IP Editing: Privacy Enhancement and Abuse Mitigation/IP Editing Restriction Study/Farsi Wikipedia

From Meta, a Wikimedia project coordination wiki

The Persian (Farsi) Wikipedia community decided to block IP editing on the article namespace between October 20, 2021 and April 20, 2022. The Wikimedia Foundation's Product Analytics team tracked the impact of this change on Farsi Wikipedia on a weekly basis in order to learn more about the impact of IP editing on a Wikipedia project.

This is a follow-up to a similar experiment on Portuguese Wikipedia, starting October 2020.

Overview[edit]

The quantitative measurement mainly focuses on the impacts on editors, edits, and administrator actions. This report aims to share the metrics and community feedback that were collected regarding the impact of this change on the project's overall health.

The metrics we captured include:

  • Number of new accounts
  • Number of active user editors
  • Retention rate
  • Number of edits
  • Number of reverts
  • Number of net non-reverted content edits
  • Number of blocks
  • Number of protected pages
  • Checkuser requests

In this report, we will go through the trends of above metrics for the duration of the experiment. We will also share the qualitative data of the community feedback about this change.

Takeaways[edit]

On the outset we should note that the Farsi and Portuguese Wikipedia experiments were very different in their structures. Portuguese community chose to block unregistered editors from all namespaces except the Discussion and Help namespaces. On the other hand, Farsi community chose to block unregistered editors from the Main/Article namespace only. The other important factor to note before reading the metrics is that the pandemic has played out differently in different parts of the world and has made a significant impact on metrics across the board.

Highlights[edit]

We can say that the restriction on Farsi effectively reduced vandalism on the wiki. We can say this based on the fact that reverts were down 68% compared to the previous six months and down 70% compared to same time period last year. Blocks were also down by over 50% in both comparisons. This trend was consistent with the Portuguese Wikipedia experiment.

However, the restriction also prevented good-faith edits. The total number of content edits was down -24% compared to the previous six months. This was a much larger decline than what we saw in Portuguese. On Portuguese, content edits declined -15% over the unusually high number in the previous six months, but it was generally in line with the previous couple of years. Farsi's decline in edits was well below the previous years.

Unlike Portuguese Wikipedia, the partial restriction on Persian Wikipedia did not drive new account creation or active logged-in editors.

A survey of Persian Wikipedia editors found that about two-thirds of the editors surveyed had a positive opinion about the experiment, while the rest felt that it was harmful to the project.

Using the previous 6 months for comparison[edit]

Note that the Portuguese Wikipedia numbers below are computed for the same 6 months after restriction as Farsi Wikipedia and compared to prior 6 months to provide a direct comparison.

Category Metrics Persian Wikipedia Portuguese Wikipedia (for comparison)
Editors Number of new accounts No change (+0.3%) Improved (+9%)
Number of active logged-in editors Fewer editors (-4.5%) Improved (+70%)
Retention rate No change (Diff +0.06%) No change (Diff +0.9%)
Content Number of reverts Fewer reverts (-68%) Fewer reverts (-49%)
Number of net non-reverted content edits Edits declined (-24%);
Matches wikis without IP editing restriction
Edits declined (-15%)
Administration Number of blocks Fewer blocks (-57%) Fewer blocks (-72%)
Number of protected pages Fewer protected pages (-32%) Fewer protected pages (-77%)
Number of checkuser checks More checks (+97%) Fewer checks (-7%)

Using Year-over-Year as comparison[edit]

Note that the Portuguese Wikipedia numbers below are computed for the same 6 months after restriction as Farsi Wikipedia and compared to the previous year to provide a direct comparison.

Category Metrics Persian Wikipedia Portuguese Wikipedia (for comparison)
Editors Number of new accounts Fewer accounts (-18%);
Matches wikis without IP editing restriction
Improved (+23%)
Number of active logged-in editors Fewer editors (-11%) Improved (+87%)
Retention rate No change (Diff -0.26%) No change (Diff +1.1%)
Content Number of reverts Fewer reverts (-70%) Fewer reverts (-45%)
Number of net non-reverted content edits Edits declined (-18%);
Matches wikis without IP editing restriction
Edits declined (-5.4%)
Administration Number of blocks Fewer blocks (-54%) Fewer blocks (-84%)
Number of protected pages Fewer protected pages (-20%) Fewer protected pages (-74%)
Number of checkuser checks More checks (+34%) More checks (+24%)

A more thorough report with an explanation of how each of the metrics below are defined and computed can be found on on MediaWiki.org.

