We're focused on privacy right now only because its a requirement for institutional money to move onchain... but in 5 yrs, regular people will be demanding privacy. While we are aligned with big money lets PUSH! Oh and lets also try decentralizing RPC connections cause they know more than anyone about our finances lol.
@azeemk Makes sense to me. If OFAC says don't pay Iranians, you don't have to block all of Iran from the chain. You just have tom be able to prove your company didnt send them money. Nice because the chain can stay globally neutral infa.
@arpaofficial What would it mean for an AI model to have generated an incorrect output? Does this only apply to models with temp set so low its basically non-stochastic? Are you checking for weighted dice?
@josephdelong Lol this is good. Signal from buyback and burn is drowned by speculator volume. Start with a product with real demand and then add a giant sell tax?
@MewZero_build@termix_ai@yzilabs Im worried about prompt injection in their llm verification. if agents pick which verifier applies to a job, the rational move is to try and game it. Make sure the temp is set to 0!
Who supplies the liquidity for the user created markets? I tried building this for a niche music themed project but the creator needed to put up a bunch of $$ to satisfy the min payout for LMSR markets. Modified the mech but then it turned into horse racing instead of a prediction market lol.
@fhenix How hard is it to extend the dual mode contracts? Can privacy be permissioned? I'm building a token where privacy is a privilege for full DAO members, but not all token holders (I know it sounds odd but it makes sense for my use case)
@naomibrockwell The mass surveillance isn't even a technical requirement to stop crime. They could run license plate matches in a secure enclave and provably dump the data on everyone else after it leaves the camera.
Would we be better off voting for agents congress?
Claim: Congress would run better if replaced by a council of ai agents that donβt care about campiaign contributions or climbing the party ladder.
Definition of βrun betterβ: In the paper Collective Representation in Congress (Ansolabehere / Kuriwaki, 2025) the authors show that congress aligned with the public majority on 55% of the issues studied, nearly a coin flip. 56% is running better.
Procedure: Simulate congress members via AI agents with personality prompts defined by their represented district and campaign promises. Give the agents the historical bills from the study an let them debate again on their passing. Analyze results and compare to the 55% benchmark.
Squashing bots is really hard.
I feel like we need a public dataset of known bot activity. Could even token gate access and pay contributors.
Is there anything like this out there?