Jump to content

User:Wugapodes/How volunteers are paid

From Wikipedia, the free encyclopedia
  • Introduction
    • What are human resources?
    • Overview of forms of capital
      • Financial capital: what people typically think of when they hear "capital", i.e. money
      • Cultural capital: embodied practice (habitus) acquired by members of an in-group after a period of enculturation
        • On-wiki this would be things like using standard acronyms (WP:Glossary), proper indentation in threads, and properly employing policies (and their shortcuts).
        • More advanced forms of cultural capital include more obscure parts of wiki syntax like [[link]]ing and using special pages
      • Social capital: acquired personal relationships that create expectations of reciprocity, i.e. owed favors
        • When discussed on-wiki this is often articulated as "trust" and is particularly prevelant in user rights discussions.
    • Jalt and related topics
    • Wikipedia in historical context
      • A brief introduction to historical materialism
      • The productive forces that drive volunteers to create free encyclopedic content on-wiki
      • The relations of production off-wiki which subsidize volunteer labor
        • Early on, these relations privileged white, male, and tech-savy contributors which created an enduring social structure
        • Current relations have led to diversification and challenges to the existing structure on many fronts:
          • Concerted efforts to remedy the gender gap and other systemic biases
          • Attempts to attract editors from underrepresented demographics
          • A collective reconning with how our systems handle harassment and abuse
      • How the community currently handles reports of abuse and harassment
        • Overview and representative examples of failure modes
    • Thesis: Capital largely determines dispute resolution outcomes
  • Elite editors have amassed capital
    • Early adoptors were able to centralize power and capital
      • See above section on "protecting in-group"
      • Admin corps much older than editor corps
      • Inertia and stagnation
    • The "old guard" metes out capital to new users who
      • Produce "good" (in the guard's opinion) content
      • Do not pose a substantial threat to the established order
      • Most obvious in user rights grants
        • The culmination is RfA
  • Dispute resolution system protects the in-group from outsiders (by design)
    • I think meatball's wikilifecycle has related content under something like "barbarian incursion"
    • Those without sufficient cultural capital
      • May not know how to use or find DR
      • May be given less credence due to formatting or atyle
    • Social capital (in the form of trust and deference) can be used to avoid most serious consequences
  • Volunteers are not properly incentivized to handle reports of abuse
    • HR departments receive financial capital to handle reports of abuse
    • Editors who handle reports of abuse can only hope to gain social capital
    • Editors stand to lose capital if they are in the dissent
    • This situation leads to "jalt" and negative outcomes for the project
  • Outcomes are largely determined by capital
    • Early adoption of Wikipedia had material connections to financial capital and the wider mode of production
      • Centralization of social and cultural capital in early adoptors recreates and reproduces external biases and hegemonies
    • Newer users have less social and cultural capital (and given changes in relations of production, likely less financial capital)
      • Older editors can use their capital to WP:BITE newbies with relative impunity (c.f. Fernandez 2019)
    • Lack of capital incentive discourages bystander participation or reports which biases discussion participation towards the editor with greater social capital
      • Number of user talk watchers might be a good proxy
      • Requirement that subjects of AN threads be notified on user talk skews this further
    • Reports are rarely successful if not made by an experienced user as their accumulated capital
      • garners deference through trust
      • can be invested to incentivize others to perform labor to evaluate the merits (c.f. bounties on Stacked change)
      • increases the social cost of retaliation
  • Conclusion
    • Basically all the stuff below
    • Any reasonably fair process for handling conduct issues needs to mitigate the role of capital in determining the outcome



I stated in the request (and still believe) that ArbCom should take any good faith request of administrator misconduct, and strongly oppose remedy 5 and revisions. The community is able to handle administrator misconduct itself, but our options are limited and the costs are high. The community's most powerful meatball:TechnologySolution is the block. This is ham-fisted in many situations, but blocked administrators are still able to use some tools despite the block (c.f. the Fram desysop). If any community restriction on an administrator were violated, we would struggle to adequately enforce them ("yeah badmin violated their TBAN, but I'd have to block them from everything which is a net negative so let me just give another warning"). This brings me to the next point: the costs of enacting a community solution are high. Sanctioning an administrator requires significant social capital in order to marshal community membors to take on the risk of publicly criticizing a prominent community member. Consider the David Gerard incident brought up by PR above; how many people do we think could get away with an expletive-laden tirade? I publicly confronted oppose !voters, and called them out on their bullshit rationalizations. I behaved like a teenager on an internet forum because reasoned argument had gotten us nowhere. The pathos was intentional: I knew sufficient will resided in the community, but it needed to be roused. It is disappointing (but not surprising) that the facts I merely summarized were not sufficient to convince the community, and dangerous that our primary method for dealing with misconduct is so susceptible to passion. We knew what needed to be done, clearly, but the risk of publicly stating it is, in many cases, higher than not doing anything. By nature, I'm contrarian, and am willing to take flak for others if it gets us to the right solution. We should not ask others to take on that same risk, let alone design a performance-review system around it. But functionally that is what remedies 5.X recommend.

