Research:Wikipedia Administrator Recruitment, Retention, and Attrition/Report

Administrators (or “admins”) are users with extra rights who do work beyond editing, such as settling disputes and preventing repeated vandalism.^[1] Across all Wikipedia editions, there are more than 3,500 administrators.^[2] Tens of thousands (sometimes hundreds of thousands) of administrative actions are performed each month.^[3]

Since 2008, there have been reduced editing rates on Wikipedia and only recently have we turned our attention to administrators as a distinct body of editors. English Wikipedia in particular has been concerned about administrator attrition for a long time, having published multiple Signpost articles about it.^[4] We know that administrators are vital to the healthy operation of any language version of Wikipedia.

This report aims to close this knowledge gap; this project investigated patterns of administrator retention and attrition and explored reasons administrators come, stay, and leave the role. We obtained baseline metrics of admin in/outflow and activity with unified definitions, and conducted a mixed method, cross-comparative study of potential, current, and former administrators, allowing us to produce a comprehensive study of administrator experiences across three major milestones of the administrator lifecycle. Our research included multiple components: quantitative analyses of admins and admin activity across 21 shortlisted Wikipedias;^[a] surveys of more than 2,000 current and potential admins across 6 Wikipedias; and interviews with 12 former admins across 5 Wikipedias.

File:(Final Report) Administrator recruitment, retention, & attrition (SDS1.2.2).pdf

How to read this report? For a quick read, we suggest reading the Section 1 ‘Key Findings’ headlines and short explanatory text, and then proceeding to Section 2 ‘Recommendations’. For a deeper understanding, the reader will find that the Key Findings section text contains links to corresponding sections in our full report, available on Commons.^[5] This is where significant expansions can be found, including additional analyses, figures, and tables.

For this abbreviated on-wiki key findings and recommendations report, we also use plain-text references to sections in the full report, where certain key findings or recommendations are further expanded.

1 Key findings

Please see background, methods, limitations and definitions for more information about how we arrived at these findings, including how we defined key terms such as administrator. You can also read more about what a Request for Adminship (RFA) is and how de-sysopping (removing admin rights) works. (See § 3.3.1, 3.3.2 of full report)

1.1 General state of administrators on Wikipedia

1.1.1 The number of monthly active administrators on large Wikipedias has been declining since 2018, with some exceptions. For smaller Wikipedias, administrator numbers are mostly stable or increasing since 2018.

Declining adminship may be more of a “large Wikipedia problem.”^[b]

From the sampled group of 21 (larger) language versions (see selection rationale):

over half show declining numbers of unique monthly active administrators since 2018 (e.g. English, Russian, Portuguese)
a third show stable numbers (e.g. Swedish, Czech, Norwegian Bokmål)
a few show rising numbers (e.g. Italian, Indonesian, Ukrainian)

This confirms past reports of declining active administrators on projects such as English Wikipedia. It also points to some notable counterpoint trends. In order to understand admin patterns in the broader context, we also examined 234 additional Wikipedias.^[c] This analysis shows that all but one (i.e., Telugu, which has decreasing admin numbers) have had increasing or stable admin activity since January 2018. (See § 3.4.1 of full report)

1.1.2 Admin inflow has been decreasing yearly since 2018 for most of the shortlisted 21 Wikipedias.

Though many Wikipedias show admin outflow to be decreasing yearly as well, the outflow has still eclipsed inflow, resulting in decreasing yearly net change. Exceptions include Italian (has had steady inflow), as well as Catalan, Ukrainian, and Indonesian, which saw inflow spikes between 2020-2022. Unsurprisingly, these four Wikipedias have also had an increasing number of monthly active admins since 2018. (See § 3.4.3 of full report)

1.1.3 Yearly admin net change (i.e., admin inflow minus admin outflow) varies across the 21 shortlisted Wikipedias, and the majority of Wikipedias show the biggest period of inflow between 2007 and 2015.

