the detail of the comments should go in threaded discussion on the talk page. --SmokeyJoe (talk) 05:56, 21 March 2018 (UTC)
- Thank-you to User:Insertcleverphrasehere for setting this up. The 250 drafts choosen were truly random (every 10th waiting) and the 100 processed were also random. The percentages can be fairly applied to pages waiting, but not all submissions because most 0-1 days easy accepts and rejects get processed quickly so escaped our sample. I agree with the conclusions presented.
- We can eliminate the backlog by making some key changes:
- More quickly accept the imperfect but notable and verifiable. This will reduce the submit-reject cycle. We are dealing with new users that don't know how to fight for their notable topic and may not be up to the task of building out a perfect page. They often quit after just one or two submissions which is a waste of both creator and reviewer time.
- look at the page like a NPR amd Seek immediate deletion of unacceptable pages. If it can be CSD'd (G11, G12, G2 amd G3 mostly) CSD it. If the CSD is declined or not CSDable but needs deleting, take to MfD. It is a little extra effort but this will cut down on resubmissions and save another editor later figuring out why to delete it.
- In theory in the hands of a good reviewer nearly every submission could be promoted as notable or deletion sought on the first AfC pass. Let the creator work on the borderline cases for a second submission perhaps. No page should get to 4 to 11 declines like the ones on this list. User:JJMC89_bot/report/AfC_decline_counts as that is just a big waste of reviewer time. Legacypac (talk) 07:02, 21 March 2018 (UTC)
- Don't at all agree that we should be lowering the quality of articles accepted to mainspace. Do we see a sufficient level of improvement of articles already here, no years-old cleanup tags, everything kept updated, etc? We do not. There should be some expectations laid on new article creators, pointing to help and guidance and so on. A "decline" is nothing like a deletion: it's like a publisher returning a MS saying "thank you, we'll accept this, but it first needs a bit more work on your part". A bit of BURDEN should be applied here:
The burden to demonstrate verifiability lies with the editor who adds ... material, and is satisfied by providing a citation to a reliable source that directly supports the contribution.
That's on a policy page, and why wouldn't it apply to a new article as much as to a change to an existing one?: Noyster (talk), 09:15, 21 March 2018 (UTC) - On a similar point to Insertcleverphrasehere's initial post, the AfC reviewing instructions don't say explicitly that reviewers must do a WP:BEFORE search when reviewing an article. (There is the "would probably survive a listing at AfD" language, but the connection there is not explicit.) As a project, perhaps we should iron out exactly how much work reviewers should be doing to assess the potential notability of a draft. Maybe it is adherence to WP:BEFORE, maybe it is a lesser standard or something else entirely, but at any rate we should have a larger discussion around this. /wiae /tlk 09:50, 21 March 2018 (UTC)
- I want to add a second point that slipped my mind when I posted earlier. There is additional tension between the notion of reviewers following WP:BEFORE and the current notability decline reasons. Take the "bio" decline reason for example: "This submission's references do not adequately show the subject's notability". The way this reads, a reviewer can logically decline an article on a notable person if the current state of the article's references don't demonstrate the notability. (Indeed, this is basically what Noyster alludes to above.) So if we want to codify adherence to WP:BEFORE or some other "is this topic objectively notable" process in the reviewing instructions, the notability decline reasons should probably be reworded. I guess I'll put a plug in here for Wikipedia:WikiProject Articles for creation/Decline comments workshop, as the outcome of this discussion could affect what goes on at that subpage. Thanks, /wiae /tlk 12:44, 21 March 2018 (UTC)
- You could start by tagging first and second person spam submissions for deletion. I occasionally come across pure spam AFC submissions that were reviewed but not tagged. It boggles the mind. MER-C 10:48, 21 March 2018 (UTC)
- I must agree with Noyster, we should not be lowering the quality of submissions by accepting everything that's notable no matter what mess it's left in by the author OR the reviewer - there's nothing wrong in a reviewer providing feedback saying 'this topic is notable, but your draft needs a little more work before it can be accepted'. I'd also wonder if Legacypac and Insertcleverphrasehere are looking at this correctly - instead of looking at notable content being declined, a more useful metric (particularly from the much talked about and fatally flawed editor retention point of view) is whether a more rigorous AFC process makes articles less likely to be nominated for and/or actually deleted. Nick (talk) 12:15, 21 March 2018 (UTC)
