Talk:IP Editing: Privacy Enhancement and Abuse Mitigation/Login Required Experiment

Active discussions

Spanish Wikipedia RequestEdit

After a 2 week period for comments on the Spanish Wikipedia (permalink), most involved users have explicitly supported temporarily turning off IP editing. A formal vote will be taken up, if the team approves this request. CC @NKohli. Best regards, Sgd. —Hasley 00:21, 15 September 2021 (UTC)[]

@Hasley thank you! This is great. I will forward this request to the team and we will get back to you soon. – NKohli (WMF) (talk) 17:12, 15 September 2021 (UTC)[]

Farsi WikipediaEdit

Hello NKohli (WMF),

Today Farsi Wikipedia banned IP editing using edit filter[1]. I think this may be interesting to you. Your team may be able to restrict IP editing through other methods.

The community has already reached a consensus for implementing this change for a two-month period: fa:ویکی‌پدیا:نظرخواهی/تغییر روش ویرایش کاربران گمنام#پیشنهاد چهارم: ممنوعیت موقت ویرایش کاربران گمنام. Thanks 4nn1l2 (talk) 06:36, 14 September 2021 (UTC)[]

  Support Yamaha5 (talk) 15:44, 14 September 2021 (UTC)[]
FYI I turned off the filter; it was throttled anyway. It is impractical to implement this as an edit filter, as the filter will inevitably be throttled in a matter of minutes/hours. I will see through the config changes needed for this to temporarily disable IP editing on fawiki. Phab to follow. Huji (talk) 22:55, 14 September 2021 (UTC)[]
See phab:T291018 Huji (talk) 23:01, 14 September 2021 (UTC)[]
Hi @Huji @4nn1l2. Thank you for bringing this to our attention. I appreciate the RfC Farsi Wikipedia has gone through. It is clearly an issue a lot of people on the project deeply care about. Thank you for the config patch, @Huji. We were not planning on using an edit filter to disable IP edits so I am happy to see this patch.
I am slightly concerned that 2 months is a very short period for the experiment. Some data points like the retention rate will not show up in that time. On our end, we will be able to collect metrics like we did for Portuguese Wikipedia. If there are additional metrics that the community would like us to capture, we would be happy to accommodate that. I would like us to work closely with farsi wikipedia on this. NKohli (WMF) (talk) 17:11, 15 September 2021 (UTC)[]
NKohli, tell me how much time your team needs, and I will try to build a consensus for that. 4nn1l2 (talk) 17:37, 15 September 2021 (UTC)[]
@4nn1l2 thanks -- I think a minimum of 6 months will be needed to get some conclusive data about metric trends. The longer term it is, the better so we can rule out impacts from things like editing campaigns, school/university changes etc. -- NKohli (WMF) (talk) 19:26, 15 September 2021 (UTC)[]
@NKohli (WMF) and STei (WMF): Here you go, a consensus to conduct the experiment for a six-month period: fa:Special:Permalink/33252131#افزایش مدت آزمایش محدودیت ویرایش آی‌پی به شش ماه. ‎4nn1l2 (talk) 07:51, 30 September 2021 (UTC)[]
Thank you for spending time to do this @4nn1l2. The team will get back to you and your community next week. Have a lovely weekend when you get to it. STei (WMF) (talk) 17:07, 30 September 2021 (UTC)[]
Thanks so much @4nn1l2! We are beginning our work to collect metrics and will share them with the community as we gather confidence in the data. This work will be tracked in this ticket. Please reach out to us if you have any feedback or questions. -- NKohli (WMF) (talk) 19:46, 7 October 2021 (UTC)[]

Technical implementation of experimentEdit

If this experiment will be done in any other community, it should be done through proper MediaWiki means, and not through JavaScript hacks and abuse filters like was done in ptWP. It is, frankly, disappointing that WMF had no input on the technical implementation of ptWP ‘experiment’, which replaces edit links only for users with JavaScript (even now), prunes edits from unassuming users through unfriendly abuse filters, shows ‘Edit’ link where no editing is allowed and ‘edit’ leads to ‘log in’, and probably has some other hacks that I did not notice now. If IP edits would be disallowed in some other wiki now, they should be disallowed properly. stjn[ru] 18:13, 9 September 2021 (UTC)[]