Data analysis[edit]

Methodology[edit]

For each of the following metrics, we compared the data during the 6-months of IP editing restriction to the following:

1. Prior 6 months in the same year (21st April 2021 to 19th October 2021)

2. Same months in prior 3 years

3. Same months at similar wikis (selection method outlined in detailed report): Arabic Wikipedia, Azerbaijani Wikipedia, Central Kurdish Wikipedia, Hebrew Wikipedia, Persian Wikisource and Persian Wiktionary

Comparison to similar wikis allows us to factor in cultural, political and other changes which might have caused changes to editing patterns.

Number of new accounts created[edit]

Hypothesis: New accounts would increase when IP editing is restricted as users who want to make an edit would sign up to be able to do so.

Outcome: Number of new accounts created did not go up and instead remained relatively flat compared to the previous 6-month period. However, in a Year-over-Year analysis, we saw that the number of new accounts created dropped by about 18%. It is possible that this drop in new accounts is unrelated to the IP Editing restriction itself and is influenced by other external factors. It is worth noting that other wikis in the region (which did not turn off IP Editing) faced similar drops in account creations: Arabic Wikipedia (-21%), Azerbaijani Wikipedia (-19%), Hebrew Wikipedia (-40%) and Farsi Wikisource (-20%).

More details here

Monthly new accounts
New Accounts Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 45,790 45,912 1 1.21 1.12 1.17 0.81 1.03 1.04
1 year prior (20/21) 51,448 55,686 1.08
2 years prior (19/20) 45,107 50,795 1.13
3 years prior (18/19) 42,160 47,189 1.12

Number of active logged-in editors[edit]

Hypothesis: Number of active logged-in editors would increase, driven by a surge in account creation.

Outcome: We did not see any increase in logged-in editors (correlated with no surge in new account creations).

More details here

Monthly active logged-in editors
Active Logged-in Editors Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 83,310 79,525 0.95 0.95 1.08 0.88 1 0.99 1.23
1 year prior (20/21) 78,538 89,573 1.14
2 years prior (19/20) 66,827 69,569 1.04
3 years prior (18/19) 54,926 65,154 1.19

Retention rate[edit]

Outcome: Retention rate peaked at 9% in November 2021. The retention of the newcomers joined in the first month of restriction is historically high. The 6 month average of the retention rate during restriction is 0.06 percentage points more than the prior 6 months, and 0.26 percentage points fewer than the same period last year.

More details here

Retention Rate
Retention Rate in Avg Persian Wikipedia
year Apr - Sept Oct - Mar 6Mo6M Diff
Current year (21/22) 7.21% 7.27% 0.06%
1 year prior (20/21) 6.19% 7.53% 1.34%
2 years prior (19/20) 6.15% 6.21% 0.06%
3 years prior (18/19) 5.89% 6.38% 0.49%

Total number of edits[edit]

Hypothesis: Total number of edits is expected to decline with a restriction in IP editing.

Outcome: The total number of edits (including bot edits, reverted edits and deleted edits) have decreased since the restriction was enabled. The 6 month comparison shows that total edits declined 29% during the restriction.

More details here

Monthly Total Edits

Data[edit]

Edits Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 1,575,165 1,120,883 0.71 1.03 0.84 1.13 1.12 1.05 1.36
1 year prior (20/21) 1,358,544 1,659,739 1.22
2 years prior (19/20) 1,480,958 1,407,900 0.95
3 years prior (18/19) 1,624,038 1,256,574 0.77

Reverted edits[edit]

Hypothesis: Number of reverted edits is expected to decrease as fewer drive-by vandalism edits get through.

Outcome: Number of reverts on content pages decreased during the time when IP editing was restricted. The 6 month comparison shows that the number of content reverts is 70% less than the prior 6 months.

More details here

Monthly Reverted Edits

Data[edit]

Reverted Edits Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 186,841 60,479 0.32 0.89 1.32 0.8 1.33 0.68 0.94
1 year prior (20/21) 138,073 202,461 1.47
2 years prior (19/20) 115,745 143,937 1.24
3 years prior (18/19) 115,351 106,696 0.92

Net non-reverted content edits[edit]

Hypothesis: The net non-reverted content edits is expected to decline after the restriction in IP editing.