- Approximately 10% of articles that were accepted at AFC get nominated for deletion, with about 5-7% eventually being deleted. Insertcleverphrasehere, do you have the numbers for articles created directly in mainspace? Primefac (talk) 12:36, 21 March 2018 (UTC)
- data pre ACTRIAL showed about 20% of AutoConfirmed user creations are deleted. A few more get redirected, which would include some merges. Legacypac (talk) 15:52, 21 March 2018 (UTC)
- Point #4 has always been a personal philosophy of mine, and one that I've encouraged others to hold to, but it might be worth codifying (for exactly the reasons you've given). Primefac (talk) 12:42, 21 March 2018 (UTC)
- I'm struggling to see the value of this study. The links that appear in the brief description of the "expected" AfC and NPP results are merely links to the project pages that contain the instructions. How were these instructions translated into numerical percentages? And these "expected" percentages are then compared to the authors' own opinions as to what the "correct" results should be. In all, I'm struggling to see how this study does anything more than produce a pretty graph that illustrates the authors' opinions.
But the more basic issue is the false dichotomy of "accept/decline". AfC is an interactive process and I wholeheartedly agree with Nick's statement above -- there truly is nothing wrong with declining a submission but then immediately opening a discussion with the new editor on the Draft's Talk page. In my experience, new editors who are writing non-promotional articles actually do want to learn how to produce a better piece of work and are quite happy to receive specific advice on how to do it. Also in my experience, these editors don't take more than two or three days to implement the improvements. A two- or three-day delay is a small price to pay in return for a new editor who will now go out into the wider Wiki-world and not repeat the same basic errors that appeared in the draft. And all it takes to get this trade-off is a group of AfC volunteers who are willing to talk to the new editors in a constructive fashion. NewYorkActuary (talk) 13:55, 21 March 2018 (UTC)
- To clarify; we followed the linked reviewing instructions for both projects to the best of our personal judgement on 100 random drafts from the AfC backlog, thus indicating the percentages for each result (the data page is linked to the word 'data' in the first sentence). A quick look at the list of abandoned drafts indicates that unfortunately many editors don't resubmit after being declined. This then eventually results in G13 deletions of their article on a notable topic, which doesn't seem acceptable to me. I get where you are coming from, and if every reviewer collaborated the way you describe with new editors, AfC would be a better place, but it would be nice to implement a system that ensures that notable submissions are not lost (unless they would be CSD/BLPRODable in the main space). Perhaps if the reviewer was required to do a notability search, and inform the author of the result (i.e. "your submission seems notable, but it needs a couple more references, try searching on google news, I fear this might get deleted if moved to main space". If you combined this with a change to the resubmit box so that it contained a message like; "AfC is an entirely optional process, you are free to move this page to the main article space yourself." Some food for thought anyway. I'm enjoying reading the comments here, some interesting perspectives and ideas. — Insertcleverphrasehere (or here) 19:08, 21 March 2018 (UTC)
- Question - Can someone summarize for me where this eagerness to delete drafts comes from? I'm pretty sure the total effort put into the MfD discussion is greater than the AfC reviewer pool having to do multiple declines on the draft before the author gets the message and the draft drifts off without drama through G13. ~Kvng (talk) 14:39, 21 March 2018 (UTC)
- Oppose 1 - We don't need to make reviewing more difficult or time consuming. Authors should not be deprived of the learning experience of finding, evaluating and adding references. Most of the previously-declined drafts I review have been improved in some meaningful way. Notability searches can be done in G13 salvage review by those WP:VOLUNTEERs interested in doing so. ~Kvng (talk) 14:45, 21 March 2018 (UTC)
- Non-viable, tendentiously resubmitted drafts should be deleted via MFD. I also support the suggestion by User:SmokeyJoe to remove the resubmit button for hopeless drafts (though I argue that they should be deleted promptly). MER-C 15:33, 21 March 2018 (UTC)