I would like to make a comment, related to stjn's thoughts. If the WMF will ultimately test this, please make sure to run a proper experiment. This really, really cannot be overstated. Surely anyone who has had lectures on Experimental Design (even at an undergraduate level) will agree that the actions from and the subsequent WMF report cannot be called an experiment. If the current plan is to just shut down editing for 8 months and use Microsoft Excel to compare a number of YOY trends, this whole endeavour will be fruitless. If you'll do it, do it properly and consider control groups through geoblocking, challenge checkpoints, etc. Thanks. Ñ (talk) 10:47, 13 September 2021 (UTC)[]
Indeed. Folks may also be interested in more extended discussions around methodology at Talk:IP Editing: Privacy Enhancement and Abuse Mitigation/Impact report for Login Required Experiment on Portuguese Wikipedia - Fuzheado (talk) 17:11, 13 September 2021 (UTC)[]
@Stjn: About the MediaWiki means, this may answer you question. ━ ALBERTOLEONCIO Who, me? 03:53, 14 September 2021 (UTC)[]

  Comment stjn, Ñ and Albertoleoncio, about the technical implementation of the experiment, Product is not planning on using an edit filter to disable IP edits. Please see the discussion in the section on Farsi Wikipedia above. Thanks for your feedback. – STei (WMF) (talk) 14:23, 17 September 2021 (UTC)[]

  • STei (WMF) I realise that it might not always be practical for different departments/employees of the WMF to collaborate directly, but rather than you guys trying to reinvent the wheel, I thoroughly recommend that you and NKohli (WMF) liaise with DannyH (WMF) to learn exactly about the excellent help his team provided for the ACTRIAL experiment. Their hyotheseses, data production and analysis was comprehensive and objective. An eventual ban on IP editing will be just as important as ACTRIAL was, and a trial will need some careful planning. Kudpung (talk) 06:05, 21 September 2021 (UTC)[]
    @Kudpung Thanks for calling that out. I was on the team which conducted the ACTRIAL experiment and am familiar with the research that was conducted at the time. We will look into the work that was done and seek out any learnings we may want to replicate here. Thanks. -- NKohli (WMF) (talk) 19:02, 21 September 2021 (UTC)[]

Research goalsEdit

A not-so-short word also on research goals NKohli (WMF) & team :) I appreciate that last year you've been in a delicate position taking the project more or less on the fly, and it's great that you want to continue research on the matter. However, the goals of the project (which I understand as "trying to gain more insight on the effects IP editing ban has") are only answering part of the problem.

How hard would it be to look at it from the other side of "T0": what caused the community to want this ban? How did the number of reverts evolve over the years, compared with anonymous edits/total edits? Did the numbers/trends change when certain features (e.g. AbuseFilters) were introduced? I believe we also need to understand the internal landscape and dynamics leading to these radical decisions so informed decisions can be taken on what alternatives can be provided.--Strainu (talk) 15:04, 11 October 2021 (UTC)[]

@Strainu you bring up an excellent point. It would indeed be helpful to know what led the community to this change. If I recall correctly, I heard that ptwiki lost a few crucial bots which exacerbated their vandalism problem. Having more facts about this would be helpful for us to understand what we can do to help in those areas.
There are also external factors at play. IPv6 has arrived. More people than ever before are using proxies and VPNs because of increasing vigilance around private information.
We have talked about gathering this research but due to time constraints we haven't been able to prioritize it. I will capture this request in phabricator (T293038) and hopefully we'll be able to get to it soon. NKohli (WMF) (talk) 20:05, 11 October 2021 (UTC)[]
Return to "IP Editing: Privacy Enhancement and Abuse Mitigation/Login Required Experiment" page.