Consensus
Sat 28 Sep 2024 5:02PM

the "how can we proactively address racism"/"what can we learn from Cohost" discussion

AB Alyaza Birze Public Seen by 36

as we move toward launch and into the beginnings of the Website League, it's time to start confronting probably our hardest question(s):

  1. how can we proactively address racism?; and

  2. what can we learn from Cohost's failures on this front and others?

there are probably other questions that will flow from these two, but in the interest of not being overwhelming let's start with those two.

My feedback

my feelings on this, and the specific failures Cohost had are elaborated at length in my postscript to Cohost So White but, to summarize:

  1. Cohost had an incredibly white founding userbase, and almost everybody that userbase invited was also white. this made it nearly impossible to invite, attract, and retain minority users

  2. Cohost's founding community had aggressively parasocial and protective elements that often reacted negatively to criticism of the site, and victim blamed or dismissed users for "creating their own problems" and "using the site wrong"

  3. bad actors were not dealt with and the moderation was, in staff's own words, cowardly at times, leading to an accumulation of bad outcomes. this might be an individual failing or a systemic one; but either way it's a failing.

  4. Cohost fell victim to this dynamic: We also need to be hyper aware of minorities which are relatively rare due to the law of large numbers. [...]let’s say only 1 in 100 people are trans - we now have an equal amount of trans people and jerks on the internet. If even 1 of these trans folks don’t want to engage with a jerk, a single bad actor can quickly drive all transgender people off a platform like this one.

  5. Cohost's anti-harassment measures were not useful and failed from first principles. the issue was more abstracted and hidden for regular users, but still apparent to anyone targeted by a harassment campaign or just the center of attention generally.

  6. in the absence of technological measures, many of the insinuated benefits of Cohost—the things that enabled its lack of harassment—were purely cultural. this culture broke down frequently, especially with users who did not at least superficially resemble the fabled Founding Cohoster in demography

  7. the sum of this was to make Cohost's community feel hollow, fake, and performative in its anti-racism.

External feedback

i've also solicited feedback from other relevant users (and those in close contact with them) of what they feel the problems are/were. here are some of the responses as of now:


  • you can’t have different moderation standards for your friends

  • knowing whether a report hasn’t been seen or has been seen and decided isn’t an issue can be helpful. with multiple instances, it could encourage someone to stay/leave their instance, which is actually a choice in this system vs cohost. the instance i use doesn’t do this for individual reports but does do it for instance bans, so it might not be necessary if there’s enough trust in the moderation (i have this trust from simply not seeing shit that i’m aware is happening, because they already banned it), but i would wonder if coming from cohost if that trust might be fragile?

  • you gotta moderate besides just reports… again, i assume this is much easier with a smaller instance where you can plausibly just browse and check things out!


  • Leaving wild-ass comments unchallenged and chastising people who pointed them out for tone (thinking about the “neotenuous” commenter in particular)

  • Dismissing criticism about racial (or really any) issues as either unhelpful whining or active sabotage

  • Using the prospect of disunity being a threat to the project as a whole as a way to shut down criticism


I don’t think the anti-discovery / anti-harassment measures on cohost really had much to do with the site’s culture over the fact that registration was heavily restricted to friends and family for a long time.

I think in many ways it was a drawback because it didn’t actually stop targeted harassment and made it harder for any sort of support network to form around most of it, because the visibility was kept low. I described it to the effect of an anti-solidarity measure.

It’s not actually possible to engineer a system that prevents abuse of itself, which is a major reason why human supervision (i.e., moderation) is necessary and needs to be proactive. I think cohost believed that having these measures would allow for less time and attention spent actively moderating content. This would match up also with the presumption that it was more important to moderate for tone / trying to prevent fighting rather than to actually manage long-term outcomes that these disputes are tied to (which, go figure, tended to hurt racially marginalized users more).