Among various Wikipedia editions since 2018, yearly net change has been:

Mostly positive: e.g., Persian, Indonesian, Ukrainian
Relatively stable (i.e., either consistently close to zero or fluctuating evenly): e.g. Norwegian Bokwal, Polish
Mostly negative: e.g., English, Spanish, Portuguese

During which periods are Wikipedias experiencing the highest inflow? (See § 3.4.4 of full report)

2007 - 2015: for majority of Wikipedias, including English, French, Spanish, Russian, and Polish
2013 - 2018: for a few Wikipedias, including Persian and Portuguese
2019 - present: Wikipedias such as Indonesian and Ukrainian

1.1.4 Admins on the shortlist of 21 Wikipedias are less likely than admins of other (smaller) Wikipedias to be a sysop on more than one Wikimedia project.

From the shortlisted group of 21 Wikipedias, there are currently 2,063 members of the administrator user group; of those, 174 (8%) have admin status on more than one Wikimedia project.^[6]^[d] In contrast, when we look at population data across all Wikipedias, we find that there are currently 4,063 members of the sysop user group; of those, 661 (16%) are a sysop on more than one Wikimedia project.^[e] (See § 3.4.5 of full report)

1.1.5 Administrative actions and responsibilities are dispersed or concentrated differently across Wikipedias.

The basic assumption that ‘administrators do administrative work’ is much more nuanced and complicated in reality. Not all administrative rights are associated solely with the sysop user group, and therefore some user groups other than sysop can perform administrative actions. Some Wikipedias (and other Wikimedia projects) allow other user groups to have some of these rights in addition to sysops; on Russian, Japanese, and Persian Wikipedias, for example, about a third of users performing administrator actions are not in the sysop user group.

This points to some degree of dispersal within these Wikipedias, where administrative responsibilities (and therefore, the user rights needed to take on those responsibilities) normally associated with sysops, are extended to other user groups. This has strong implications for approaches to lighten administrator workloads, as well as research into administrative workflows. (See § 3.4.2 of full report)

1.1.6 The priorities ranked highest by English Wikipedia survey respondents include recruiting new administrators and support in conflict, with policies for common conflicts and automation nearly equally ranked as a third priority.

There’s remarkably little variation in the priorities of English Wikipedia admins, aside from less interest in providing admins the ability to focus on specific tasks. (See § 3.4.6 of full report)

1.2 Administrator recruitment

1.2.1 Administrator candidacy guidelines vary substantially across language versions of Wikipedia.

This summary of administrator candidacy requirements for the 21 Wikipedias studied as part of this project shows that candidacy guidelines based on editing history vary substantially. For example, account age guidelines range from none stated, to “a few months”, to a full 2 years. Factoring in both formal and informal guidelines, main namespace edit guidelines for administrator candidacy similarly range from none stated, to 100, to 10,000. Language versions may also list additional formal or informal guidelines, such as membership in a user group (e.g., patroller, rollbacker, autopatrolled), presence of an email address, or evidence of participation in anti-vandalism work. (See § 3.5.1 of full report)

1.2.2 While multiple Wikipedias seem to recognize significant declines in administrator inflow, this does not always produce speedy interventions aimed at increasing administrator inflow.

Spanish and English Wikipedias have both long acknowledged major issues in administrator inflow; English Wikipedia has a decade-long history of Signpost articles to that effect.^[4] However, major reforms to the English Wikipedia RFA process, and administrator role removal policy, have only come about in 2024.^[7] Spanish Wikipedia, to our knowledge, has not significantly changed aspects of its RFA or de-sysopping process, despite single-digit RFA successes in the last six years.

1.2.3 The RFA process is routinely characterized by administrators as stressful, opaque, and something to be endured.

Interview and survey respondents from across all surveyed Wikipedias consistently characterise the RFA process as highly stressful and difficult to understand. Beyond formal written standards, which can vary significantly across Wikipedia editions, every Wikipedia that we studied has unwritten rules that are critical to determining a candidate’s success. These informal expectations are related to a candidate’s social recognition and standing with the existing group of administrators or eligible RFA voters. Survey results show that English Wikipedia administrators regard the unofficial RFA requirements as the biggest barrier to a successful adminship candidacy. They often necessitate finding an existing administrator in good standing as a nominator for the RFA process, maintaining a positive reputation among the existing administrator body, and enduring intense scrutiny of one’s entire on-wiki history. (See § 3.5.2 of full report)

1.2.4 Current English Wikipedia administrators say it is too difficult to become an administrator; more think the RFA process is unfair than fair.

