As possibilities expand, the "ATmostphere" can be improved. Maybe in two years everyone wants to go to a new Personal Data-Store that offers new features X or Y, or Bluesky is sucking because of Z...

This architecture of making users sovereign is what's really at stake with the 'user = site' idea; the meaning of that is powerful & clear to techies for what it ongoingly unlocks & enables, but most users arent expected to share that technical fervor. It insures the ATmosphere doesn't get trapped or corralled into some local maxima, can't be enshittified since there isn't the same switching cost as all other networks. And it enables growth & extension of the ATmosphere, by allowing innovation at the edge.

More generally, I'd argue that users don't have to be familiar with and attracted to features for them to be features. No one knows every function in their spreadsheet software and there's some they probably will never use, but the spreadsheet is known & respected because it has such a broad library of functions to enable so so much calculating, in so many different ways. Users can grow & change where they are in the adoption curve over time.

Features evolve on healthy & fetile platforms. Giving users sovereignty & mobility enables new platform to be created off-Bluesky/main: it is a meta-feature to allow new features.

However, the more I learn about what Jay, Paul, and team are doing, the more I realize that the protocol is actually important. Them being federated protocol nerds, while also having to focus on supporting millions of users is super interesting to watch unfold. This could be really cool because they really thought about the underlying factors, from go.

There are many user issues like "when I block someone, they should no longer follow me" which seem so simple from a user POV, but are actually complex from the federated protocol POV. I am learning a lot by watching what happens here.

If that doesn't work, just substitute "account" for "user". "Every bluesky account is a site."

  > still struggling to sustain 2 million DAU.

Seriously. Do we need every service to have a billion users? I'm glad there's places that are open to everyone, but at the same time I'm sad that a lot of smaller and niche places have disappeared. Those are the places I've found communities. It's possible on bigger sites but it's much harder. Your voice is one in a billion, it's not much better than having no voice. It's even happened more as HN has grown. And in any of these big places it's more likely that sensational posts rise to the top drowning out sensible or accurate messages (maximizing engagement doesn't maximize user happiness, communication, community, nor many other things). It's also harder to communicate because every community develops their own language and priors. A mutual understanding and that's what builds good faith conversations. I think if we've learned anything from these mega sites it's that even when people use their real names they still view themselves as "anonymous" (and conversely, in small sites people can hide their name, but it means something to the community, and they still have a reputation to defend)

I'm not convinced we are measuring the right things. It just feels like people are measuring what they think they should, and are not thinking about what the actual goals are. These are not the same thing. But that's Goodhart's Law I guess

other than a few VC heads, in no world is millions of dailies a bad thing. i’d be shocked if hn had 2 million daily. and thank your god it’s not full of everyone screaming at once. it’s what makes it such a rad place to be.

i moved to mastodon when the infosec community migrated there and honestly, its OmgSoMuchBetter that it has less people. it’s like here on hn, the signal to noise ratio is fucking glorious.

i’ve mentioned this multiple times in the past, but in reality we (people) prefer multiple different spaces, and for very very good reasons.

it i want a quiet night with a fancy dinner, i’ll go to a quiet nicer restaurant.

if i want a loud night out with the friends, we go to a club.

if i want a goth night i go to the goth club

if i want to hear blues i go to a blues bar.

etc…

the big sites suffer from their own ridiculousness and still, years later haven’t figured out that trying to be everything at once is hilariously stupid. it just makes you bad at everything.

blue sky is incredible, at least until eternal september hits. Mastodon’s different servers hit so much better than the last years of twitter did. the minute everyone flocks to either of them, im out. we like different spaces for different moods, it’s common sense.

could you imagine how awful going out would be if you’re talkin with your friends and over and over some totally random people were constantly coming to your table and screaming shit at you? “defend yourself!” … no, because that’s not normal… bar owners would be like “bro, that guy who keeps screaming at random people has to go… fuckin weirdo.”

that’s the experience these same group of VC people keep trying to tell us is normal and what we should want. no thanks. thats weird af.

We started as a protocol design company. Shifting to a product & services company has been a process, and "moving quickly" as a new org in an established market-space is a comical challenge. We got pitted against Threads, which was able to reuse the scaling and moderation infra of Insta and take on 100M signups in their first week. The only way to be competitive here is to find a differentiated & novel execution which finds PMF.

The protocol work is the PBC mission, but new technologies need PMF to gain distribution. We're not trying to sell people on ideology or ethics. I do want folks to know how the technology works, however, so I will post threads like the one linked here.

https://paulgraham.com/die.html

As in, if Mastodon got say 2 million new users in a few days, then each instance would have to sync the entire DB. Most instances could not handle that. In AT's case, that is handled more efficiently. That is correct, isn't isn't it?