Content
@
https://warpcast.com/~/channel/opacity
0 reply
0 recast
0 reaction
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
Yes we can https://dwr.email
6 replies
6 recasts
18 reactions
Royal
@royalaid.eth
Do we have some supplementary reading material on how?
1 reply
0 recast
0 reaction
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
Would need some design iteration with @sanjay before i stood by a specific implementation. Celestia wants to use zkTLS for an identity based priority ordering for sequencers. Prove your bank balance is at least 5k and your txs are prioritized. —- My gut tells me a similar approach would work. You’d define the identity signal scoring somewhere and high signal accounts would be prioritized in the mempool. If the mempool is just casts with no connected identities the system behaves as it is today.
1 reply
0 recast
1 reaction
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
@v I’m guessing this is a defensive measure against spikes? Because the per year pricing doesn’t keep mempool healthy
1 reply
0 recast
0 reaction
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
@ncitron.eth wdyt
1 reply
0 recast
0 reaction
ncitron.eth
@ncitron.eth
I suspect a combination of things like WorldID and zkTLS could achieve a decent spam metric, but the bigger question is should these systems actually be built at the protocol level. Honestly don't know the answer. Dan seems to get the risks though of popular clients having defacto control of what gets reached in a non-protocol based spam detection. But if we did it at the protocol level, would we be buying anything of use? Clients can still put their own filtering on top of that, so arguably a protocol level spam filtering solution would only provide a base level of filtering and doesn't enforce what subset of the post-protocol-filtered content is shown. What does protocolizing that actually buy us then? It doesn't enforce any notion of censorship resistance (because clients can still filter you). I don't see the benefit of buying this "base filtering" from a company vs protocol.
1 reply
0 recast
3 reactions
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
My guess (@v correct me): Snapchain does pricing as per user/year, not per cast. Tradeoff for better UX is no mechanism to keep mempool healthy. If Dan wrote that article because they were trying to solve for this, then it would have to be at the protocol level.
1 reply
0 recast
1 reaction
ncitron.eth
@ncitron.eth
Well there's the storage unit system as well. But you're right that the existing protocol doesn't really charge for usage during bursty periods. Using some identity system for priority during these periods is actually a good idea.
2 replies
0 recast
1 reaction
EulerLagrange.eth - shitpost/acc
@eulerlagrange.eth
This identity scoring contract could also double as an airdrop center.
0 reply
0 recast
0 reaction
christopher
@christopher
Sounds like y'all are describing user scores, which are a solved problem in traditional social, though applied privately. Content is much harder to filter because context matters and is $$$$ at scale. But there is a lot of progress in fraud ML and how fraud rankers fold scores into the consumer interfaces. cc @gt
0 reply
0 recast
2 reactions