A majority of English Wikipedia admins surveyed believe that “It is too difficult to become an administrator on English Wikipedia” (57%; n=184), as compared with 6% who believe it is too easy. Similarly, 36% (n=118) of current English Wikipedia admins surveyed consider the RFA process to be unfair, 25% (n=82) consider the process fair, 22% (n=70) are neutral on the question, and 16% (n=52) are unsure.

1.2.5 Most potential English Wikipedia administrators say they are familiar with adminship; fewer are familiar with formal RFA requirements or the RFA process.

Most (68%) potential English Wikipedia admins say they are moderately (29%, n=177) or very familiar (39%, n=237) with the role of admins on English Wikipedia. Only a comparative few say they are not at all (3%, n=18) or slightly familiar (11%, n=68). About half (49%) of potential English Wikipedia admins are moderately (22%, n=136) or very familiar (27%, n=163) with the formal requirements for an editor to become an English Wikipedia admin. Conversely, about one-third (33%) say they are not at all familiar (18%) or only slightly familiar (16%) with these requirements. Potential English Wikipedia admins are relatively less familiar with the RFA process. Twenty-seven percent (n=164) say they are very familiar with the process, 19% say they are moderately familiar with it, while about two of five (38%) say they are either not at all familiar (21%, n=130) or slightly familiar (17%, n=104) with the RFA process.

1.2.6 Interest in adminship is particularly low among potential administrators on English and Polish Wikipedias, but appears stronger among those from other Wikipedias.

Overall, interest from potential admins in becoming an administrator varies across Wikipedia language editions. Interest is particularly low among potential admins on English and Polish Wikipedias, with over 60% and 58% respectively expressing little to no interest. In contrast, about half of potential admins on Spanish, French, and Russian Wikipedias are interested to some degree in becoming an admin on their respective Wikipedias. The potential admin sample from Indonesian Wikipedia is small (n=14 answering this question), but only one respondent reported no interest. (See § 3.5.3 of full report)

1.2.7 Potential administrators are most likely to cite a desire to focus on their work as an editor as a reason for lack of interest in adminship.

In all six surveyed Wikipedias, potential admins were most likely to cite their desire to focus on editing as a reason why they were not interested in becoming an admin. The second most common reason was viewing admin work as too time-consuming (English, Spanish, Indonesian, Polish, Russian) or the potential for conflicts with editors (French). These three reasons were very prevalent across all six Wikipedias. However, only 19% of English Wikipedia potential admins cited fear of harassment as reason for their lack of interest in adminship. Comparatively few potential English Wikipedia admins cited any aspect of the RFA process as a reason for a lack of interest in adminship.

1.2.8 Potential admins interested in adminship most frequently cite a desire to improve and support Wikipedia; few cite friendship with current admins as a reason for interest.

Among potential English Wikipedia admins who are at least “moderately” interested in adminship (n=193), the overwhelming majority (86%, n=166) cited an opportunity to help improve English Wikipedia as a reason, followed by a belief that admin tasks are essential (72%, n=139). Although half of current English Wikipedia admins said they had a friend or mentor who was an admin prior to pursuing adminship, few potential admins (7%, n=13) cited this as a reason for doing so. Potential admins on French, Polish, and Russian Wikipedias are also most likely to cite improving Wikipedia as the top reason for pursuing adminship.

