we are fairly solidly against ActivityPub-style ACL on content in a public repo (which is basically really all public, the ACL is just friction!) or encrypting records in the public repo (don't persist cyphertext!) as specific design approaches.
there is a large design space and we will get there!
that isn't to say we are against group-private content being part of atproto, it should just be a different mechanism from "public repos". maybe multiple repos? maybe PDS-to-PDS? still designing. but probably not a "hack" or tweak to the current repo architecture
for repo content, the atproto philosophy is pretty strongly that it is either accessible to the whole network or none of it. we don't want to do access control on public repo content. infra and services can filter out specific repos or content, but that is different from the account deciding
on the one hand, "self-takedown" is basically how we are planning to implement account de-activation. we do want that to be distinguishable from an actual takedown or deletion though.
a thing that would be good is somebody doing shared-hosting or helping labelers get started, similar to skyfeeds; but have it be an "apply and approve" flow, not just totally self-serve?
we have discussed doing such a thing, but wanted to ensure *some* fulling indie/self-hosted early on
apart from bad actors / harassment, there is some responsibility to respond to appeals and be accountable. do want communities to jump in, but don't want well-meaning folks to get in over their heads and/or burn out.
yeah, we definitely discussed and thought about this when designing labelers, and how much effort we wanted to put in to the self-hosting setup.
eg, feed gens we really want low-friction, labelers is a balance between accessibility and some friction
have you ever gone to a bar and gotten a tuna head and spent a couple hours picking out yummy bites while drinking beer? that is a fav of mine
and, summoned by mention, had another 200+ come through just recently (!)
thanks for the write-up! in interesting inter-net situation.
if all of nostr was bridged automatically, would we need to moderate all of nostr? hrm.
awesome!
we only enabled proxying recently, and for self-hosted PDS instances I think just pushed out that feature in the past couple days?
sometimes DID docs are cached for up to 24 hours as well
support@ is humans and the canonical inbound support channel.
if you have technical experience you could try opening a github issue in the app repo, or poke a front-end dev like paul?
december 26, 1996 after a good meal
june 3, 1997 is incredibly @dril.bsky.social
we would get joe wenderoth on here
archive.org/details/lett...
alex grey painting of somebody sending a big line, colorful vibes emanating outward
one of the great use-cases of the not-E2EE DMs we are working on is to privately share contact info. take it to Signal!
eg, is "Bluesky the app" an aggregator like FriendFeed? or does somebody build an aggregator which consumes both Bluesky content and whtwnd content (and video, podcasts, etc build on atproto) and combine them?
en.wikipedia.org/wiki/FriendF...
but maybe that is where things are today: Bluesky (app) *is* the main experience on the network, and thus closer integration is necessary.
in the future maybe there will be more profile-oriented views/apps (as mentioned in the essay), or generic feed-reader UIs which aggregated many apps
a discomfort for me is that it feels like devs shouldn't be reliable on any other team (like Bluesky) for their atproto app/service to be successful.
and also a blurring between "bluesky the app" and "atproto the network"
a negative example is farmville and zynga on facebook: it changed the experience in a big way, and facebook eventually pulled the rug.
to be clear, whtwnd and blogging are positive!
one train of thought I have is that social media design has some similarities with game design. I want developers who are very opinionated about user experience to be able to build stand-alone apps on atproto and have near-total control over the experience, similar to game designers
cross-linking a related conversation thread you might be interested in @laurenshof.online
@bnewbold.net Hi!
I've written a reading material about atproto.
Please feel free reading it when you have free time!
It might not directly answer your questions since I don't have clear idea either.
I am a developer who wants to create a content platform.
I am thinking about how to energize my platform.
First, I decided to think about what my target users are thinking and how they can contribute....
cross-linking a related conversation thread
I think for that to happen the bsky app needs to allow other lexicons in the app. But since they want to recreate twitter, they wont. Starting up a new lexicon on ATproto currently sucks since you cant leverage the buildin userbase.
thank you so much for writing this piece @knksm5.final-techblog.com. it touches on some really important parts of our goal with the protocol which we discussed a lot in the beginning, and raises a bunch of new (to me) questions.
I am sure I will re-read this blog post many times in the future!
""I would like to see something that appears attractive to someone swinging a knife, making them want to put the handcuffs on themselves"
incredible quote!
it depends on what folks want to do, and when we design for some random MVP thing it becomes normative.
eg, maybe mods want to give a bunch of examples for each label, with screenshots? or other exemplar images, but then those need content/trigger warnings, or over-18 concerns
we ended up cutting down the description length of labels and labelers significantly, which removed space for this sort of context in-app.
I sorta feel like out-of-app website or gdoc or something would let folks really get in to details w/ rich media, formatting, i18n, etc?
endorse. great to have this story and timeline laid out clearly!
Okay, I wrote a big thing regarding Jack Dorsey and Bluesky. It explains why he's actually more correct than people give him credit for, but still absolutely wrong about Bluesky. www.techdirt.com/2024/05/13/b...
There was a bit of news in the world of decentralized social media over the past few weeks. It kicked off with the announcement that Jack Dorsey had left the board of Bluesky. This was followed by …
www.techdirt.com
there are def some things we would do differently in a v2 of post schemas.
going forward, should always include the full pristine original record as a separate field in views; this is important for extension points
(handing off appview-to-appview is maybe technically possible but feels off to me)
the appview stores and applies labels, which is important for performance and reliability (don't want an ozone server to melt down from traffic), but they are "canonically" stored by labeler.
important for new appviews to have backfill access to all old+important labels
there are some new atproto apps being built now, like whtwnd.com and a voice/audio one
stayed up late for aurora but no success.
think we got a bad batch, threw up a couple times but no visuals
ok, think we addressed the bleeding on this.
these are nostr accounts being bridged in via the fediverse. they are obviously very low-effort. have more levers if it escalates in to a cat-and-mouse game.
ok, think we addressed the bleeding on this.
these are nostr accounts being bridged in via the fediverse. they are obviously very low-effort. have more levers if it escalates in to a cat-and-mouse game.
ok, think we addressed the bleeding on this.
these are nostr accounts being bridged in via the fediverse. they are obviously very low-effort. have more levers if it escalates in to a cat-and-mouse game.
cc: @snarfed.org @rabble.nz @npub1wmr34t36fy03m8hvgl96zl3znndyzyaqhwmwdtshwmtkg03fetaqhjg240.momostr.pink.ap.brid.gy for nostr/bsky dynamics interest
we expect spam/abuse over bridges and will try to handle this account-by-account unless it starts causing harm
ok, think we addressed the bleeding on this.
these are nostr accounts being bridged in via the fediverse. they are obviously very low-effort. have more levers if it escalates in to a cat-and-mouse game.
I can download that repo CAR file, but when I try to list the records or unpack with go-export-repo, I get:
error: loading root from blockstore: ipld: could not find bafyreighnzge6oitpthd7upzfslpom5lw6mpekwgc677gs7fajihsvw3qe
some related tools:
atproto.tools/records?did=...
bsky.jazco.dev/walker (failing?)
github.com/bluesky-soci...
I have a little "fetch entire repo and validate structure + records, by DID" CLI tool, almost done but not quite
nice! ooo an event in croatia, very tempting
OAuth is in the works; a bit complex but there is a full proposal (linked) and implementation has been underway for a while.
Bluesky proposal discussions. Contribute to bluesky-social/proposals development by creating an account on GitHub.
github.com
... but not a bsky *takedown* at the appview.
we intentionally created many shades/layers/rungs of takedown, to give room to maneuver, both for ourselves and other devs/projects in the network