Ancestors

Written by notplants on 2025-01-31 at 17:02

i feel like some descriptions of crdts are making impossibly utopian promises in confusing technical and human "consistency guarantees" (or maybe im missing something)

=> More informations about this toot | More toots from notplants@sunbeam.city

Toot

Written by notplants on 2025-01-31 at 17:03

e.g. this article https://vlcn.io/blog/intro-to-crdts (not making the claims, but as a background for talking about this)

"""

A CRDT is a data type that can be:

- Copied to multiple machines

- Be modified independently by those machines without any coordination and for any length of time

- All divergent copies of that state can be merged back together in any order and by any machine. Once all machines have seen all divergent copies, they're guaranteed to have all converged to the same final state.

"""

however, if you think in more human terms, if a group of people is collaborating on a document (e.g. a document of by-laws), and two different sub-groups modify the document in different ways, there wouldn't really be a guarantee that their changes could be merged in a coherent way without an additional human process of conflict resolution. this is a not a technical problem, but rather an inherent property of resolving collaborative decision making processes.

=> More informations about this toot | More toots from notplants@sunbeam.city

Descendants

Written by notplants on 2025-01-31 at 17:03

"One of the simplest CRDTs is a set that can only grow"

-- this makes sense, but not every collaboration or process can actually be mapped onto "a set that can only grow"... that is a very specific subset of all collaborative problems

so what I'm really saying here, is using CRDTs is not just a technical issue, its also an issue of figuring out what collaborations actually make sense as CRDTs, which isn't something ive yet seen written about much

this feels like the "not every problem needs a CRDT" lemma from the similar "not every problem needs a blockchain" issue

=> More informations about this toot | More toots from notplants@sunbeam.city

Written by notplants on 2025-01-31 at 17:15

this thread could be seen as an argument towards more solutions like @cblgh social-norms-based crdt-replacement idea

=> View attached media

=> More informations about this toot | More toots from notplants@sunbeam.city

Proxy Information
Original URL
gemini://mastogem.picasoft.net/thread/113924046021684216
Status Code
Success (20)
Meta
text/gemini
Capsule Response Time
261.355544 milliseconds
Gemini-to-HTML Time
1.073455 milliseconds

This content has been proxied by September (3851b).