- I have an issue regarding adding sourcing to drafts. Authors of these drafts could simply come up with a notable subject, and write a line or two with a source; and as reviewers we wouldn't be able to deny it, and would have to work on the article ourselves. Surely notable topics that haven't been written well/sourced well should be left in draft. There are plenty of "Requested articles" in WikiProjects, which are usually notable topics that haven't been written yet. But an article shouldn't be written purely to strike that entry off the list. I feel as though our goal should never be to encourage poor articles, but to help new users write better articles. Lee Vilenski (talk • contribs) 16:13, 21 March 2018 (UTC)
- Even a one line stub on a clearly notable topic submitted to mainspace can actually be an acceptable submission that shouldn't be deleted, people create these all the time, and while NPP often adds tags to them informing the author that they need additional references etc, often nothing is done about it. The same is true in draft space, editors often don't resubmit, resulting in the loss of a notable topic start. I don't see much reason why a new user would be any more inclined to fix the issue in draft space vs article space, given that they are informed of the issue even if they just go the NPP route, but at AfC their submission is held hostage, even if they can't figure out how to add references. I've seen lots of notable topics declined by reviewers despite good references just because the author had only used bare urls and not included them as incline citations (we don't even require inline citations in main space). — Insertcleverphrasehere (or here) 19:08, 21 March 2018 (UTC)
- If the one line stub is un-sourced, then it's not an appropriate submission, no matter how notable it might be (to actually assess notability in the absence of any sources, you would be forcing the reviewer to find and assess sources, some or all of which may well differ from those the author used to write the content - far from ideal, maybe even far from possible if it's a historical work for which dead tree media is being used). The last thing we want to be doing is removing the fair, reasonable and friendly approach AFC provides for sourcing, and in particular, sourcing of BLPs, and instead letting new users out into the article space where they're going to be finding more deletion notices (CSD, BLPPROD and AfD) and they're going to be getting warnings and sanctions under the BLP policy. Just remember for a moment that there are ArbCom Discretionary Sanctions available for those violating the BLP policy - do you want new users being hit with DS notifications and then sanctions because they can't work out how to add a reference, or do you want them confined to Draft space in the company of helpers and reviewers who can point them in the right direction and help them reference their new content ? Nick (talk) 11:24, 22 March 2018 (UTC)
- }Per WP:NEXIST and WP:BEFORE, that isn't quite right. A a notable topic with a one line unsourced stub that establishes context, and also contains a credible claim of notability and is not a BLP (Avoiding CSD on A1, A7, and A9, and also avoiding BLPROD) should actually survive deletion and should not even be nominated in the first place. Per WP:BEFORE, reviewers are required to search for sources, and if they are found, are not supposed to delete the article. My preferred solution in these cases is to quickly copy a couple of references to the end of the first line (using the proveit auto reference gadget can do this quickly), or else tag it with the (
{{sources exist}}
) tag. — Insertcleverphrasehere (or here) 20:40, 22 March 2018 (UTC)- The problem with relying on guidelines like WP:NEXIST and WP:BEFORE is that you're relying entirely on the reviewer being able to find sources, to assess those sources, and to make a determination on the notability based on what they find. That sometimes works when sources are digital and online, it's useless when you're looking at an article or draft written around print source material and difficult when digital sources are behind a paywall. WP:NEXIST and WP:BEFORE provide little or no meaningful protection for many of the articles we see submitted through AFC, and even if an article isn't ultimately deleted, the process of dealing with CSD, PROD or AFD notices on your new article is a significant deterrence to continued contributing to Wikipedia. Nick (talk) 23:31, 22 March 2018 (UTC)
- }Per WP:NEXIST and WP:BEFORE, that isn't quite right. A a notable topic with a one line unsourced stub that establishes context, and also contains a credible claim of notability and is not a BLP (Avoiding CSD on A1, A7, and A9, and also avoiding BLPROD) should actually survive deletion and should not even be nominated in the first place. Per WP:BEFORE, reviewers are required to search for sources, and if they are found, are not supposed to delete the article. My preferred solution in these cases is to quickly copy a couple of references to the end of the first line (using the proveit auto reference gadget can do this quickly), or else tag it with the (