Just because the community can deal with administrator issues does not mean that it should. Having two independent processes for dealing with administrator misconduct is a feature, not a bug. It provides reporters with multiple options that they can choose from based on their particular situation and fears. To some degree, this is what remedy 5.2 gets closest to, but it loses touch with reality. RexxS had been given feedback on this---the crux of his RfA was civility concerns, and the promoting crats assumed (given the margin) he would take them on board. Editors had been in touch with him about his conduct, and the evidence page has examples of this. Of note is when valereee and Zero0000 raised concerns about Rexx's conduct on his talk page; not only was it removed without comment the reporters were berated by a third party and accused of raising the concerns in bad faith. Prior to reporting their concerns at ANI, the now-vanished editor had discussed their concerns with Rexx on a talk page where another administrator also raised concerns.

Looking at the evidence, are we to believe that prior resolution methods had not been tried? Was RexxS was unaware of concerns? Would further talk page discussions or AN(I) threads likely result in anything? On what grounds would the answer to any of those questions be yes? Remedy 5.2 writes against a reality that did not happen and advises the community to do things it already did. It resolves nothing, because the fundamental premise behind remedy 5 and revisions is flawed. They are correct in stating that arbitration is in addition to, not instead of, community solutions, but in what realistic circumstances has someone's venue of first resort been WP:RFARB instead of a talk page or noticeboard? None. Because the situation remedies 5.X write against is unrealistic, these remedies function largely as a recommendation that we waste our time, emotional energy, and camaraderie on repeated discussions that historically have gotten us nowhere. Imbalances in power and social capital make sanctioning an administrator by community consensus outrageously difficult and opens the reporter up to further harassment. Sometimes it works fine, and the community is working to improve our success rate. But we should not pretend that our best developed structure for limiting power imbalances in reporting misconduct is off-limits to most situations. It is acceptable to take no action after accepting a case, but remedy 5.0 in particular suggests accepting a case is tantamount to a presumption of guilt and establishes an undefined threshold of "desysop-worthy". I have said, and I maintain that the committee should take all good faith administrator conduct cases, because doing otherwise harms our community by disincentivizing reporting and prolonging conflicts creating a toxic environment. The committee rightly acknowledges that it should not take every case that it can; now extend that same wisdom to the community.


HJ Mitchell has a good response below that I think warrants a clarification. His point about the social capital needed to file a request for arbitration is well taken---he is correct that at present the social capital required to mobilize editors in support of a case request is high---but that is precisely why I believe ArbCom should take all good faith cases regarding administrator misconduct: we should not pretend that our best developed structure for limiting power imbalances in reporting misconduct is off-limits to most situations. By lowering the barrier to entry, we decrease the chance that misconduct goes unaddressed for long periods of time.

Harry is further correct that investigations take time and energy, and in fact this is the crux of my social capital analysis. Social capital is not only getting people to agree with you, and it is not knowing the right places and norms (that is cultural capital). Like economic capital, social capital is used to do things. If I want to start a podcast, I can use economic capital to buy equipment, hire audio engineers, and pay for advertising. Alternatively, I can use social capital to borrow equipment from my musician friends, have them put me in touch with their mixers, and tell all my friends about it. Depending on whether I have more money or more friends, one of these strategies will be more effective. Harry is correct that ANI does a bad job of investigating patterns of misconduct, and this is because mobilizing volunteers to comb through thousands of edits is hard; go look at the WP:CCI backlog. Investigative departments pay people to do that; they use financial capital to mobilize human resources to spend time and effort on a particular task. We do not pay people. We use social capital to mobilize our human resources, and we pay people in prestige, trust, and owed favors.

