the standard is described in IETF RFC5988

https://www.rfc-editor.org/rfc/rfc5988.txt

I would say that the HTTP `Link:` header w/ `rel={first,prev,current,next,last}` is the most standard way of implementing pagination. Typically cursor/limit style pagination is most efficient and is increasingly common. Linking to numeric pages is problematic because list size/ordering can change.

Got it working!

at the moment, trying to install some software from 1987, in DOSbox, for research purposes.

6% battery life and brew is playing chicken running clang. will the code compile before lights out? place your bets with the bookie.

yes. labels are useful, but they are just the lowest level data. it's like saying "we've built a system that can store and retrieve bytes! you could totally build a file system on this"... yep, but I need... files, directories, symlinks, rsync, permissions, etc. and kind need them now, not later.

how do you create a set of policies, as a community? once created, where is the logic of those policies encoded such that it can be distributed and applied in different contexts? a label is a product, but we need the means of production as data. we need the community and its values on the protocol.

to expand on that a bit: what exists from a data perspective is a label. a label is a 128chr string w/ a DID of the creator, a timestamp, and a URI to thing it's labeling. https://atproto.com/lexicons/com-atproto-label

the policies used to create these aren't modelled. how do you trust a label(er)?

yep. that said, labeling is not enough. it's the tiniest of baby steps.

Those communities, belonging to no one, kneeling to no corporate kings, should be portable across and alive upon the network. They should be able to make their own decisions, and define their own policies, according to their values.

The protocol should only encode, transmit, store, and apply that.

Moderation should not be the responsibility of the entities/people running the services that the data flows through, wether that's one or many.

The communities we form here, which exist both online and offline, and across many services, all have different values, and want/need different policies.

Further, BlueSky/AT Protocol would be missing a huge opportunity for progress if it was just a "better version of Mastodon/ActivityPub".

Moderation in Mastodon is still the responsibility of each server to implement. Many tiny fiefdoms, each with different kings and different laws. Yech. Fail.

This unfinished protocol is missing an essential part: moderation.

The question we should be asking is not "what policies should BlueSky implement" but "how can moderation be done in a decentralized manner".

How do we build that into the AT Protocol and how do we share this responsibility?

BlueSky's role is to create a full featured *protocol* for *decentralized* social media. Not an app. Not a single perfect service with a set of perfect policies. The app is here to test and inform the design of the AT protocol, nothing more.

And that protocol is still incomplete.

BlueSky is not and should not be a drop-in replacement for Twitter.

There's a sentiment that "if only we could do Twitter over again, knowing what we know now, that it would turn out better".

I entirely disagree with that. Twitter wasn't perfect, but they did their best with a centralized model.

any interesting bits from the #WWDC videos?