blakemscurr

@bennhoffman @noworkfunction @eigenrobot Yup, it'd be great if there were a timestamping service as widely used as, say, archive.is. Could at least limit the scope of discussion. Second solution feels flimsy, but I believe there are protocols in the works for it. @boneh/using-zk-proofs-to-fight-disinformation-17e7d57fe52f" target="_blank" rel="nofollow noopener">medium.com/@boneh/using-z…

@noworkfunction @eigenrobot Easily to establish existence at a point-in-time. Other solutions probably involve hardware signing of media as it's generated with manufacturer controlled keys, which poses new problems.

it's possible that the next time something like this happens we won't know which videos are real and which are pure ai-generated it's further possible that we won't know whether any of them at all are real contemplate this on the tree of woe

@dystopiabreaker Didn't this already happen like last year?

it seems pretty apparent that “prompt engineering” will just go away because model pipelines will just git good

@Jeyffre Yeah, and these vary hugely in how fast you can get up to speed. I think for an average interested developer it's something like: Type 4: days Type 3: weeks Type 2: months Type 1: years

Just like there are 4 types of zkEVMs, there are four types of learning zero knowledge proofs: Type 4: ZK DSLs. This means coding in languages that try to abstract as much of the zk away, like Noir, Cairo, Zokrates, Leo, etc. You are “learning zk” but only in a loose sense. Type 3: Designing circuits. This means you are actually creating constraints to encode the application’s functionality. This is a very different kind of skill, as “thinking in constrains” is not a thing developers do regularly. Type 2: Coding proving and verifier systems This is where most of RareSkills focuses, how to prove a circuit is pretty orthogonal to designing the circuit, and both aren’t easy. Type 1: Research You’re publishing academic papers and writing and math proofs. You write formulas and the engineers implement it. My suggestion is to be intentional about which type you are studying, try to get a good sense of one before moving to the other.

@portport255 @p_tsanev @pashov @code4rena @zksync Makes sense, can you say anything about the language/proving stack?

@p_tsanev @pashov @code4rena @zksync The scope for this contest includes more than just the circuits. You can think of the circuits as connecting the L2 to the L1: a ZK Proof is created from the circuits, and then verified on Ethereum. We'll be releasing more documentation soon :)

The folks at @code4rena just announced a $1.1M decentralised security contest. Correct me if I'm wrong but this is the biggest reward pot we have ever seen? Time for some people to go all-in on ZK studying

@odysseas_eth @ProofOfZk @ameensol With plume I think you could take a snapshot of key holders at the time of poll creation, and the proof would show that you own an address in that list. That should solve the registration time issues🤞

@odysseas_eth @ProofOfZk @ameensol I think you could replace semaphore with plume and you wouldn't have to interactively create nullifiers. A plume signature is unique for a given user and message (in this case poll id), and it's still unlinkable. Wallet support is an issue though. github.com/plume-sig/zk-n…

I've looked around and I'm pretty sure this isn't possible with existing zk tech 👇 maybe some day, oh well

@dystopiabreaker Huh, seems about right. 2 people is a model 3 is ~2000lbs of vehicle weight per person, and a R32 (subway car) is ~1500lbs/person with all seats full and no one standing. Still, cars bad: noise, killing people, concrete wasteland.

(per passenger)

two people in a tesla is more energy efficient than most urban rail and far more energy efficient than amtrak in the US

@justanowlet Beethoven's 7th symphony 2nd movement. Tannhäuser overture. Mozart requiem in D minor (the lacrimosa part is particularly epic). Grieg, Åse's Death.

I'm on my classical music era, recommend something epic

@aryehazan What has changed lately that makes the wet market more likely? In mid 2020 I was pretty convinced that it was a lab leak for the reasons you stated, and the lack of evidence for the wet market hypothesis at the time and I haven't updated much since.

We live in an age where an intelligent, curious, generally informed person can truly have no idea how to process various important, pivotal events. [This was probably always true, but a lot less apparent with the near-monopoly on information outlets.] One such example is the lab

@nullity00 I think there might be a small mistake though. Instead of `g^{s+hash(msg)}`, I think you need `g^{hash(s+hash(msg))}` or something. Otherwise you could just pick a new message and change `s` accordingly.

@nullity00 Oh nice, I feel like it could be a good game mechanic. Secretly burn eth (as an offering?) to make a move in the fog of war. Curious to see if there are more practical defi/protocol use cases though.

👀

@HamishDoodles The lonely lonely water hemisphere

@DrugGovoruna @orthonormalist Yeah, @orthonormalist, if they happen to have studied IPs or PCPs they should be able to get a cryptography-ish position working on ZKPs at an Ethereum L2 like @Starknet @aztecnetwork @zksync @Scroll_ZKP and many others

@orthonormalist Something having to do with crypto and ZKP?

I have a friend with a computational complexity PhD but not a ton of years of industry experience on the market. What would be the interesting problem and income maximizing route for him to take?

@levs57 go to. That's not a uniquely attributable fault - I don't think mainnet can tell who is withholding data from whom. If we can't tell which successor is acting badly, I don't know if we can do a cascade like you describe.

@levs57 Hey, sorry for the late response! So how is failure defined here? My main confusion is that if there are two successor validia that are trying to claim from the same predecessor, they might claim the same specific coins, but withhold the proofs that show who the coins should...

Talked about it with @BlakeMScurr, I think it is a really neat idea, because it sidesteps the DA problem completely. Ideally, we could even have multiple such validiums so exodus is not on a main net, but to other such validium. If enough independent parties run this, it will

@donnoh_eth @kobigurk favour doing things offchain where the throughput is practically onbounded.

@donnoh_eth @kobigurk @donnoh_eth in ArcPay we're not actually trying to determine who's at fault for a DA problem, we're just using L1 as a fallback to post data in case someone wants to forcibly get the data. That means we're limited to L1 level *forced* data throughput, but the game theory should..

ArcPay by @BlakeMScurr is an interesting design - it’s (currently) limited to payments, but has a trustless exit /shutdown mechanism that avoids the usual problem of locked funds in validia (love this term, saw it in the post) @donnoh_eth maybe it’s better than a magma chain :)

@bkiepuszewski @optimismFND Is there a good post explaining why they had to rebuild their whole stack? Looks like a little discrepancy between the L1 simulation contracts vs what the sequencer was running - I'd have thought those bugs could be squashed more incrementally.

Hey, anon, if you are passionate about @optimismFND and their fraud proofs strategy, this is an essential read that provides a lot of nuance and important context. Note the date of the article. Seems like from years ago ... medium.com/infinitism/opt…

@eth_call I love reading this without knowing hoon, it's like a Lewis Carroll poem. A contract is a door with two arms?

@BlakeMScurr Yeah, it's the smart contract language for Uqbar: uqbar-network.gitbook.io/dev-docs/devel…

once I learn Hoon it's over for all of you