Outcome: Net edits on content pages contributed 70% of total net edits. There was an observed decline in net non-reverted edits once the restriction was enabled. The net non-reverted edits on content pages is 24% less in second half of the current year(21/22) than in the first half of the year. In a Year-over-Year analysis, the number dropped by 18%. It is worth noting that there was a similar drop in net non-reverted content edits observed on other wikis in the region which did not turn off IP editing: Arabic Wikipedia (-23%), Azerbaijani Wikipedia (-32%), Central Kurdish Wikipedia (-27%) and Hebrew Wikipedia (-29%).

More details here

Net non-reverted content edits

Data[edit]

Net non-reverted content edits Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 657,228 498,571 (adjusted) 0.76 0.89 0.92 0.63 0.93 3.75 1.65
1 year prior (20/21) 553,994 611,259 1.1
2 years prior (19/20) 452,333 497,591 1.1
3 years prior (18/19) 419,097 467,074 1.11

Number of blocks[edit]

Hypothesis: The number of blocks will decline once the IP Editing restriction goes into effect as the amount of drive-by vandalism will be reduced.

Outcome: The number of blocks on IPs and IP ranges saw a huge reduction in the restriction period.

Note: There were a number of blocks carried out by bots which has been excluded from this analysis as those bots were copying over blocks made on other projects.

More details here

Excluding blocks by bot admins, the number of blocks decreased 57% 6MoM and decreased 54% YoY.

Monthly Blocks (Excluding Blocks by Bot Admins)


Blocks excluding bot admins Persian Wikipedia
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio
Current year (21/22) 9,504 4,125 0.43
1 year prior (20/21) 4,995 8,968 1.8
2 years prior (19/20) 7,623 4,719 0.62
3 years prior (18/19) 3,555 15,257 4.29

Number of protected pages[edit]

Hypothesis: The number of page protections will decline once the IP Editing restriction goes into effect as the amount of drive-by vandalism will be reduced.

Outcome: The number of page protections declined during the course of the restriction.

Note: This metric spiked in October 2021, in which 5,867 pages were protected by one user. Per a community member’s request, those protections were excluded from the total of protected pages in the first half of the current year (21/22). After the adjustment, compared to the prior 6 months, the number of protected pages decreased 32%.

More details here

Monthly Protected Pages

Data[edit]

Protected pages Persian Wikipedia Other Wiki Projects 6Mo6M Ratio
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio Arabic Wikipedia Azerbaijani Wikipedia Central Kurdish Wikipedia Hebrew Wikipedia Persian Wikisource Persian Wiktionary
Current year (21/22) 3,487 (adjusted) 2,372 0.68 0.95 0.97 1.09 1 - 1.5
1 year prior (20/21) 2,874 2,964 1.03
2 years prior (19/20) 3,542 2,739 0.77
3 years prior (18/19) 1,452 1,828 1.26

Number of checkuser checks[edit]

Outcome: The number of checkuser requests made during the IP editing restriction saw no significant change and stayed close to prior levels.

Checkuser Checks

More details here

Checkuser checks Persian Wikipedia
year Apr 21 - Oct 19 Oct 20 - Apr 20 6Mo6M Ratio
Current year (21/22) 389 765 1.97
1 year prior (20/21) 535 570 1.07
2 year prior (19/20) 623 564 0.91
3 year prior (18/19) 234 287 1.23

Community survey responses[edit]

We conducted surveys during the study and we had 24 respondents. 62.5% (15) people had a positive view of the restriction and 37.5% (9) of them had a negative view of the restriction. We have shared some of them below.

Positive[edit]

Please pardon any text that is incorrectly translated. The translations were made through Google Translate for speediness.

Some editors who had positive opinions about the restriction

[fa]
تأثیر مثبت داشته است، چون اکثر IP ها متأسفانه مفید نیستند.باعث کمتر شدن بار نگهبانان شده است

 — Editor 1, Community Survey


It has had a positive effect, as most IPs are unfortunately not useful. It has reduced the burden of guards

 — Editor 1, Community Survey

[fa]
باعث کمتر شدن خرابکاری ها شده و محتوای مقالات قابل اعتمادتر شده‌اند. جنگ ویرایشی از طریق خارج شدن از حساب و ویرایش با آی‌پی دشوارتر شده که خود باعث می‌شود جامعه‌ای سالم را شاهد باشیم.