If mods have rules of engagement, then those rules should be equally applied across the board. [...] Whether its by reports or alot of people complaining, mods engaging with a potential issue should be equally applied across the board.

WM

walking mirage Sun 29 Sep 2024 8:05AM

@v1s1n suspending someone for 6 hours or a month is a lost art

R

ruby Sun 29 Sep 2024 7:51AM

From Shel's Cohost Eulogy & Retrospective:

So when a small group of white people invite only their friends to join a new community space, it is statistically likely that the community space will be 90% white to start. That makes it very hard for it to become more racially diverse later.

I'm... definitely thinking about this. Especially along with point 1 in the OP, I am questioning my decision to run my node as "invite only to people I already trust". My rationale for this is simply that I'm not certain that I can handle the moderation load that comes with having people I don't know well and don't already have a base level of trust built with. However, this is almost certain to run up against these 2 points here, on account of me being white and not having a whole lot of PoC friends (especially not ones that would be interested in this project).

It seems to me like it's worth reconsidering whether "invite-only to friends" is something we should permit within the League. I'm not confident there's any way we can reliably mitigate the concentration in racial demographics that this would inevitably bring to the League. Bringing in (for lack of a better term) "unknown variables" to an node is likely to come with more work on that node's staff, which is something we want to avoid (we are all volunteers, after all), but at the same time... it's like, that workload is kinda what you signed up for.

S

sirocyl Sun 29 Sep 2024 1:34PM

@ruby very good points. the way I see it is that instances should be for "less than five" (i.e., between zero and five people that the instance owners want to bring in) or "greater than fifty" (i.e., inviting in far more than one's own comfortable in-group).

The idea here is to reduce the "association multiplier" one has by inviting an in-group.

If one white person adds all their friends, that might be twenty or thirty largely white people, overall, inviting their twenty-to-thirty white friends.

If, instead, they keep it very small if it's going to be friends-only, then the impact of those few friends will be less than the dozens to hundreds of friends-of-friends.

And if they open to a more general public, then people outside these mostly-white "bubbles" will be more likely to join.

S

Shel Mon 30 Sep 2024 8:21PM

@ruby I think it's worth pointing out that "We won't have many problems if we just keep those unruly problem people out of our community, and only allow in civil reasonable people" is like... a classic white mentality when it comes to community-building. "I won't really have to moderate if I only have my friends/close circle on here" does not jive well with "moderators not giving preferential treatment to their friends."

White people fled big cities to live in small suburbs that would be "safe" because those communities wouldn't have any of "those people" and would only selectively allow in "good folk." If joining Weague requires finding someone with strong IT skills who wants to invite you into their neighborhood, it will be more difficult for people who are different from these admins to find a neighborhood that will allow them in. The entire network will just feel unwelcoming. I think a lot of federated/alternative social media has an unspoken assumption of "If everyone present is sufficiently like-minded and similar to each other, it will be more harmonious" but this predicates on homogeneity as a core element of the community. You can't seek to create a homogeneous community in all things but race, and then expect to end up with a racially diverse community.

Historic efforts to resist racial integration on the parts of white people usually downplay the racial component and just say things like "we are a tight-knit community of like-minded folk and we don't feel safe welcoming in all these strangers who could be ruffians and criminals." What is your anxiety of someone you don't know making an account on your node and how does it compare to the anxiety of a stranger, or person who is different from you, moving into your neighborhood? Embracing diversity classically requires getting over those anxieties. Racial integration efforts require white people not just allowing families of color into their neighborhoods but actively welcoming them and making them feel like a genuine full part of the neighborhood community whose presence is appreciated. When a stranger opens an account on your node, they should have the neighborhood association show up with fresh baked muffins and a strong desire to get to know them, not suspicion.