Potential Spanish Wikipedia admins are relatively less likely to cite improving Wikipedia (31%, n=68) compared to admin tasks being essential (45%, n=98) and mentoring other editors (41%, n=89). Potential Polish Wikipedia admins are most likely to cite improving Wikipedia (78%, n=56), but are relatively less likely to cite admin tasks being essential (42%, n=30). Potential Polish Wikipedia admins are also relatively more likely to cite “intrinsic” motivations such as adminship being the next step on their editor journey (56%, n=40) and adminship being a validation of their accomplishments as Wikipedians (35%, n=25). Our sample of potential Indonesian Wikipedia admins was very small for this question (n=7), but these respondents were least likely to cite friendship with current admins (n=2) as a reason to pursue adminship.

1.2.9 Current and potential admins strongly resemble each other in aspects of gender, primary spoken language, and education level.

Survey results suggest that potential and current administrators share key demographic characteristics. A comparison of gender identification among potential and current administrators suggests that gender-based participation disparities may be both large and pervasive among admins of Wikimedia projects. Similarly, potential and current admins are overwhelmingly likely to say that the project language is their primary or native language. Both groups are also highly-educated. One differentiating demographic factor was age. Current admins tend to be older than potential admins, and younger cohorts (such as 18-29) may be underrepresented in current admin groups.

Many of these demographic traits mirror those of Wikipedia readers and editors, such as the gender gap^[8]^[9] and in education.^[10]^[11] These disparities may be downstream of the same demographic disparities among Wikimedia contributor communities; that is, gender gaps among current Wikipedia admin respondents in the projects we studied do not appear systematically greater than those present in the corresponding potential admin respondents. (See § 3.5.5 of full report)

1.2.10 Current administrators may be longer-tenured and older than potential administrators.

When we compare current admin respondents to potential admins across each of the surveyed projects, it becomes clear that admins are disproportionately likely to be long-serving editors and conversely, proportionally less likely to be newer editors. For example, current English Wikipedia admins are nearly three times as likely to have first started editing in 2001-2004 when compared to potential English Wikipedia admins (33% of current admins vs. 12% of potential admins). This disparity appears even starker in other surveyed projects (57% of Spanish Wikipedia admins began editing in 2005-2008 vs. 17% of potential Spanish Wikipedia admins). (See § 3.5.6 of full report)

1.2.11 More on factors positively and negatively affecting recruitment.

In our full reporting section, you can read more about factors positively affecting potential administrator recruitment, and factors negatively affecting potential administrator recruitment. (See § 3.5.4, 3.5.5 of full report)

1.3 Administrator retention

1.3.1 Administrators are motivated by a strong sense of purpose and are proud of the work they do on Wikipedia.

Participating administrators in this study identify a sense of purpose in their work as their primary motivation. Many interview participants expressed that they view their role as a responsibility and feel a strong sense of duty to fulfill it. Very few interviewees reported negative sentiments related to their time as an administrator, even among those who had no interest in pursuing adminship again.

“It’s fun, and I’m contributing to something great.” —Former Administrator, English Wikipedia

The survey data echoes sentiments from our interviews with current and former administrators. Eighty-five percent (n=261) English Wikipedia admins are proud of the work they do (42% strongly agree; n=129). Virtually none (2%; n=5) disagree. Most English Wikipedia admins (54% total agree; n=167) agree that the broader English Wikipedia community appreciates their work as an admin.

1.3.2 Most administrators see themselves staying for the next two years.

According to the survey, an overwhelming majority of English Wikipedia admins expect to still be serving as admins two years from now, while only 3% disagree or strongly disagree. Seventy-one percent (n=218) of English Wikipedia admins say they “never” consider quitting their roles, and English Wikipedia admins are more likely to say they have considered reducing or pausing their work as admins, than quitting.

Across all other surveyed Wikipedias, clear majorities anticipate staying over then next two years; however, compared to English Wikipedia, there is more variance across responses of admins for other Wikipedias, suggesting comparatively more uncertainty about the future of their role. (See § 3.6.3 of full report)

1.3.3 Admin activity is largely performed by a small subset of highly-active, human moderators. Many Wikipedias utilize bots with administrative rights to sporadically handle one-off, high-volume administrative tasks.

