Read on Twitter
there's a lot of reasons why twitter as constituted is likely unfixable but this exchange more or less sums it up
note that the "wow this escalated quickly" moment occurred during an argument about a....cryptocurrency company
I wouldn't get too worked up about how serious Costolo is, as I said its kind of impossible to make good-faith interpretations of this at this point https://aelkus.github.io/culture/2019/11/08/idle.html
The point I'd like to make about Costolo's tweet (he is the former Twitter head) is that it shows a lot of things we already 'knew' about Twitter itself and its moderation difficulties but casts them in much greater relief:
First, for all of the discussion about bots, trolls, and anonymity/pseudo-anonymity we tend to consistently see that real names and verified identities are really not sufficient to create an environment where this sort of thing doesn't happen regularly
Second, every moderation decision at this point requires untangling multiple layers of subjective context behind apparently inflammatory (at face value) statements, something that is hard for English PhDs with backgrounds in Gadamer much less taskrabbits and robots
By tomorrow afternoon, this will likely have spiraled into a larger clusterfuck (it is spiraling as we speak) as people not affiliated with Costolo's tech set (in the original convo) join in because they're looking for a fight. Containment fails at the outset.
Naturally, one man's nuance is another man's excuse-making. It's easy to see (look at the QTs of Costolo's tweet) how it is perceived as an "aha, gotcha!" moment akin to catching the sheriff robbing a bank or two https://twitter.com/Aelkus/status/1311495678129049603
But if every attempt to enforce a rule runs into this level of interpretive difficulty, its impossible to have anything resembling a coherent and legitimate procedural authority.
Hence, moderation decisions mostly end up being subject to your pre-existing level of social backing. Who you can call on when you get suspended, how much Twitter loses by banning you. https://www.theatlantic.com/ideas/archive/2020/09/what-i-learned-twitter-purgatory/616144/
Again, I should emphasize that this runs counter to the popular image of social media being overrun by people deceiving others about their identities. After all, no one cries when a bot gets the heave-ho. https://twitter.com/Aelkus/status/1311494323196878848
This argument about inequality, however, only goes so far. Because if we assumed optimal policy enforcement regardless of status, we'd quickly find the policies themselves were incoherent and unenforceable.
The only way to fully and fairly enforce the policies without subjective interpretation of intent is what I call the "Prince Escalus rule" after the titular authority of Verona in Shakespeare. Start a fight in the street, Escalus either gives you capital punishment or exiles you.
"I will be deaf to pleading and excuses;
Nor tears nor prayers shall purchase out abuses:
Therefore use none: let Romeo hence in haste,
Else, when he's found, that hour is his last. "
Nor tears nor prayers shall purchase out abuses:
Therefore use none: let Romeo hence in haste,
Else, when he's found, that hour is his last. "
But we also know that the major flaw of Escalus' policy, even if it had some understandable origins (can't stop Montagues and Capulets from fighting, have to do *something*). It's a blunt instrument and it leads to a double teenager suicide.
Leaving the iambic pentameter behind and moving back to the main point... https://twitter.com/Aelkus/status/1311491719561703424
Twitter has a lot of problems individually, but they combine in particularly pernicious ways.
I've already mentioned that every moderation decision potentially may involve an enormous amount of subjective context. The larger the clusterfuck, the more the various factions involved and the more different layers of belief and beliefs about beliefs have to be considered.
But Twitter itself is not built on artisinally handcrafted free-trade green new deal sensemaking or whatever. It's instead, as @generativist discussed here, built on simple and shallow ambient social cues perceivable at the mass level https://generativist.substack.com/p/dunbars-number-is-quadratic
Moreover, big accounts (like Costolo's) are optimized for pushing out instantly to enormous amounts of people, *taking away* attention from others
https://generativist.substack.com/p/the-real-supernode-problem
https://generativist.substack.com/p/the-real-supernode-problem
The mismatch now becomes obvious: to make a "fair" decision about Costolo's tweet, one would need to give it a level of individual attention and care that runs at variance with the way the platform uses social cues and distributes big follower tweets https://twitter.com/Aelkus/status/1311504563522936833
Any decision will be (unsurprisingly) second-guessed by legions of people that know that the platform lacks a coherent policy and does not fully enforce it https://twitter.com/Aelkus/status/1311501218343251969
Knuth's warning against premature optimization is in context of understandable belief that overarching decisions about data structures and algorithm design limit how much benefit you will get out of hand-optimizing code.
I think there's a similar thing to be said about platform design and rule enforcement. Twitter made initial technical decisions about how to build its platform that John ( @generativist) discusses at great length. They, unsurprisingly, had consequences! https://twitter.com/Aelkus/status/1311450176058527750
These technical decisions -- as well as the way in which the company's leadership have repeatedly compounded them and doubled down on them --- mitigate against rule optimization much in the way the wrong algo or CPU won't make code run faster no matter how many loop unrolls u do
which is part of how our friend Costolo now ends up tweeting about revolutions and putting people up against the wall and such while angry people (with varying levels of sincerity) read him the TOS back to his face https://twitter.com/Aelkus/status/1311492496560713729
and why any moderation decision that will emerge from this will likely leave both his fans and enemies unsatisfied, angry, and convinced the fix is in.
It would be one thing if this was just some obscure elitist message board, for, i dunno, neopets fanatics with no larger cultural/political consequences to fuckups of this magnitude. but....uh...its actually the preferred medium of the President of this fine old country of ours
so the reality that its probably unfixable does hit a bit harder, doesn't it? everything from here on out is about triage in the short term, mitigation in the near term, and thinking up a better technical design for a place like this over the long term (hence me plugging John)
PS: I could have easily just done the usual "Bluechecks bad! There's one rule for the elites, another for the rest of us! Hypocrisy!" thing and waved a red shirt. And I wouldn't entirely be wrong to do so.
I did this prolonged stream because there's plenty of other places you will get that and I want to offer my followers something unique besides weird Doom memes. And the actual problem this represents doesn't seem to be going away, so maybe we can try to collectively diagnose it.
