Thoth

And he didn't even agree to the TOS.

I reread the watch books recently and the shift from Night Watch to Thud was stark. The entire book is drenched in incredible atmosphere.

(I mean that's not to detract from Night Watch, which is an easy top 5 discworld book for most fans, but it's doing something very different)

I say this as the proud owner of ---.leaflet.pub.

And the regex to validate them isn't too painful (although you do technically have to sneak a length check in there to avoid a lookahead assert and sometimes people forget the special rules about the first and last characters of a segment)

You just said it couldn't do anything new and then said it's entirely novelty?

If you're trying to say that it has zero useful real world applications... I'm sorry, but you are just objectively factually incorrect. For good or ill, people are getting real work out of these things.

The stories I've heard about early twitter don't paint him in a particularly flattering light either.

(Like, a revisionist framing on Jack's part to be clear)

IIRC he left as a direct response to bluesky doing, like, real moderation rather than just the bare minimum. Or at least that's what I remember him saying at the time. So this seems like a bit of a revisionist framing? I dunno.

I am evaluating usefulness, not novelty.

I mean Kalanick was a real piece of work across the board so I'm pretty sure any robin-hoodery was entirely accidental.

They can negatively impact people prone to psychosis and that's a concern worth taking seriously, but no, AI does not "rot your brain." The study you are probably referencing was questionable in its methodology and never even made the claim that AI rots your brain to begin with.

"You can thoughtlessly generate code without even looking at it" does not imply that everyone who uses AI does/will do that! I don't think "all AI use is a slippery slope to vibecoding" makes sense!

No this is 100% correct. Like, even when you explain that not all AI assistance in programming implies "vibe coding" as originally defined and you can be more thoughtful about it, people will just be like. "no."

The fact that atproto devs broadly seem to think the average user's problems are worth solving is a big part of why I'm here and not on the fediverse!

The thing that makes me optimistic is that the dev community feels pretty aligned on making credible exit more accessible to the average user and agrees it's a problem worth addressing.

Well yes. But if normal nontechnical users get screwed we all suffer. Ask not for whom the bell tolls etc etc.

Any plans to support linking to code on @tangled.org? Seems like the one possible atproto integration that could be helpful for scientists that hasn't been mentioned.

Being able to generate more code faster means needing to think harder about what code is worth generating as otherwise you will drown in a pile of garbage.

It surprises me more people haven't made this connection yet.

tbf this is an extreme exaggeration of what he actually said.

Not that what he said isn't crazy, but it's. Less crazy than this?

I do worry about them being able to swing enough premium users to become net-profitable. That's gonna be a tough problem.

I think there are good opportunities for non-abusive monetization, especially as a PDS host, but even so, running a social media site with lots of images and video is costly.

I mean, legal status can't protect you from your actual leadership being relentlessly profit focused.

(not saying being a PBC is a cure-all, but the failure modes of OAI and Bluesky look very different imo because Jay isn't sama.)

I'm less critical of the bsky team than most because I can see the reasoning behind a lot of their decisions even when I think they're wrong, and I don't think the team is ill-intentioned, but. Uh. Boy do they know how to screw up and not know how to do comms.

(It is my theory that most of the motivations people might have that lead them to WANT to run a large-scale social media site tend to make them unsuitable for the position because doing it well is deeply unpleasant and not very rewarding)

I suspect a part of the problem is that bluesky didn't really want to be running a social media site, alas.

Equally unfortunately I strongly suspect that if the company that built this place HAD wanted to run a social media site the decisionmaking would be even worse.

That is actually an important point, yeah. Jay can freely prioritizine the protocol over profits without having to worry about it being a breach of fiduciary duty.

Hopefully having a real comms director helps...

The team's propensity for airing their frustrations to the public is a big problem even when their actual decisionmaking is sound, yeah. Like, everyone working on a UGC site has a lot of these frustrations with their users probably, but they don't say it out loud!

...But we are definitely not at the point where the ecosystem can stand on its own two feet if the investors pull the plug yet, so. Uh. Well, again. Here's hoping.

I do think it's possible to build profitable things on atproto, just. Perhaps not to VC expectations.

(This is not entirely down to bluesky—in particular, @bad-example.com's work on the @microcosm.blue service stack in terms of both making public instances available and making it readily self-hostable was a real step forward in terms of ease of building real apps)

Which... I think they're broadly moving in the right direction and the progress in the past year in terms of both making it easier to spin up independent infrastructure and easier to consume that infrastructure to build real applications more easily has been faster than I expected. So here's hoping.