Each month, the majority of administrative actions are generally done by humans. For most of the 21 shortlisted Wikipedias—with few exceptions—each wikipedia’s top 15% most active admins performed the majority of administrative actions each month, over the past year. Bot-performed admin actions spike in non-regular intervals, which suggests we should understand bot-performed admin actions as primarily an event-based or for-purpose phenomenon. These one-off events are accompanied by (some) Wikipedias having low “ambient” levels of bot-performed admin actions, from tools such as AbuseFilter or custom anti-vandalism bots. ^[f] (See § 3.6.4 of full report)

1.3.4 English Wikipedia has the highest percentage of administratively inactive admins^[g] over a 30 day period. ‘Inactive’ admins appear less common on other Wikipedias.

A substantial proportion (38%) of current English Wikipedia admins self-report as not having conducted any admin actions in the past 30 days. Queried data suggests about 60% of English Wikipedia admins have been administratively inactive during the month of November 2024.

Survey data suggests that administratively inactive admins appear to be somewhat less common for Spanish, French, Indonesian, Polish, and Russian Wikipedias. This aligns with patterns in our queried data, suggesting that inactivity among administrators may be a more prominent issue on some Wikipedias, such as English Wikipedia, than others.

It is important to consider that there may be seasonal variations that influence both the survey and queried data, as the timeframe for activity was limited to “over the past 30 days” in the survey.

1.3.5 Experiencing some interpersonal conflict as a part of their work as an admin appears nearly inevitable for English Wikipedia admins. Moreover, many have occasional or frequent experiences with interpersonal conflict that could result in potentially serious harm.

More than 99% of English Wikipedia admins say they have experienced some interpersonal conflict due to their work as admins and they report a relatively high incidence of conflict with editors. In contrast, they are less likely to experience interpersonal conflict with other admins. Nearly half have faced abuse or harassment occasionally or frequently. These incidents sometimes involve serious psychological and physical safety issues such as doxxing, personal attacks in both on-wiki and off-wiki virtual spaces, and in-person harassment. (See § 3.6.5 of full report)

1.4 Administrator attrition

1.4.1 Reasons for resigning from administrator roles are multiple and largely individualistic.

Insights into attritional factors were drawn primarily from interviews with former administrators (n=7), who shared personal experiences about why they left their administrative role. Those reasons were:

Inactivity on-wiki, falling short of self-perceived acceptable boundaries for an administrator, often as a result of changing life circumstances such as increased schoolwork or jobs.
Interpersonal conflict with other administrators or experienced editors.
Loss of intrinsic motivation for participation, often as a result of the two reasons above. (See § 3.6.4, 3.7.2 of full report)

1.4.2 Among all the Wikipedia language editions surveyed, positivity is highest and burnout is lowest among English admins.

Across the different Wikipedias surveyed, admins report varying levels of positivity and burnout related to their work. On English Wikipedia, only 9% (n=41) of admins surveyed report being “almost always” or “frequently” emotionally drained by their work, and 52% (n=266) of English Wikipedia admins report feeling positive about their contributions to English Wikipedia. French Wikipedia admins report low burnout, and relatively less frequent positive feelings about their work. Spanish Wikipedia admins report higher burnout levels and less positivity. Indonesian and Polish respondents fall between Spanish and English— reporting less frequent positive feelings and somewhat more frequent feelings of emotional drain. Conversely, Russian admins report high positivity, but also relatively high burnout levels.

1.4.3 Harassment and abuse drive English Wikipedia admins to reduce their roles or quit.

Our interview and survey data both confirm that harassment in all of its forms and conflict take a significant toll on the admin population, leading many to reduce their administrative roles or consider quitting. While fewer consider quitting outright than reducing their time, those who do consider quitting are more likely to have experienced abuse, harassment, doxxing, personal attacks in both on-wiki and off-wiki virtual spaces, and conflicts with other users. Notably, 5 of 7 former admins (across multiple Wikipedias) interviewed cited conflict with other admins as reasons for resigning, and nearly all of the 12 admins interviewed, both current and former, spoke about interactions with other “toxic” users as demotivating forces that impact their work. (See § 3.7.3 of full report)

2 Recommendations

We have grouped our recommendations into three sub-categories, based on their relevancy to key groups.