It might be challenging to convince the stewards of, but maybe we should seriously consider banning invite-only nodes. Set a size-cap, sure, but there needs to be easy go-to welcoming nodes that will accept you and welcome you if you do not know anybody. In fact, those nodes should probably be the majority and need to be seen as important. On the fediverse, such nodes were often derided and treated with disdain. ".social user" was an epithet. This sort of elitism is unhelpful for diversity. So long as owning a node is tied to your socioeconomic status and level of education in specific trade skills, allowing those people complete control of the community and who can join it will replicate the fediverse feudal system and will perpetuate the White Techies Only vibe.

If someone is unprepared to moderate a node that has strangers on it, maybe they should not be making their own node and appointing themselves to the status of community leader.

S

sirocyl Mon 30 Sep 2024 9:47PM

@Shel If there were a vote on banning "invite-only" nodes, I would probably agree with you there. Perhaps it's a good idea to start a poll about it, too.

My scope for "less-than-five" isn't so much "invites" as it is "literal households" or "teams". People who are already formed into a group independent of the group-forming characteristics of social media.

Everything in the "invite your friends and friends of friends" scope, between the <5 and "mass acceptance" buckets, is what I'm okay with considering out-of-target for operators on the League, and something we should possibly downregulate, if not outright forbid.

S

Shel Mon 30 Sep 2024 10:20PM

@sirocyl I wonder if we could classify nodes? It feels weird instituting zoning laws but something like "Nodes larger than 5 people may not be invite-only. Nodes must be open registration until they hit a ratio of moderators to users of M:U. Nodes may not accept more than U users per M mod." We could call <5 user nodes "households" and >5 user nodes "neighborhoods."

I sort-of wonder if calling them "neighborhoods" instead of "nodes" would get people thinking about being welcoming in a different way. Your neighborhood should not be your clique of friends, it is a place people are looking to live. There are only so many neighborhoods in the league, so room for accounts is like housing, if each neighborhood only has so much room, and everyone who wants to participate in the league needs an account on a neighborhood that has room, then we need to be treating this more like the housing market and ensuring that people can get accounts where there is room for them and feel welcome and not like an undesired outsider.

S

sirocyl Mon 30 Sep 2024 10:39PM

@Shel The thing is, I don't know if we should codify it in particular, so much as make it a progressive encouragement. In League Code terms, a "SHOULD". There could be local/team groups larger than five, and I don't think it's so much the "size" that we're trying to regulate (though, that is a good indicator) - it's the establishment and acceptance factor. We want teams, households, polycules, companies (not corps, but in the classic sense, like a theatre company) and cooperatives.

Like, we could prefer that anything that isn't already an established small group like this, does not grow except by organic and open merit to all, even toward people said group may not consider friends, or acquaintances, within reason.

S

Shel Mon 30 Sep 2024 6:05PM

Some thoughts:

  1. Having the technical know-how to run a node does not in any way qualify someone to be a community leader or moderator. I recommend not just moderation guidelines be created but possibly a manual, or even a training. We need to support our moderators in understanding best practices, the basics of anti-racism (this should not be assumed knowledge just because someone identifies as an anarchist or queer), etc. and should probably have regular moderator check-ins across the league where mods just talk about how they're doing, what they're up to, etc.

  2. There should never be only one person doing anything important. This is a principal I learned in my offline community organizing and it was crucial. One Mod Systems do not work. If your small node only has one mod, you should elevate a second person from your node or maybe ask someone from another node to pair with you to double-check your work. This also creates a failsafe to ensure moderation is still happening if someone gets sick, overwhelmed, hospitalized, etc. we don't want appendicitis resulting in an unmoderated node with no awareness that this has happened. The fediverse infamously ends up with "ghost admins" who continue to pay for a server but haven't been seen in years and provides no moderation.

