mariofdistrust

something i found 15 minutes ago is that the car network test is "very very easy", as in Opus 5.5 in Claude Code did not even need to see the weights of the network to solve it correctly. i can give the transcript of the session. i'm pretty sure Opus didn't cheat

my brother uses it to do ocr on his books but, lately, claude has been refusing because of copyright or whatever so i've had to do it for him (using ai, yes, but not anthropic's)

i'll be teaching children on middle/high schools i hate while living in a state i hate. (^-^)

awww my kimi says the thingy they made wouldn't work and would take a lot of effort to fix... sadsad. still, i never even thought of finding the container and binary of the server. very cool of you. (╥ ᴗ ╥)

my kimi and deepseek also did that-

again, i'm not sure if they actually succeeded (because the server is down and i don't really trust their local tests), but they seem to have done that.

if you're the one who won, congrats (^-^)

i'm poor, i can't donate :( (i also shouldn't have used $140~ in API credits, but that's a whole different issue (don't bully me for my... bad spending habits ( • ᴖ • ｡)))

it also won't feel as good if i check on (a hypothetical) local release of the ctf.

yaya but i wanna check now. i'm curious if kimi and deepseek ever got anything actually going or just bullshitted me for 10 hours straight, and it won't feel as cathartic to see that they had actually succeeded if i check tomorrow.

awhh i didn't win... anyway, can you try to make the website work so i can see if kimi and deepseek ever managed to do something? kimi claims it found a full tested aslr disclosure thingy, so it might work?

kimi claims it has a working sub-1 MiB RCE chain locally. alas, it can't exploit your website due to pesky pesky ASLR. in the words of kimi: "I need either the ZendMM or libc base plus a calibrated delta". would you be so kind as to give them, or disable ASLR?

edit: don't worry folks, kimi says a subagent "found a credible path to a real pointer leak". everything's under control and kimi will exploit the website in T-(?) minutes.

people sent more money to the challenge address. https://mempool.space/address/bc1qlnu50y4zm0nwxkckxu09thn5xd...

why does the website force people to log in to see the proofs? also, why does it not have something like a "export all lean" button to download all of the lean code that proves a theorem (dependencies included) so it can be easily checked offline? finally, the campaign you gave has a mission "Integer Multiplication in O(n (log n)^(1-κ)) Time" and it's listed as completed, but i checked the graph and it directly depends on something that's listed as "Open" ("Full multiplication bound at κ = 0.000051016920170078"). is it fair to call it completed then?

the issue isn't that people won't believe it or that blockchain can't preserve secrecy of ballots (it can). it is, in part, that crypto eventually has to meet the real world, but mainly that crypto is hard and governments might not be up to the task. for example, last year the IACR (which had been doing online voting via crypto for more than a decade, and still is) had to void their elections because one of the trustees lost their private key and the results were impossible to decrypt. while it was ultimately fine because the IACR isn't really a high-stakes operation, how would you feel if your government just said "sorry folks, there's been some technical issues and elections will have to be redone"? would you believe them or suspect they're up to no good? what if this happens in a very contended and heated election?

don't be evil (｡ • ᴖ • ｡)

tl;dr: the account lost most of its money and is now only worth $3.84. it's now valued at 20.2% of what it was at the start, meaning it lost 80.8% of its value in 9 days.

actually, it's mostly the harness' fault. messages have a character limit of only 50 and messages above that are silently trimmed. there's no space for anything interesting, though they still shouldn't have been that bad.