TLDR: the AI community is unable to solve a collective action problem and chooses to ignore harms as long as those harms do not hit them directly.
The point of this article is not to suggest that you should go to bluesky, or to say that everyone who continues to use twitter is a monster. I just want to talk through what our failure to quit says and why I think we should probably give ourselves a little more moral side-eye about it than we do. What does it mean that we've normalized a space with such bad norms?
I'm going to start with a few operative assumptions about you as a reader.
You broadly agree it is a bad thing that whenever an immigrant researcher becomes too visible on twitter (or someone says something positive about immigration), they get swarmed with variants of "go home" / just straight up slurs / disgusting GenAI images of themselves generated form their profile picture.
You broadly agree that it is a bad thing that anyone with a Jewish last name receives a series of antisemitic slurs anytime a post of theirs goes viral.
Now perhaps you don't agree that these things happen! In all likelihood, you have not personally experienced it. Unfortunately, I cannot point you to one of these threads without putting the corresponding researchers on blast. So I'll use myself as an example instead. Whenever a post of mine touching immigration topics went viral, back when I used twitter, I would get some mixture of:
being called a kike/heeb/yid, etc.
death threats indicating that they knew my work place.
variants of "huh, what an interesting last name he has. Wonder where it's from" (a clear way of signaling to other antisemites that I am Jewish and should be gone after)
Unfortunately, looking back for an example the worst comments appear to have eventually been cleared up, several weeks later, so you'll have to take it on faith. I think I'm generally pretty credible but if you don't believe me, what am I to do.
Now, I am not particularly hurt by these comments, as my identity does not rest on what antisemites think, but what I would like to interrogate is the implication of the community continuing to remain on a platform where they know this is happening.
So let us say you know these things are happening, then I think you would also broadly agree that it would be better if AI discourse continued on a platform that is not twitter. So what could explain its persistence and what would each explanation say about the community?
It certainly is the case that if you only use your following feed, and you never go viral, and you are completely immune to clicking on rage-bait, that it is possible to not see these things. However, if you know it occurs (and I personally find it beggaring belief to think that people do not know it occurs on a platform where its main LLM, after fine-tuning, went on a spree of calling itself mecha-hitler and its image generator was used to make CSAM and exploitative imagery to the glee of its owner) and continue to use the platform, then the presumed explanation is you think one of the following:
as long as it is not happening to me, it is not a problem
although it is happening, it is not a big problem. This is not a significant harm to the people who experience it.
my use of it does not cause any of these harms to occur, therefore there is no harm in my continuing to use it.
the net upside of my continued twitter use outweighs this harm.
I think 1 is trivially immoral, so I don't even need to argue it, and 2 is silly though I accept that people can disagree about this. However, my personal experience, talking to people who seem to regret their twitter usage, is that they are cognizant of the harm and find themselves in camp 3 or 4. I think 3 is perhaps the most superficially sensible reason. Considering that you aren't the one doing the bad things, what is the harm in remaining on a platform where some of your friends and colleagues are subject to the harm? I feel like one should be able to think this one through without my hand-holding (hint, your presence makes it hard for your friend to leave).
In camp 4, whether it's that they find it useful for finding papers, publicizing their papers, staying up to date on some discourse, or simply that it's fun, they are aware that there is some damage being done but the costs of finding another way to replicate these features is too high. To me this last one reads as a classic collective action problem. We mostly agree that it would be better if a different system were in place, but we are in thrall to the current system.
And so, we continue to tolerate the harm, hoping or believing that it will never be pointed in our direction and comfortable in the expected value pointing in the right direction. The AI community is literally the most powerful a research community has ever been, what does it say that we can't successfully assert our morals on this tiny point?
Remaining points that might come up
I didn't know this happened to people!
Okay, I believe you. Now you do. So what camp are you in?
You just hate freedom of speech
Fairly often the horrible environment of twitter is described as unavoidable if one wants freedom of speech. I think this is an intentional sleight of hand to conflate freedom of speech with reach. If the government was insisting that certain things couldn't be said, that would be a freedom of speech issue and I'd usually be on the side of speech. If an institution decides that its platform shouldn't be used to host hateful conduct, that's well within the right of the institution. If someone showed up outside my house, set up a soap box and started spewing anti-semitism, I don't think it's a freedom of speech issue that we'd all try to get him to leave. I think we're all allowed to come together and say that we prefer platforms where you're not free to say hateful things with the understanding that there will be inevitable adjudication (and mistakes!) around what constitutes hate.
Why do you call it twitter? It's X.
X is a horrible name whereas twitter was a pretty good one. A random rich person doesn't get to decide what I call things just because they said so. The facebook->meta transition made sense because the company was now a much broader category of products than facebook alone so continuing to call it facebook was more confusing than helpful.
So just don't talk about immigration.
I think "you're not allowed to talk about this topic else you get swarmed with slurs" is a bad default status to allow. I'm perfectly happy with people to disagree with me about this topic, just not in that way.
Why do you care so much about this, aren't there more important things?
Sure! I talk about those too.
You're making this up, it's not that bad.
Oy vey, you don't believe me? Well, at least you got all the way to the end of my little essay.