2.1 Recommendations for the Wikimedia Foundation

2.1.1 If the WMF provides communities with better access to data that can be used to assess the impact of administrator policies on recruitment and activity levels, communities will be better enabled to measure the impact of their policy changes.

One gap uncovered in our study of administrator recruitment and attrition is the lack of effective evaluative mechanisms, and even key basic metrics. Few administrator bodies have the capacity to undertake effective auditing, for current or proposed practices; such efforts often depend on the motivation and availability of a single volunteer. Adoption of policies in administrator circles tends to follow a “do first, fix later” model as a result of this lack of information. Therefore, by providing better, consistent and cross-wiki metrics on administrators, the WMF may be able to provide communities with the data they need to make informed decisions about their own governance. Relatedly, the Foundation can pursue the following, which are important because they can affect general data accuracy:

A central place user groups and admin rights can be tracked and updated, including which user groups have admin rights per Wikimedia project (e.g. these user groups);
Updates to the mediawiki.org pages for logging table, log actions, and other admin-related data documentation;
Proposing a software change (and/or creating a processed dataset or variable) that would help to distinguish between administrative actions made by humans, bots, and AbuseFilter.

2.1.2 If the WMF provides financial and planning support for admin-focused cross-wiki exchanges, the current communities of policy makers will be able to learn from each other’s efforts around admin support and recruitment.

This involves facilitating potential knowledge-sharing between administrator bodies. We know that this already happens in the sense that policies cross-pollinate between projects, but these generally rely on personal connections and social relationships that individual, influential editors make with each other. We propose that the Foundation might be able to facilitate such exchanges through support for regional events (both online and offline) with a specific focus on administrator knowledge-sharing across different Wikipedia projects. There may also be opportunities to support grant-funded work that shares these goals. As this type of work advances, there will likely be opportunities for the Foundation to help develop ‘support packages’ or strategy toolkits that can be offered to communities with needs such as one to grow the number of admins. Finally, there are also opportunities for the Foundation to support communities as they seek to evaluate the effectiveness of their efforts aimed at supporting current admins.

2.1.3 If the WMF improves automation of repetitive tasks performed by administrators, they can more evenly distribute, and even reduce, the admin workload.

The WMF should continue current efforts aimed at automation of admin tasks. The Foundation should also consider taking on the maintenance burden of widely-replicated tools (compare Automoderator’s relationship with third-party anti-vandalism bots).

Relatedly, the WMF Product and Technology department should explore ways of integrating administrator tasks with current editing workflows; by making administrator tasks feel more like editing, admin work in general may be more accessible and appealing to a larger pool of editors.^[h]

2.1.4 If the WMF better understands the nature of conflict faced by administrators, it will be better poised to offer effective, meaningful and equitable support.

Our findings seem contradictory: administrators may report facing severe forms of harassment and abuse, yet such abuse (or the possibility of it happening) does not seem to deter them from continuing in their roles. While there is a positive correlation between experiencing more severe forms of harassment and intentions of quitting, even these admins are still more likely than not to say they wish to continue being an admin. At the same time, our interviews suggest that interpersonal conflict between administrators is simultaneously less likely to be severe (or pose a threat to personal safety), yet is significantly more demotivating to administrators.

In line with research done on effective anti-harassment remedies, we ought to understand this finding in several directions.^[12]^[13]^[14] The ability to report their experiences to a perceived authority such as the WMF, whether by survey or interview, validates administrators’ experiences of harassment. Letting the WMF know that conflict, harassment, and abuse exist is different from expecting support for every potential case of conflict that happens over the course of adminship. Finally, different forms of conflict produce different forms of harm, which in turn may be best remedied by a wide variety of actions; this will require the broad participation of multiple groups within the Movement as a whole.

