Eduardo @hackerdocc

0x1A yo ; verifying zero-knowledge to make sure it's actually dumb ; organized TYPED Floripa ; prev SRE @bookingcom and Rust & cryptography @opactprotocol hackerdo.cc Floripa Joined July 2020

Tweets

8K
Followers

759
Following

1K
Likes

182K

Eduardo @hackerdocc

12 hours ago

@misterclayt0n i mean it's easier to either verify another language by having a formal semantics + extractor or to write an embedded dsl in a theorem prover that outputs code of the target language you want

0 0 4 134 1

View Details

Eduardo @hackerdocc

23 hours ago

@Aron_Adler this claude is autistic let him write his proofs

0 0 1 14 0

View Details

Eduardo @hackerdocc

a day ago

@Shinywcott 🤟😤

0 0 1 75 0

View Details

Eduardo @hackerdocc

a day ago

today i just conducted an experiment and the magsafe cable i thought was broken actually worked with my switch 2 brick. it appears the port on my macbook brick maybe got dirty? funny cuz i swore it was bad contact fixed by fiddling with the cable, maybe the issues compound?

0 0 0 67 0

View Details

Eduardo @hackerdocc

2 days ago

someone needs to fix this fucking mathlib cache situation. versions that aren't even old already have no cache being served. sometimes transient failures kick in and it fucks with CI

0 0 4 172 0

View Details

Eduardo @hackerdocc

2 days ago

@TheYisusOne champions graphics only appear good because pokémon graphics have been shit for more than 10 years now. it's profoundly unambitious. and they even cut themselves some slack by restricting the available pokémon

0 0 92 4K 1

View Details

Eduardo @hackerdocc

2 days ago

@craftsoldier for when they break

0 0 0 12 0

View Details

Eduardo @hackerdocc

3 days ago

who's making: apple care+ but it covers your magsafe cable?

1 0 1 77 0

View Details

Eduardo @hackerdocc

3 days ago

@AstleDsa nvm this is actually so fucking interesting? i may have an inductive bias myself as to which theories i tend to believe, but what they claim here is huge! i wonder if it could lend itself to optimize optimizers other than Adam

0 0 0 9 0

View Details

Eduardo @hackerdocc

4 days ago

i'm so excited to witness the scientific theory of deep learning start to exist

1 0 7 437 0

View Details

Eduardo @hackerdocc

3 days ago

@AstleDsa i was referencing this actually: arxiv.org/abs/2604.21691 but that was a great read too!

0 0 0 17 0

View Details

Eduardo @hackerdocc

3 days ago

the expression load-bearing is done for

0 0 0 60 0

View Details

Eduardo @hackerdocc

3 days ago

@JoeMerrick sure they have some animations and sure it's over a thousand models i still think they are lackluster. some animations are cute sure but the idle poses for many models are depressing

0 0 0 44 0

View Details

Eduardo @hackerdocc

3 days ago

@JoeMerrick no the models have sucked for more than a decade and there's no excuse for pokémon games to have this little care put on them. other game franchises which gross much less have better graphics, and random 3d artists online consistently outperform official devs

1 0 0 186 0

View Details

Eduardo @hackerdocc

4 days ago

who's going to extract an AGENTS.md to Lean and formally verify it's AGI?

0 0 1 106 0

View Details

Eduardo @hackerdocc

4 days ago

formal and cryptographic proofs are two different faces of an epistemological phenomenon regarding computation

0 1 17 2K 3

View Details

Eduardo @hackerdocc

6 days ago

@danrobinson @paradigm isnt this the puzzle that someone else had found the answer to, but the author had to legally threaten them to not reveal it as to not jeopardize the auction?

0 0 0 794 0

View Details

Eduardo @hackerdocc

6 days ago

@lucasleandro @OpenAI a minha chegou hj tb, n tava esperando q ia ser tão bem feita minha namorada odiou que agr o chatgpt sabe nosso endereço kkkkkkkkkkk

1 0 2 119 0

View Details

Eduardo @hackerdocc

7 days ago

sorry i don't know you and your previous obtuseness in engaging with the points i was laying out made me think you were bullshitting me in that case i just think your position is not intellectually honest then, reads like BigAudit dissuading people from verifying, especially when you cite a big practical win of formal methods authored by yourself while complaining it won't solve contrived theoretic instances that no one gives a shit about you were prompted multiple times to elucidate what you mean by "formal verification doesn't scale" and the only thing you came up with was citing that integer program infeasibility is coNP, and you couldn't even be bothered to write it yourself i'm not invested in trying to convince you anymore, since you know the theory and the practice, if you are still claiming that the existence of adversarial inputs for constraint satisfaction problems has any bearing on the practice of program verification, your biases got the best of you and there's nothing i can really argue that will change your mind the day i can't know whether my software has a bug because it's equivalent to proving an integer program has no solution i'll shout to the 4 winds that you were right all along