- If the one line stub is un-sourced, then it's not an appropriate submission, no matter how notable it might be (to actually assess notability in the absence of any sources, you would be forcing the reviewer to find and assess sources, some or all of which may well differ from those the author used to write the content - far from ideal, maybe even far from possible if it's a historical work for which dead tree media is being used). The last thing we want to be doing is removing the fair, reasonable and friendly approach AFC provides for sourcing, and in particular, sourcing of BLPs, and instead letting new users out into the article space where they're going to be finding more deletion notices (CSD, BLPPROD and AfD) and they're going to be getting warnings and sanctions under the BLP policy. Just remember for a moment that there are ArbCom Discretionary Sanctions available for those violating the BLP policy - do you want new users being hit with DS notifications and then sanctions because they can't work out how to add a reference, or do you want them confined to Draft space in the company of helpers and reviewers who can point them in the right direction and help them reference their new content ? Nick (talk) 11:24, 22 March 2018 (UTC)
- Even a one line stub on a clearly notable topic submitted to mainspace can actually be an acceptable submission that shouldn't be deleted, people create these all the time, and while NPP often adds tags to them informing the author that they need additional references etc, often nothing is done about it. The same is true in draft space, editors often don't resubmit, resulting in the loss of a notable topic start. I don't see much reason why a new user would be any more inclined to fix the issue in draft space vs article space, given that they are informed of the issue even if they just go the NPP route, but at AfC their submission is held hostage, even if they can't figure out how to add references. I've seen lots of notable topics declined by reviewers despite good references just because the author had only used bare urls and not included them as incline citations (we don't even require inline citations in main space). — Insertcleverphrasehere (or here) 19:08, 21 March 2018 (UTC)
Examples
There is a long held idea starting with Jimbo Wales himself that one editor puts in an idea and another adds a little and over time an article builds. Joseph_Bishop is a page I accepted based on meeting WP:PROF amd tagged with "Sources exist". Well overnight the page has been greatly sxpanded from the nice friendly page I accepted. Legacypac (talk) 16:22, 21 March 2018 (UTC)
- Nice example. New editors actually learn from the edits that others make to the pages they create as well, as most of them will follow along to see what happens to the article they created. This collaboration and 'teaching by demonstration' is generally lost with the current AfC process. — Insertcleverphrasehere (or here) 19:11, 21 March 2018 (UTC)
- And we're all just gonna ignore the fact that it's a massively huge copyvio, I guess. CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 13:57, 22 March 2018 (UTC)
- oh, you're no fun anymore..... Ritchie333 (talk) (cont) 14:07, 22 March 2018 (UTC)
- If my work with copyvios makes me 'not fun' anymore, I'm okay with it. This was when it was accepted. CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 14:11, 22 March 2018 (UTC)
- Yes, I know - I dealt with it and threw in some Pythonesque humour for a bit of a laugh - we're not allowed to link to the actual video on YouTube because it's a .... well .... Ritchie333 (talk) (cont) 14:24, 22 March 2018 (UTC)
- Oh well, new users will be new users. — Insertcleverphrasehere (or here) 14:26, 22 March 2018 (UTC)
- New users will be new users isn't an excuse or acceptable reason for an established AFC reviewer accepting at least (so far, in the last 24 hours alone) 5 copyvio drafts and moving them into mainspace without even acknowledging the CV. CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 14:34, 22 March 2018 (UTC)
- Oh well, new users will be new users. — Insertcleverphrasehere (or here) 14:26, 22 March 2018 (UTC)
- Yes, I know - I dealt with it and threw in some Pythonesque humour for a bit of a laugh - we're not allowed to link to the actual video on YouTube because it's a .... well .... Ritchie333 (talk) (cont) 14:24, 22 March 2018 (UTC)
- On the specific draft a good portion of it was university positions and dates which is not really copyrightable. If Bishop was President of X School from 1964-1967 and the Dean of Y school from 1967-1969 there are only so many ways to write that. I also note all the additions by mainspace editors were removed as copyvio which seems highly unlikely they were copyvioing in a few words at a time.
