🏅 Top traders

Thread of shenanigans: https://x.com/ptrschmdtnlsn/status/1939750416243466544

The secret was 101000100010111011011001101100111000001000111101, so YES, it ends in a 1. Thanks for playing, everyone! I'm going to post a Twitter thread with all of the highlights of people doing shenanigans tomorrow morning. If you don't want me screenshotting your trades, for some reason, I guess let me know.

Started cracking last evening, woke up to my crack being complete and apparently lots of other people also. hashcat reported a full bruteforce on the full 48 chars to take 16 hours and with the known 2 first bits 8 hours, apparently having fixed chars roughly halfed my hashrate.
This was on a 7900 xtx goind about 4500MH/s with no predetermined chars and about 2200MH/s with the first 2 chars locked in

@JohannesJensen Yeah every known bit changes the search-space by a factor of 2^-1. The trick I, and a few others made, is as a human making a guess at the last digit and doing that search first.

preimage is 101000100010111011011001101100111000001000111101 (out of mana and want to win the related >90% market as well)

@robertvc Everyone, just buy the next day and wait till after midnight.

@ThomasMach good idea, although I'm not sure if the other market is liquid enough / whether it's possible to fill enough volume into the other market to justify the spend required to pull this one down? I guess depends on who finds this in the next hour

@robertvc You are probably right, unfortunately.

@robertvc I'm surprised it has taken this long for a public solution.

1. I found it almost a day ago
2. Since the Q is "is the last digit a 0 or 1" one could make a guess that it was 0 and solve it in half the time
3. The yes-side has been dominated by new users (such as myself) making big bets, indicating great certainty

@MartinSundhaug broadly agree, which is why I guessed 1 ;)

I assume for example that you guessed zero and therefore learned the answer but did not get the proof (preimage)

@ThomasMach btw, you're still on the "NO"-side with a huge investment

@robertvc Correct. I sent Peter a Twitter DM about it yesterday, and then verified it a few hours later. After I had the solution (though I wasn't sure I had done it correctly, since it was a matter of exclusion), I also told him I thought he had decided to make it an "is it odd" Q after making up the string

@PeterSchmidtNielsen will you resolve this market early now that we know? 101000100010111011011001101100111000001000111101

@dcxStep I will wait until this market has a clear resolution: https://manifold.markets/PeterSchmidtNielsen/when-will-my-question-about-finding

Then I'll resolve it after that.

@PeterSchmidtNielsen good stuff, hopefully before the end of the season :-)

@MartinSundhaug I was trying to protect my bet that the market here will not be above 90% for more than an hour on 29 June. This was cheaper, than loosing all my mana on the other market.

Looks like this can be done in a couple of hours on some 4090s, so I'm going to get cracking.

@robertvc certainly an idea

@robertvc Lucky crack! Took only 50% of the expected time.

@Na3wt54zergd then why are you buying NO

@Na3wt54zergd Huh? the market isn't for "does the preimage exist", it's for whether the last digit of the preimage is 1

@robertvc Uh... I got a different answer from you.

@jessald lmao, the jig is up! buy buy buy

@robertvc lol nice
I didn't make much but it was a fun learning experience.
$1 to rent 4x RTX 5090's on vast.ai for about an hour

@jessald sounds about right, I spend about a dollar in electricity at SF prices 😭

Also a cute trick is we can halve the search space by searching only hashes that end either one or zero but not both (!)

@robertvc If you do that and get unlucky (and don't have the exact answer), it's slightly harder to convince others that you are right.

@robertvc Yeah that heuristic was part of what helped me, I'm surprised nobody reacted to people making big bets