ATProto sounds a bit like Pubky but without pkarr and pkdns, but with added complexities, which makes me wonder if Pubky can scale to Bluesky's current numbers. Just armchair speculation from me.
Also the costs and complexity seem like an overall L for ATProto and it seems clear that it's "decentralized" between large entities only.
Thing is nobody seems to have a useful scorecard for decentralisation. For ATProto, Pubky, Nostr, Farcaster, ActivityPub, Keet, etc., each has certain "decentralisation categories" where it's the winner, and each has certain decentralisation categories where it's the loser.
Far as I can see ATProto's ethos seems to be that relatively few people in the grand scheme are going to put up with this fragmentation stuff and therefore fragmented social networks are never going to get past the size of, say, Mastodon. And if they never get past that size then what’s the point?
The scorecard is keet is on top and everything else is less decentralized
Far as I know, as of today there aren't any alternative clients that are interoperable with keet via holepunch, giving it a bottom score on the client rug-abliity metric. I don't even think you can export contacts and whatnot out of the keet app. Unless things have changed.
I believe they have published enough source that it's quite possible to make an app that could interact with keet, I have seen others do such things since a long time ago
Your "contacts" in keet are the rooms you are a member to, so what ever keys that allow that plus your own local hypercores. The function to make a copy of these "contacts" is to link another device which will then replicate them all on that device which is technically just another instance of keet whether it's on another device is not actually important. And the action of linking the "device" is actually a new identity joining each room with one overall identity linking each of your devices/identities in the room. Interesting point your identity per room is unique opening the possibility to have a different screen name etc per room but this has yet to be implemented, it's just the foundation has been built for it
If you wanted to get into somantics it would be possible with keet to store the info necessary to join any room in text and then store that in keet or any app
As to rugability, it's built on a completely open source and P2P foundation meaning there is no servers to maintain ie it can't be shutdown, yes parts of the app are not currently open but there is certainly enough open to build chat apps and you should be able to make them keet compatible, if you have the keet install files which are publically available you can keep installing and using keet even if holepunch the parent company went bankrupt and stopped, which you can not say for things like signal, WhatsApp or telegram and most other apps
This all may be true but it demands a LOT of work and know how. For 99% of keet users, if keet goes down then that person's holepunch world goes down with it.
Same for Bluesky, for 99% of users if Bluesky the app goes down their ATProtocol world goes down with it, even though in theory you can restore your PDS to an outside host, appview and relay running on independent infra and it's also all open source too.
None of this is the case with nostr clients, one goes down then the bulk of the users can simply access their nostr world from another client, in five seconds, no technical knowledge needed.
Keet can't go down tho? Do you understand how it works? It's P2P, if you have the app it works, it has no servers
If you have the install file you can install it and it works, again no servers
If you join a room and use it you make another copy of it, you can give that copy to anyone
How are you imagining Keet stops working or goes down?
Holepunch Inc, the company, ceases to exist, or pulls the app from the stores.
For 99% of iOS users, that is effectively the end of their holepunch expereince. They cannot invite friends (who also not tech savvy) to install and join them for a chat, and if they lose their local copy of the app (get a new phone, etc.) that's the end of their own connection too.
Yes 1% might compile from code and install via their own personal testflight or whatever, but when talking about general adoption of decentralised tools there is no sense talking bout that 1%. Most people don't even know what the word "key" means, let's not forget.
iOS is a stawman argument, it's a small platform with limitations, where as the other platforms can keep installing if holepunch disappear
The vast majority of normies (or semi-normies, seeing as normies aren't there yet) using Holepunch are using Keet on iOS.
I myself am not interested in protocols that normies will never care about. Sure some tiny number of developers can do whatever they want with some open source code they pulled from Github before a repo went down. But that's of very little interest to me, it's essentially little more than an open hackathon.
Developers are fond of saying "you can just do this" and "you can just do that" but we all know that normies will never ever "just do" these things.