- If my work with copyvios makes me 'not fun' anymore, I'm okay with it. This was when it was accepted. CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 14:11, 22 March 2018 (UTC)
- oh, you're no fun anymore..... Ritchie333 (talk) (cont) 14:07, 22 March 2018 (UTC)
- I look for CV by searching key phrases and I often find it. I've never been able to get earwigs tool to load properly though. Maybe ChrissyMad's diligence should be put to good use by her checking more submissions for cv. I really appreciate the reviewer who marks drafts as CV checked. Legacypac (talk) 16:05, 22 March 2018 (UTC)
- Yes. Let’s be more careful with copyvio please. It’s no secret that I’m a fan of Legacypac’s approach to AfC and draft space, but moving CV to mainspace shouldn’t happen. TonyBallioni (talk) 14:46, 22 March 2018 (UTC)
- New users will be new users is an excellent example of why keeping AFC a little more on the strict side is important - it gives reviewers the ability to check for copyright issues. That is of course aided when there's sources against which one can check not just for a copy and paste copyright violation but also for less apparent issues such as close paraphrasing or image copyright problems. Nick (talk) 15:02, 22 March 2018 (UTC)
- I look for CV by searching key phrases and I often find it. I've never been able to get earwigs tool to load properly though. Maybe ChrissyMad's diligence should be put to good use by her checking more submissions for cv. I really appreciate the reviewer who marks drafts as CV checked. Legacypac (talk) 16:05, 22 March 2018 (UTC)
Everybody looking at NPP and AfC needs to arm themselves with importScript("User:The Earwig/copyvios.js");
in Special:MyPage/common.js - every time you see a big article with flowery language outside of Wikipedia's style, hit the button and check for copyvios. Ritchie333 (talk) (cont) 15:11, 22 March 2018 (UTC)
- For the initiated (Me), I currently run Earwigs Copyvio as a bookmark and run it as a separate page. How does the script work differently? My knowledge of coding is very poor, sadly. Lee Vilenski (talk • contribs) 15:47, 22 March 2018 (UTC)
- It's basically a shell around the web version - it takes the current page and auto-fills the relevant parameters across. Ritchie333 (talk) (cont) 15:53, 22 March 2018 (UTC)
- For the initiated (Me), I currently run Earwigs Copyvio as a bookmark and run it as a separate page. How does the script work differently? My knowledge of coding is very poor, sadly. Lee Vilenski (talk • contribs) 15:47, 22 March 2018 (UTC)
- Yes, it is by far one of the most useful scripts on this project. TonyBallioni (talk) 15:38, 22 March 2018 (UTC)
- For once, we agree on something Ritchie333 ;) I'm actually a bit surprised it's not automatically built into the AFCH script (or maybe it is and we don't advertise it well enough?) In any case, my normal routine with reviewing drafts is to open the CV check in an ew window (via the tool) and look at the draft while it's running. I do it for every draft (the obvious exception being nonsense) even if I'm declining. CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 15:43, 22 March 2018 (UTC)
- We should catch copyvio wereever we can but it is potentially more quickly spotted in no index mainspace than no index draft space because more eyes on the page. Legacypac (talk) 16:05, 22 March 2018 (UTC)
- How do you figure there would be more eyes on an unwatched page in mainspace, than an unwatched page in draft space? SQLQuery me! 16:36, 22 March 2018 (UTC)