Empathetic, effective support of admins will likely require deep consideration of the following:

Heuristics of harassment incidents
Decisions about whether to address the issue of conflict and harassment at a systemic level, or at an individualized level in the case of severe harassment or abuse, especially by well-resourced actors (such as states)
The needs of the group being addressed, whether this is individual or at a community level
Resource and time considerations on the types of support that can be provided

2.1.5 The WMF should improve visibility and ease of access to support in cases of major threats to safety.

The WMF should prioritize greater support for admins who are threatened with physical harm, and improve expanded outreach to administrator groups for better visibility of these options.

2.1.6 The WMF should avoid the lure of one-size-fits-all solutions, and acknowledge areas that are not well-suited for WMF efforts.

A recurring theme from our study is that, even among Wikipedias that share several characteristics on paper, the challenges faced by their administrators are very specific and sensitive to local contexts. In addition, there are certain areas where effort is not advisable; for example, focusing on minimizing administrator outflow is not advised. All of our data points towards administrator outflow and attrition linked to factors that are beyond the ability of any given volunteer community to seriously influence (e.g. school, work). On the topic of conflict and adminship, it is unclear if it is either possible to avoid conflict as an administrator, or if such conflict avoidance would even significantly impact administrator outflow. From a values-driven perspective, it makes sense to try to support people targeted for serious abuse, but doing so is unlikely to be a solution to admin outflow.

2.1.7 If the WMF continues efforts to increase representation and diversity of editors, there’s a good chance this will have a downstream effect of increasing representation present among administrators.

While the acquisition of administrator rights comes with its own set of norms that may filter out certain types of individuals, increasing demographic diversity of editors upstream will increase the odds of improving representation among admins. One potential exception may be age—while we do not observe consistent demographic gaps between current and potential administrators by gender, educational attainment, or primary language, younger users (age 18-29) do appear to be generally underrepresented among current administrators across the studied projects relative to the sample of potential administrators. This suggests that there may be some specific challenges to recruiting younger administrators that are distinct from those in attracting younger readers or editors. This is particularly notable given previous research indicating that readers and editors of Wikipedia either match or skew younger than the age distribution of the global population.^[10]^[11]

2.1.8 If, in close coordination with communities, the WMF Product and Technology department explores ways of surfacing the opportunities and benefits of adminship, more potential admins will discover these opportunities earlier and present stronger RFA applications.

Helping editors identify these opportunities earlier, and then learn what editing histories and characteristics are prioritized in RFA processes may also help improve the RFA experience for incoming admin candidates. In part, because understanding what characteristics are emphasized for administrator candidates will help those individuals prepare stronger cases. For example, efforts aimed at making community-building opportunities and locations of on-wiki village pumps (especially if active) more visible for potential admins, may help serve future admins and underrepresented groups.

2.2 Recommendations for community-adopted policies

2.2.1 Distributing (unbundling) administrator rights may help disperse the admin workload while providing more opportunities for potential admins to learn and succeed with admin work.

Several of the Wikipedias we analyzed distribute administrator workloads more broadly across multiple user groups, such as Portuguese Wikipedia allowing rollbackers to block vandals for 24 hours.^[15] The suggestion of distributing (namely unbundling) administrator rights was also explicitly named as a potential improvement by multiple interview participants on multiple Wikipedias. This could potentially have numerous helpful effects, such as:

In the short term, distribute workload for the simplest administrator tasks
Increase the visibility of simple administrator tasks
Longer-term, provide a clear on-ramp to administrator work without requiring individuals to start with the full RFA process
Allows individuals who might not want to “give up” editing or other non-administrator work, to help reduce maintenance burdens without feeling obligated to focus entirely on administrative tasks

Communities would have to decide on which user groups would fit an expansion of user rights, and the subsequent mechanisms for ensuring that they were not misusing these potentially more consequential rights.

2.2.2 Lowering the barriers to de-admining can be explored as an enabling factor to help make the RFA processes less prohibitive.

