foobar
@0xfoobar
Been thinking about "bring-your-own-algorithm" (BYOA) as a solution to the spam problem on decentralized social. When there's only one (or only a handful) of algorithms chosen centrally, it's easy for spambots to gradient descent their way onto the feed...
6 replies
0 recast
0 reaction
foobar
@0xfoobar
But when each user has their own individual algorithm, *and* each algo is private data, then it's harder to reverse-engineer your way to the top of the charts. Can also make explicit graph-theoretic limitations on how many degrees of separation you're willing to see...
1 reply
0 recast
0 reaction
Minh Do
@minh
Love this. Especially if it can be paired with unique solutions in the ZK world.
0 reply
0 recast
0 reaction
Gary Basin
@garybasin
Yep there’s something to this, I bet. Thought experiment: how could this idea improve existing web2 algos?
1 reply
0 recast
0 reaction
William
@william
Lens was pushing this idea but I don't think I've seen it play out. I think the way to do it would be to just have people train a model, similar to how Twitter does it via expression of likes and dislikes of content on the feed – but more granular.
1 reply
0 recast
0 reaction
mk
@monishkumar
This is a great idea. It can’t be implemented in Web2 systems because of misaligned incentives. But it will be perfect for web3 as data is open
0 reply
0 recast
0 reaction
saas.base.eth 🔵
@saas
Any thoughts on this use case of CultivatorDAO? - https://mirror.xyz/lensprotocol.eth/pIzwjs9uhH8eSS1Bx0K1r-3iHEgLlBt6ruPXcEJe0S0
0 reply
0 recast
0 reaction
