nit: not a worker bee

had to share these bizarre shops that someone made of me & @samuel.fm (tho i can't quote the post)

bsky.app/profile/luft...

idk what to make of it. it's up there with the time someone wrote erotic fan fiction about me & @pfrazee.com

ooh! yeah i've wondered what an ideal atproto handler would be 🤔

just resolving the data? not a great user-facing feature but useful for devs. maybe there's a way that a user's PDS could proffer some type of UI for the data?

in other words, atproto is a stateful universally addressed data protocol where the data outlives any particular invocation of a capability

(this isn't necessarily at odds with UCANs. but does imply a particular flavor of how to use them)

- related^^ but permissioned data isn't created through "invoking" a capability. it's published via the authority of the author & then included in the space by the fact that the space owner included the author in the space

- having a capability doesn't help syncing services know where they need to go to sync the data for a space. having an enumeration of all members is helpful

- spaces can have arbitrarily complex governance structure. both *on top of* the space in application semantics, and *under* the space with account ownership facilities in the space host. i think not trying to capture this in the protocol is actually the best & most flexible option

- spaces can reconstitute in a similar way to how you describe! all of the data is on members' PDSes. if a DID is lost somehow, members can all migrate their data to a new space DID. tough social coordination problem, but i don't think any more difficult than the example you give

- i don't think records are a good place for UCANs to live. they are fully public & end up leaking a lot of metadata about the going ons of permissioned spaces

- this provides a mechanism for how revocation works but it doesn't solve the hard problem of revocation. ultimately you still need to verify every UCAN in the chain any time a capability is invoked. that's expensive on every write & complex during network partitions/outages as your post discusses

a few particular things:

- I'm not sure i totally get the diff between space & circle DIDs in this writeup. sounds like a circle owns one or more spaces? do spaces then operate similar to my existing proposal or do they operate off of UCAN as well?

hey thank for the write ups & engaging so deeply with the design here!

i worked on the original design & spec for UCAN and still 💙 them. but i'm not sure it's the right fit here

ACL v capabilities is an argument as old as authorization. idk if it'll ever resolve one way or the other 😅

Protocol Governance & Hard Decentralization - ATmosphereConf 2026 YouTube video by AT Protocol Development, Tech Talks, and Events

yes!

www.youtube.com/watch?v=9z0z...

i have a half-finished blog post sitting around about exactly this that i'll hopefully publish At Some Point Soon™️

anisota is of of the most interesting things in the atmosphere to me rn. not because i think it'll be the big breakout app (it's almost allergic to that), but because it gives the blueprint for the kind of hyper-specific, bespoke, artistic, or contextual apps that are only possible in the atmosphere

if you came to my talk over Dame's, you made a grave error in judgement that you can finally rectify by watching this video

@ 🤝 🌎

modelo time w the good folks at germ 🍻

bsky.app/profile/mack...

what's your ai strategy?
what's your quantum computing strategy?
what's your ww3 strategy?
what's your collapse of consensus reality strategy?
what's your restructuring of global supply chains strategy?
what's your pedophile billionaire cabal blackmail ring strategy?

"The atmosphere has a window, perhaps a narrow one, to become that kind of foundation, to build infrastructure whose affordances will shape what becomes possible for communities that do not yet exist and cannot yet advocate for themselves."

this weighs on me heavily^^ (in a good way)

this captures it perfectly

pleasure to meet and talk with @laurenshof.online in person

> grown man experiences a brief moment of whimsy

"the only solution to this is ground troops in iran"

just curious, do you mean at the protocol layer (something like multi-sigs) or at the implementation layer (something like multi-user control of an account on the PDS)?

yeah i think that would make sense

websocket framing is one of the big things @bnewbold.net wants to consider changing

Too early for that, tho the wg can always recharter

If the wg did permissioned data, we might not have the protocol for another 3 years 😉

it’s on a branch

And someone can run a “sync hub” or relay for a space and accept space credentials but that’s not described by the protocol. Maybe it eventually could be if it becomes common/necessary?

That’s basically right. I expect server side by applications to be the main (and default) path. Tho client applications can hold a space credential in just the same way (folks from @habitat.network are working in roughly that model & we chatted about it a lot at atmosphere conf)

happy for folks to follow along, but *please* don't over-index on this. it will have massive refactors & breaking changes

i'll try to be upfront about how stable the code is as i go & the current answer is "0% stable" lol