 — Editor 2, Community Survey


It has reduced vandalism and the content of articles has become more reliable. The editing war has become more difficult by logging out of the account and editing with IP, which makes us see a healthy community.

 — Editor 2, Community Survey

[fa]
کاهش ویرایش های غیرمفید چون در ویکی پدیای فارسی بیشتر ویرایش های آی پی در راستای گسترش محتوا نیست. کاهش نیاز به گشت زنی.

 — Editor 3, Community Survey


Reducing useless edits because in Persian Wikipedia, most of the IP edits are not in line with content expansion. Reduced need for patrolling.

 — Editor 3, Community Survey

[fa]
میزان خرابکاری در مقالات پایین آمده است. تاثیر مضری قابل رویت نیست.

 — Editor 4, Community Survey


The rate of vandalism in articles has decreased.

 — Editor 4, Community Survey

[fa]
میزان خرابکاری ها به شدت کم شده است و به نظرم اگر برای فضاهای نام دیگر نیز بسته شود عالی است.

 — Editor 5, Community Survey


The amount of vandalism has decreased drastically and I think it would be great if it is closed for other namespaces as well.

 — Editor 5, Community Survey

[fa]
خرابکاران قابل پیگیری شدن . امکان گشت زنی بهتر گردیده، ویکیپدیای فارسی ظرفیت گشت زنی ویرایش های آی پی خود را ندارد.

 — Editor 6, Community Survey


Vandals can be tracked. The possibility of patrolling has been improved, Persian Wikipedia does not have the capacity to patrol its IP edits.

 — Editor 6, Community Survey


[en]
I think blocking IPs has improved the security performance of Wiki Fa. I think with the reduction of the need for patrolling, the opportunity to pay attention to the main pages has increased.

 — Editor 7, MediaWiki

[fa]
به کاهش چشمگیر خرابکاری ها منجر شده است. بر این اساس موافق با استمرار محدودیت آی پی‌ها هستم. درک من آن است که محدودیت آی‌پی‌ها به کاهش داوطب‌های ویکی منجر نشده است و اعمال محدودیت بر آی‌پی‌ها صرفا با کاهش خرابکاری‌های منجر شده است.

 — Editor 8, Community Survey


It has led to a significant reduction in vandalism. Based on this, I agree with the continuation of IP restrictions. My understanding is that the restriction of IPs has not led to a decrease in wiki traffic, and the restriction of IPs has only resulted in a decrease in vandalism.

 — Editor 8, Community Survey

[en]
The volume of vandalizing and disruptive edits was much less during this period, and those users who fight vandalism were less burned out. The data reported by the Dashboard that WMF created suggests that despite the IP edit restriction, the volume of IP edits in talk pages did not increase, nor did account creations. This further strengthens the idea that this restriction really reduced the volume of malicious edits by uncommitted users, without impacting the role of committed users in adding useful content to the wiki. The community more or less took it for granted. Which is sad.

 — Editor 9, Community Survey

[fa]
این ممنوعیت کار درستی بود. با توجه به فضای وپفا بیش از ۹۰ درصد ویرایش‌های آی‌پی ها برای خرابکاری و دورزدن قطع دسترسی است و انرژی صرف‌شده برای خنثی‌کردن آن‌ها از فعالیت مفید دانشنامه کاسته و برخی ویرایش‌های منفی برای سالیان در صفحات باقی می‌مانند و به اعتبار دانشنامه لطمه وارد می‌کنند. تاثیر قطع دسترسی از منظر بنده مثبت بود و وقت مفید کاربران و نگهبانان و مدیران سامانه صرف ویرایش مفید محتوا شده‌است.

 — Editor 10, Community Survey


This ban was the right thing to do. Considering the space of WPF, more than 90% of IP edits are for vandalism and evasion, cutting off access, and the energy spent to neutralize them reduces the useful activity of the encyclopedia, and some negative edits remain on the pages for years and harm the credibility of the encyclopedia. From my point of view, the effect of cutting off access was positive and the useful time of users, guards and system administrators has been spent on useful content editing.

 — Editor 10, Community Survey

[fa]
۸۰٪ خرابکاری ها کاهش یافته بودند و مقالات تازه داشتند از خرابکاری های به جا مانده از آیپی ها پاکسازی می‌شدند. آرامش به ویکی فارسی بازگشت

 — Editor 11, Community Survey


80% of the vandalism was reduced and new articles were being cleaned from the vandalism left by the IPs. Peace returned to Farsi Wiki.