- A page in NPR gets quite a few eyes and checks and editors that categorize and verify etc. Not so much as an AfC Draft. The page could been placed directly in mainspace vs moved into mainspace and it will get the same reviews. We all know there are plenty of copyvios on the site, and we wack them as we find them or even better, rewrite the material to solve the problem. Legacypac (talk) 19:22, 22 March 2018 (UTC)
- Legacypac If it's so easy to rewrite, why don't you do it in the first place instead of moving it to mainspace where it will be indexed and cause legal trouble? Can you please, going forward, just commit to actually making a good faith effort to check for copyvios before moving anything to mainspace? CHRISSYMAD ❯❯❯¯\_(ツ)_/¯ 19:27, 22 March 2018 (UTC)
- Ritchie333, What does the copyvios script do? I installed it but nothing new comes up. Dial911 (talk) 17:46, 22 March 2018 (UTC)
- You need to be browsing the article first (ie: not editing, viewing history or anything else) before you'll see an option. For example, if I browse to Ballynahinch Castle, and click on the link "copyvio check", I get this. Ritchie333 (talk) (cont) 17:49, 22 March 2018 (UTC)
- I think I have done something wrong. I can’t see ‘copyvio check’ button. Will check my script page once again. Dial911 (talk) 18:02, 22 March 2018 (UTC)
- Hmm, might be worth raising it on the Technical village pump. Ritchie333 (talk) (cont) 19:31, 22 March 2018 (UTC)
- It's not just you - I have the same problem. SQLQuery me! 19:44, 22 March 2018 (UTC)
- Is it possible it doesn't work because I run in MonoBook and you're running in another skin? Although that's the whole point of common.js - it's common to all skins. Ritchie333 (talk) (cont) 19:48, 22 March 2018 (UTC)
- Hah! Changing to monobook and back got it. Weird, but works. Thanks! SQLQuery me! 19:51, 22 March 2018 (UTC)
- I use Vector (default) so I tried changing to Monobook but that stacks nav links on top of each other. Still no link to earwigs tool. Is it supposed to be on the left bar or top bar? Legacypac (talk) 20:08, 22 March 2018 (UTC)
- Hah! Changing to monobook and back got it. Weird, but works. Thanks! SQLQuery me! 19:51, 22 March 2018 (UTC)
- Is it possible it doesn't work because I run in MonoBook and you're running in another skin? Although that's the whole point of common.js - it's common to all skins. Ritchie333 (talk) (cont) 19:48, 22 March 2018 (UTC)
- I think I have done something wrong. I can’t see ‘copyvio check’ button. Will check my script page once again. Dial911 (talk) 18:02, 22 March 2018 (UTC)
- You need to be browsing the article first (ie: not editing, viewing history or anything else) before you'll see an option. For example, if I browse to Ballynahinch Castle, and click on the link "copyvio check", I get this. Ritchie333 (talk) (cont) 17:49, 22 March 2018 (UTC)
- A page in NPR gets quite a few eyes and checks and editors that categorize and verify etc. Not so much as an AfC Draft. The page could been placed directly in mainspace vs moved into mainspace and it will get the same reviews. We all know there are plenty of copyvios on the site, and we wack them as we find them or even better, rewrite the material to solve the problem. Legacypac (talk) 19:22, 22 March 2018 (UTC)
- How do you figure there would be more eyes on an unwatched page in mainspace, than an unwatched page in draft space? SQLQuery me! 16:36, 22 March 2018 (UTC)
- We should catch copyvio wereever we can but it is potentially more quickly spotted in no index mainspace than no index draft space because more eyes on the page. Legacypac (talk) 16:05, 22 March 2018 (UTC)
(←) :Left bar, under tools. SQLQuery me! 20:21, 22 March 2018 (UTC)
- It works now and saves time too. Dial911 (talk) 02:50, 23 March 2018 (UTC)