Interviewees from English Wikipedia generally stated that part of the reason for English Wikipedia RFA’s strictness was because it can be quite difficult to remove admin rights involuntarily. The rationale was that, if it’s difficult to remove admin rights, then the community wants to impose very high standards on incoming admins. If it were less difficult to remove admin rights when appropriate, the bar to entry for admins could – as a result – be lowered slightly, allowing an increased inflow and opportunity for more potential admins to try out this type of work.

2.2.3 Consider adoption of programming and policy that has been used elsewhere to support current administrators and increase inflow of new administrators

There are examples of support efforts that can serve as inspiration for thinking about types of support that admin communities may want to consider a model for their own efforts as appropriate. Multiple communities have identified the opportunity to support current admins in various ways. There are various examples of support programs for admins, including on Polish, German, and French Wikipedias. For example, in 2020, Wikimedia France set up a psychological support and assistance program, and in 2021 decided to continue this experiment by offering a broader and more comprehensive service to all its stakeholders.^[16] This idea is supported via interviews from the current project with English admins as they discussed the need for more community support to encourage collaboration and camaraderie among admins to help foster support and kindness.

2.2.4 Establishing better means of support for administrator candidates can help reduce the frequent extreme stresses of the RFA process and improve the number of successful candidates.

Administrators relayed to us that emotional and mental support through the RFA process was more important than any advice about how to perform in the RFA, per se. Thus, support aimed at coaching potential candidates through the stresses of RFA attempts could increase the rates of successful RFAs, or potentially reduce the amount of voluntary withdrawals. In addition to current efforts aimed at supporting new administrators, such support could also include a proposal to try to match each RFA candidate with a current admin who could serve as a contact point and coach for them. The goal of this contact point person would not be to try and directly influence the result of the RFA, but instead ensure the candidate has basic support throughout the process.

2.3 Recommendations for future research

This project has addressed various foundational questions around the current state of Wikipedia administrators. Nonetheless, there remain a number of open questions, especially as they pertain to research opportunities. We offer a generous (but non-exhaustive) list of them here for other researchers to consider; there are opportunities for:

Evaluative research into the effectiveness of administrator support programs. Any such research should consider close partnership with affiliates or organized on-wiki bodies.
More thorough investigation of the support that administrators would find most effective and desirable.
Development of methodologies for tracking admin activities might change over time (e.g. policy changes).
Development of models for predicting admin retention and/or attrition.
Closer investigation of the relationship between administrator inflow and outflow rates. An important question in this space is whether the difficulty of removing adminship (“de-sysopping”) affects administrator inflow rates in addition to administrator outflow. Relatedly, there is an opportunity to better survey cross-wiki de-sysopping processes.
Understanding what impacts, if any, the 2024 English Wikipedia RFA and de-sysopping reforms will have on the administrator body.
Evaluating which policies seem effective at attracting new administrators, both in the short and long term.
A deeper investigation into burnout of admins on-wiki. How prevalent is it? How severe is it? Are communities aware of this as a present problem?
Focused inquiry into the topic of how people not only become admins, but become good admins. What are effective ways for administrators to learn how to use the tools available for them? What ways currently exist? Are they effective? And, moreover, how do they affect knowledge equity, given the highly skewed demographic makeup of administrators?
Comparing metrics on admin activity, retention, RFAs, and overall sentiment between Wikipedias that have administrator “term limits” or re-election procedures (such as German Wikipedia), versus those that allow administrators to retain rights indefinitely by default (such as English Wikipedia)
Using mediawiki/product data: investigate, using baseline counts, admin editing-experience over time in the population data, to compare to what we see in the survey data.
Using production data, in conjunction with structured and unstructured data on RFA pages: investigate what types of users become candidates for RFA, and which candidates are successful
- Look at these trends over time
- Compare the qualitative and quantitative aspects of RFA data, such as voting numbers and comment type or sentiment, etc.

2.4 More on product and policy implications

In our full reporting section, you can read more about product and policy implications. This includes topics related to policy-based influences on recruitment, the clash between the reality and perception of administrator workloads, as well as the lack of direct market economic concerns on attrition. (See § 3.9 of full report)