So to bring this back to ANI and ArbCom: social capital is not just about swaying people to your side. It is used from the beginning to get people to spend time considering your post and to get people to investigate the situation at all. If a new editor showed up at AN(I) claiming that an admin was misusing their tools, how deep would we investigate? We see so many frivolous complaints, many would just brush it off and not waste our time. Eventually it will get archived with maybe a few comments. In fact, social capital can be used to prevent mobilization of resources. If the reported admin quickly responds with a particular interpretation of events, they invest accumulated social capital to dissuade further investigation. In fact, look at the recent case of Tenebrae who hid their own COI by accusing others of COI editing and getting them blocked. The social capital they accumulated from generally positive editing was used to buy deference, prevent investigation, and silence critics by mobilizing human resources in their defense until it ultimately brought the project into disrepute.

Peer review is important and viable, I'm not saying it isn't, but it is not infallible or immune to the forces of capital. In fact, social capital as a sociological theory was developed to understand how organizations protect their members from outside attacks, create advantages for insiders, and ultimately (re)produce inequality; it was literally developed as a theory to understand how "old boys clubs" operate. Social capital as a theoretical framework is useful here because we are an old boys club. Our decline in active editors stopped around 2014-15 and has been increasing since 2019 (chart), but of our 1,109 current admins, 92% got the bit prior to 2015. Our processes don't work because they were designed to protect the privileges of early Wikipedians (Fernandez 2019); that is one of the privileges social capital works to produce and protect. Most of the people tasked with enforcing our norms have a vested interest in not doing so. I am not even trying to shift blame, as I said in the David Gerard ANI complaint I was aware of that problem for months before it got brought to ANI. I did nothing, because I didn't want to deal with the potential backlash at the time. Investigating the problem and getting others to look into it or take me seriously would require more effort than it was worth, and likely would have opened me up to the same kinds of harassment. Why would I volunteer for that when I could go deal with the RfPP backlog?

Why I push for a lower bar for accepting admin cases (and why I oppose remedies 5.X) is because it can help counteract these forces. Our time, as volunteers, is limited. Claims are unlikely to be investigated without some form of capital. At present, I have little incentive to publicly criticize my colleagues and the documented cases of retaliatory harassment give plenty of reasons not to do so. By contrast, ArbCom is quite literally tasked with compiling evidence and coming to conclusions and solutions based on that evidence. It has an explicit structure for inexperienced editors to follow, and a group of volunteers whose literal job is to fix formatting and inform them about the norms. ArbCom's mandate is the very thing Harry and I agree ANI is bad at: investigating patterns of behavior. But we seem to come to different conclusions. Instead of hoping someone will care enough to go through thousands of edits to establish a pattern of abuse, I believe we should use the existing committee for the job it already does using the structures it already has in place to limit the identified inequalities. By making it easier to report, we decrease the chance that problems go unaddressed for years because the newbies/victims got run off or blocked.

It is a structural problem, not an individual problem, which is why I am advocating for a structural solution. The solution--I think--is clear, and I have said it repeatedly: ArbCom should take all good faith admin conduct complaints they receive. If we do not find a way to balance the inequalities in social and cultural capital that influence our admin conduct dispute resolution process, the WMF has made explicitly clear that it will use its economic capital to pay Trust & Safety to do so.


I mean "compiling" very literally, and I mean "ArbCom" broadly to refer to the entire process not necessarily the arbs themselves. Cases usually have an "evidence" page where the community submits and analyzes evidence. From that perspective, it quite literally creates a compilation of evidence. We don't think of it that way, but that doesn't mean it's not one of the tasks included in the typical ArbCom proceeding. Who specifically does it is not really material to my point. Regardless of who does it, ArbCom proceedings differ from ANI because they have an explicit period for finding evidence structured in a way that disincentivizes the dysfunctional parts of AN(I). Accepting a case in itself is an expenditure of social capital: the committee is signaling to the community that you are taking the concerns seriously and that we should too. It mobilizes editors on both sides of a dispute to engage in the necessary fact finding because it has a reasonably well defined scope (as opposed to the open-ended inquiries of ANI that threaten to waste your time), a low risk of retaliation (unlike the public flame wars and boomerang of Damocles that plague every ANI contributor), and a cadre of people who are compensated (with social, but not economic, capital) for spending their time reviewing and acting on it (unlike ANI where your work may well just be ignored).

