> I think the way to solve that is to have a global cross-community feed and to surface active communities.

The default sort is already "last active" for communities on Satellite.

But I still agree with you discovery needs to be a LOT better

The eternal question of whether the perfect is the enemy of the good

R2 specifically. I chose R2 on account of the pricing model that charges only for storage and not data egress (bandwidth). That's actually a fairly big deal because with other options (e.g. S3) you're always at risk of getting a huge surprise bill if some hosted data gets really popular, and this forces all the services built on it to host third-party data to be post-pay with no guarantees to the customer about how much they will be charged.

By replicating R2's pricing structure, Satellite CDN is able to be flat-rate pre-pay, and therefore a good fit for lightning.

If you're concerned about censorship, that's what NIP-94 is for. For every file uploaded, Satellite essentially creates a torrent for it, and returns these params to the client. So if Cloudflare ever pulls the plug there will, in principle, be recourse to self hosting. Also, the url path of every file on the CDN is just the SHA256 hash, so to deal with the problem of dead links in the long term, clients could just parse out the hash from dead links and serve the same data from a different host.

So slick. Looks better than my UI 🤣 nostr:note1h5enh9udrl7jy8x9cjsre7uja6vz7y02yuej5m9ylrscmw0uhvestgerjp

With next update there will be a way to exclude community posts from your main timeline

That's the one

> Create a new kind and use the moderated communities NIP

Currently working on implementing exactly this

Can't wait to try it out! I've been playing with shipyard this morning - writing focused client seems obvious in hindsight (like most good ideas)

I really like the focused writing mode

WHAT

Will be fixed soon

Satellite works differently than other clients in the way that it surfaces *threads* rather than *notes* — so if someone you're following in replied to a top-level post from Alex, that thread will show up in your following feed

That'd be super cool

I'm sure this has already been discussed, but does anyone know if there's like a standardized "max request size" for relays? For example, if I open a subscription with an authors array of with 1000 pubkeys, that seems to work as expected, but trying to match against 10,000 pubkeys doesn't work at all - I'm assuming that relays just drop subscriptions with filters over a certain size, which makes sense, but I'd like to know what the limit is so I can truncate queries on the client side as necessary

If someone manages to apply LLMs to predict financial markets in a way that gives a significant advantage, there would be a strong incentive to keep that knowledge secret, which makes me wonder if it's already happening now.

The other question that arises from this line of thought is whether it would matter at all if markets became fully automated. I actually don't think we'd notice.

Wow that must have been pre-nostr, nice to see you here! Thank you 🙏

Lmk when you got something, would love to check it out

Really good idea

Super crazy

It’s interesting that asymmetric key cryptography is now the basis for both the private sphere and the public sphere via encryption and signatures, respectively

A strength in the sense that other platforms erecting barriers to their data precludes them from acting as a true "town square", leaving that niche open for something like nostr. I expect we'll see an eventual bifurcation of the social landscape into purer, separate implementations of both the private and public spheres, with many hidden invite-only groups on one hand, and on the other hand a single public global context (hopefully nostr).

Like AI is going to scape everything that's not totally private/encrypted anyway and so if there's something that isn't strictly private that's probably because you're intending to publish it, so why not maximize its reach?