Twitter hack has profound national security implications

It's also totally different than most journalists who push out stories left and right on topics they have layman understanding of. The quality and depth of work is significantly higher. I would pay for news reporting from gwern than random journalist.

Granted, Twitter is still a horrible place for long-form news (which is why it links to the full articles) and an even worse one for discussions. Which is why a different platform for instant updates could be beneficial, since news will have to adapt anyway.

Worse. It makes it seem like they shouted him down while not even delivering any concrete criticism. If that's accurate you can't blame LeCun for leaving. And looking at your typical political "discussion" on Twitter, I wouldn't be surprised if it is.

(One might say that Usenet did, at one point, contain all of academia within a few of its hierarchy-roots; but there's nothing stopping random quacks from posting to those. Imagine a version of the 80s where Usenet's sci. hierarchy would only accept posts from actual science departments in universities and research institutions—but where it was still mirrored for reading everywhere else—and you'd be closer.)

Really, what I badly re-invented here, was scientific journals, but with a 140-character publication limit ;)

Or IRC channels in +v mode, but where there's only 10 of them, and each one has millions of unvoiced members and thousands of voiced members.

The point of "discourse domains" as a concept isn't to create echo-chambers of (correct or not) majority views. It doesn't aim to solve every problem that moderation, ignore-lists, etc. solve. The point really isn't to limit what people say at all; it's to limit how they say it, per domain (or rather, to make it clear what the proper "how" is, when you're in this space vs. that space.) It's to reconstruct a sense of there being a conversation happening within a context of particular sub-cultural discourse norms at all, vs. the current situation we're in with the Internet where every open-membership website has the de-facto discourse norms of "a protest in the public square."

As for why you'd want that—if it isn't immediately apparent: people who are enculturated to a given domain's norms, see certain things as being offensive (norm-violating), and these things differ per domain (but not really per community, given two communities both "of" the same domain.) So isolating domains from one-another, means that you eliminate "impedance mismatch" conflicts where everyone is trying their best but the difference in discourse norms makes everyone think that everyone else is acting in bad faith. It especially helps out the people for whom the norms of "a protest in the public square" are quite far from the norms of the domain they're enculturated to.

(Have you ever seen a scientist argue with a politician on Twitter? That's an impedance mismatch. Each one is trying to get different things out of the conversation, and so neither side can actually say anything without it seeming to be an act of rhetorical sabotage when seen through the lens of the other side. And further, the politician is closer to the norms of "a protest in the public square", while the scientist is further, so usually the scientist is seen by Twitter users as committing the greater faux-pas in the process.)

> This seems great in theory but how do you "verify" these types of domains without also instituting some kind of gatekeeping?

I hope it's clear, here, that the point isn't credentialism / closed-membership communities. You aren't trying to detect whether someone e.g. "is a scientist"; you're only trying to detect whether someone is best modelled as being offended by the set of things scientists are offended by.

"Verification" (in the sense of Twitter's blue checkmark) is a KYC mechanism, not a credential per se. It helps people to find a specific person on Twitter by their name. It doesn't assert that they're a member in good standing of their field; someone could be "verified" for being infamous just as much as for being famous. It just ensures that the people who "squat" on names are the people you'd expect.

The suggestions in my previous comment for verification strategies (and for linking verification to "voiced-ness" within the domain), just take this a step further: they're suggestions of processes that would be applicable to anyone who really is enculturated to a given domain's norms. They make it hard for people "uninitiated" in the domain to come in, but not by requiring a credential or a skill test; rather, they just test for something trivial that people only learn to do after first enculturating to the domain's norms.

There's no academic who hasn't written a paper—you have to write one to complete a master's and start doing original research (or, if you're an auto-didact who's not part of an academic institution, papers are even more important, as they're the way you collaborate to advance the field at all. Like PRs are in software.)

And there's no professional politician or talking head who doesn't have at least a few people willing to vouch for them. (It'd be like saying you're a professional e-sport athlete with no ladder rankings anywhere.)

None of these things require effort; they're just things you do as a matter of course as part of the process of becoming enculturated to these domains of discourse. They're like those locks in fantasy books that are opened by pushing mana into them: they don't stop bad or evil wizards, or even incompetent wizards; they just stop people who don't understand wizardry and so will probably just muck everything up if allowed inside. (Imagine a lock on a chemistry lab where you had to draw a benzene ring to get in. Same idea. They don't want you blowing yourself up in there.)

-----

Note, though, that once you've got a working domain with a cohesive set of norms, it becomes much, much easier to moderate it, because everyone in a domain will agree what the social norms they're abiding by are!

That's the real point of all this: if you untangle from your social platform all the different sub-cultural domains with their different discourse norms, then each domain can be moderated independently, by people who are also enculturated to that domain's norms and so understand them implicitly. They won't be blindly applying a checklist of rules; they'll be applying the spirit of a law that both they and the people they're applying it to understand.

(There's a good comparison to the "cultural laws" of Judaism and Islam, here. Take "cultural laws" and the priests who study+interpret+enforce them; subtract the religious aspect; and you have a domain and its moderators.)

Yes, that is the charitable interpretation you could have safely assumed I meant.

This was exactly my first thought when I saw Sen. Hawley's open letter to Twitter, aghast at the idea that the President's own account could be affected by this. I share his concern, but not his determination of blame. It is not a private company's job to protect the communications of our government. It is the job of our government's defense and intelligence apparatus. They can't do that job without the requirement for communication to be done through channels they control.

Anything is possible through legislation, I'm not sure this is such a big leap.

An account on a private service was deleted. Not a brilliant move. But not something that merits more than being fired and possibly pursued for money damages.

The fact that high profile accounts were posting content not from the 'owner' of those accounts.

> Fault

A fool and his money and soon parted. These people are foolish.

I appreciate it more when people are willing to do the work.