So, yes, one way to implement this would be to have arbs themselves investigate, but I believe the quoted statement already reflects the reality of how accepting a case mobilizes editor resources to a common task. Look at the difference in participation between this case vs the AN threads PR had started. This one from February had roughly 10 unique participants across roughly a week and a half; the case request alone had 43 editors engaged. The evidence page had 17 editors participate---with more on the talk---and only three were part of the AN thread I mentioned. As I'm writing, this page has 30 editors engaging in discussion about that evidence and we are all being quite civil about it. At all stages, this case has mobilized more human resources into dispute resolution than an entire AN thread. So, yes, we could try to shift community opinion to encourage Arbs to investigate more---this long analysis and your response may well help change some minds---but if you want editors to better engage in admin conduct dispute resolution processes, I hope my example demonstrates that the solution to incentivize that really is right in front of us, and it is not any of the remedies in 5


My ArbCom statement was tailored to the matter at hand, but the implications reach much further. Social capital gives us a framework to understand a lot of problems on the encyclopedia, and that allows us to tailor our solutions to the cause rather than the symptoms. So I do want to push back and say the community does acknowledge the imbalances, but often without the academic language or social theory to connect all the dots. That's actually what fascinates me about MeatBall: the depth of social theory that the community developed is remarkable for being done by a decentralized group of amateurs. That said, our "original sin" is the exclusionary bias of early internet culture that served as the foundation for the wikis like ours. That is what I think we need to grapple with more, and it's something that took me time to accept. Robert Fernandez wrote a wonderful article for the Wikipedia@20 edited volume republished in the Signpost that at first, frankly, angered me because it touched on uncomfortable truths; but he is correct.

If you're interested, there are two important takeaways from Fernandez (2019) that radically changed my thinking on the community's role in preventing and addressing abuse. The first is volunteer entitlement, and the second is structural bias. Fernandez notes that "altruistic" volunteerism can lead to sense of entitlement within the volunteers which manifests in harm to the organization they support. He cites anecdotes from Kennedy Center staff who recount their gift shop volunteers stealing. Having provided hours of free labor to the Center, they felt a sense of entitlement that allowed them to self-justify stealing from it and ultimately harming its cash flow (and by extension its ultimate mission). Fernandez says WP:BITEing is an example of this, I'm not so sure, but since reading, that framework has helped me understand phenomena such as WP:OWNerhip of articles (leading to edit wars and blocks/protection), WP:UNBLOCKABLEs (both in their behavior and in the community's response), and opinions regarding paid editing (volunteers do not want to help someone who is getting paid for their work). We are able to self-justify actions that plainly harm the project because we feel entitled to the product rather than beholden to the mission. But knowing the cause and the symptoms, I think, will allow us to better respond to these issues structurally rather than as isolated incidents.

The second takeaway is our structural bias, or what I called our "original sin" earlier. The internet is hostile to marginalized groups, and sadly it used to be worse. Early netizens where overwhelmingly white, male, and of-means because access to technology and the arcane knowledge required to operate it was gate-kept by existing racial and economic hierarchies. I think Fernandez actually says it best: In practice, to have a voice in a Wikipedia discussion requires a combination of stubbornness and privilege. This led to the reproduction of those systems online and in the early WikiPedia which helped centralize social capital from the beginning; at its extreme, this was known as the meatball:GodKing WikiFounder role (which Wales largely eschewed). Because of our consensus-based decision-making model, structures are relatively inert. Changing anything requires a great deal of social capital which is hard to marshal and even harder to amass. I've thought about this in the context of RfC closures. At Wikipedia:Non-sysop closures, I am developing the idea of a "cline of stability" where greater deference is given to closes by elite editors---essentially the stability of a close is related to the closer's ability to marshal social capital against those who opposed the ultimate outcome. Where the division between sysop and non-sysop closures comes then is in the implicit threat of violence (c.f. meatball:NonViolence): as a sysop I not only have social capital but also the tools to silence dissent through technological means. Non-sysops can only manage dissent through capital, but sysops have both capital and violence. This distinction is hidden, and hiding it reproduces the power disparities between an aging admin corps (reflective of early power disparities) and relatively younger and diversifying editor corps. In other domains, understanding our "original sins" I believe will help us understand the factions that form in our discussions and how to effectively work across the divides: who are we not hearing from and why? It's why I think recruitment is such a dire need. I believe Levivich has been saying that one of our biggest existential threats is editor recruitment, and that perspective has drastically changed how I understand and frame the consequences of the phenomena described by Fernandez