 — Editor 11, Community Survey

Negative[edit]

Some editors who had negative opinions about the restriction

[en]
It has had a destructive impact on Wikipedia. It has denied the readers who may not be users the opportunity to easily and quickly fix its errors, and has resulted in a less up-to-date and prosperous project.

It's one of Wikipedia's advantages over other sources that enables quick, accumulation of knowledge from anyone who wants to contribute. This limitation vastly hurts this advantage.

The privilege to edit anonymously is another one that is undermined by this. You should not be forced to open an account, pick a user name, and give your email to be able to edit in a project that proclaims itself as an open wiki.

A highly negative impact. It has slowed the stream of new talent from entering the project. It has negatively impacted the notion of Wikipedia as an open encyclopedia "everyone" can edit. It is against the Wiki philosophy and hurts the perception of Wikipedia among readers. It implies the suspicion that Wikipedia is run and edited by a monopolist, exclusionary gang of writers who do not welcome outsiders, and do not foster an open and truly neutral environment for all voices to edit regardless of their views as long as they abide by the rules.

 — Editor 1, MediaWiki


[fa]
شخصا باعث شده بخش زیادی از فعالیت‌هایم در ویکی تحت تاثیر قرار بگیرد. برای ویرایش یک مقاله مدام باید فیلترشکن را فعال/غیرفعال کنم و به همین دلیل حتی اگر ویرایش نگارشی ساده‌ای هم وجود داشته باشد باید موکول کنم به زمانی که فیلترشکن خوبی دارم.

 — Editor 2, Community Survey


Personally, it has affected a lot of my activities on Wiki. To edit an article, I have to constantly enable/disable the VPN anti-filter, and for this reason, even if there is a simple editorial edit, I have to postpone it until I have a good filter breaker.

 — Editor 2, Community Survey


[fa]
زحمت مورد نیاز برای واگردانی خرابکاری ها کاهش پیدا کرده است و اطلاعات غلط کمتری وارد مقالات می‌شود. از نظر آیندهٔ ویکی‌پدیای فارسی، جذب کاربران جدید، و حفظ دیدگاه بی‌طرف در مقاله‌ها نگران‌کننده است.

 — Editor 3, Community Survey


The effort required to undo the vandalism is reduced and less misinformation enters the articles. In terms of the future of Persian Wikipedia, attracting new users and maintaining a neutral point of view in the articles is a concern.

 — Editor 3, Community Survey


[fa]
کاهش خرابکاری در محتوای مقالات. در مقابل تضعیف قسمت نگهبانی از محتوای ویکی‌پدیای فارسی، خرابکاری کاربران گمنام باعث نیازمندی به نگهبانی و تربیت کاربران گشتزن میشود، که در ادامه و با کسب تجربه به ویکی‌بانان و مدیران اضافه میشوند. به بیان دیگر مشارکت کاربران گمنام جزیی از حلقه مدیریتی سامانه است که محدودیت در آن این حلقه را عقیم میکند. بعد از محدودیت شش ماهه کاهش محسوسی در نگهبانان و به دنبال آن ویکی‌بانان و مدیران بوجود آمده است و این به مرور زمان سیستم مدیریتی موجود را خسته و فرسوده میکند. اگر منظور جامعه بیرونی ویکی‌پدیای فارسی است، داده و اطلاعات خاصی در این مورد موجود نیست تا قابل تحلیل باشد.

 — Editor 4, Community Survey


Reducing vandalism in the content of articles. In contrast to the weakening of the guarding part of the Farsi Wikipedia content, vandalism by anonymous users causes the need to guard and train patrol users, who will be added to wikikeepers and administrators as they gain experience. In other words, the participation of anonymous users is a part of the management loop of the system, the limitation of which makes this loop sterile. After the six-month limit, there has been a noticeable decrease in the number of guardians, followed by wikikeepers and administrators, and this over time makes the existing management system tired and worn out. If it means the external community of Persian Wikipedia, there is no specific data and information available in this case to be analyzed.

 — Editor 4, Community Survey


[en]
It harmed Persian Wikipedia. Number of edits got reduced, number of active users went down. It is not accessible. It harms the community in the long-term, they might not notice right away but they will get smaller and smaller later.

 — Editor 5, Community Survey