  3. Clear expectations around moderator response timelines. Users should not expect immediate instant action on reports from volunteer moderators. Mods who are constantly glued to the screen and extremely online are probably not going to be in the best headspace to make good decisions. There have been studies on the harsh mental health impacts on professional moderators, let alone volunteers. It's thankless work being exposed to the worst things online regularly. At the same time, there needs to be a cap on how long a mod can sit on a report before their inaction becomes problematic. Leaving users in the dark erodes trust and they need to know if we're in the "Be patient, they'll get to it" stage or the "my mod is just ignoring my report" stage. The one who filed a report must at the very least be given something like "Hey, we're still working on your report, it's taking longer than expected because..." Something like "Reports will be acknowledged within 72 hours and acted upon within 1 week, unless extenuating circumstances require additional time. Extenuating circumstances must be explained to a minimally satisfactory level and communicated to the userbase that reports are being handled slower right now. Efforts should be taken to find a backup mod if possible."

  4. I don't like "ban first investigate later" but I think "issue a timeout first, resolve the conflict later" is a good option. I feel like the internet has forgotten the power of "(User was put on probation for this post.)" Especially when there is an active conflict with one user being an idiot but mods disagree on whether this is bannable or not, just time them out for 48hrs and decide what to do while they're on timeout. The active conflict has been terminated in the meantime because they can't keep posting. If mods think someone is 'ignorant' rather than 'malicious' (even though malicious actors feign ignorance) then time them out and have that conversation away from the person on the receiving end of their BS.

  5. No anonymity. Anon asks can be fun but they're just a bad idea.

  6. I think it's still important to expect users to file reports. Mods shouldn't have to be trying to view every single individual conversation and interaction on their node through a panopticon.

  7. I think moderation transparency is important. I do think it's important to balance the transparency. I grew up in a cult where every single misdeed anyone did which went through the judicial committee was published weekly for everyone to read. It became public shaming. If someone is banned, then the ban and reason can be public sure, but for timeouts and mediated conflicts, those don't need to be broadcast to everyone.

  8. I think some of the issues on the fediverse and cohost stem from an assumption that being a radikewl queer person who hangs out in alt-culture kinds of spaces means you totally received the liberal arts college orientation on social justice—when in reality many of these people are computer engineers who went to schools where the humanities weren't even taught. There is no magic barbie speech that deprograms racism from white people... but some 2015-era social justice 101 educational materials for new users might actually be warranted. Given that the Autism Rate will be High, it never hurts to spell things out explicitly. New users are given a one-pager on racism, or something like that. It feels kinda cringe but like, maybe it's worth trying? 99% of users will be like "Oh yeah yeah racism = power + prejudice I know this already" but maybe it'll catch the missing stairs and create a floor for common knowledge we can expect of all users, and if they say something stupid there's no longer an excuse of having been ignorant. If they just didn't read it, then that is their fault and we can safely ban them for being someone who doesn't think it's important to read about racism.

  9. Being open to new users forming new communities who aren't like us but still agree on our base principals. Maybe they won't all be Typical Cohost Users but if they at least jibe with the style of governance we are open and welcoming. Putting in effort to promote this option in places that are already racially diverse, not just Cohost.

AS

art semaphore Mon 30 Sep 2024 9:01PM

@Shel per point 1: strongly agree. im very technically competent at running one of these nodes, but having been in a moderation position before, I know that I don't know how to do it right, but I don't know how I'd learn to do it better, short of trying, failing, and making it everyone else's problem. And, I think for similar reasons, I also don't know how to identify if someone else will be good in a moderation role or not. I have over the years observed when someone is having a hard time moderating and I feel like I've gotten a sense for a number of common mistakes folks make. But the set of "things not to do" is vastly larger than the set of "things to do", so that only goes so far.

K

Katja Tue 1 Oct 2024 12:46AM

@Shel 1–3: agreed.

4: by "ban", I meant some kind of reversible moderation action, yeah. Explicitly not Mastodon-style "your account has literally been deleted deleted" bans.

5–8: agreed.

9: yes. Absolutely. Seeing as we're already making basically a massive social and technical fork of the Fediverse here, it'd be outright hypocritical for us to not encourage people to, in turn, build off of our work.