Topic: Deleted (Read 10072 times)

Congratulations! Man that took a lot of time & effort! I was lightyears behind you guys so I gave up a long time ago. Well deserved!

well, a bit late, but hey congrats! you deserve it for the work you put in. but i guess it was a worthy experience and a nice way to learn some coding.
i really hope there are more puzzles like this out there.   

 I didn't think anyone was ever going to solve the puzzle.  Perseverance was the real key
Congratulations! 

Congratulations And I really appreciate uploading that script. I'm learning python and I expect that script will help me to learn even better

lol, thanks for sharing. my only missing character was jigsaw which i could never think of starting with j because i thought it is a puzzle piece
apparently my English sux


♯♯congrats.

Well... maybe they're not THAT safe (if we get clues )...

So at around 1509 hrs UTC on 28th September (around 0409 hrs on the 29th local)... my script spat out another possible hit... I was working a night shift and so I didn't actually get around to checking it until around 8 hours later.

Amongst the 3 new possible hits that had been generated, one started with "1qk...."... could this be it?? I quickly checked on blockchain.info... "Final Balance: 0.5001 BTC"... WOOOOHOOOO!!!!!!!   So I quickly imported the private key into my wallet and transferred my newfound loot into my own personal address.

The private key turned out to be: 5JKPapJwgyEij3sxYRAEnixyiFgxqkVhgZXv9bWWknBexegx6tM

• 5 - 1_3_5
• J - Steve Jobs
• K - Kevlar
• P - Pi
• a - Atari
• p - parachute
• J - Jet
• w - wing
• g - Johannes Gutenberg
• y - yoyo
• E - mc^2 = E
• i - iron #lowercase BLUE
• j - jigsaw
• 3 - 1836 Morse Code Invented
• s - scissors
• x - x-ray #lowercase BLUE
• Y - Linus Yale #UPPERCASE lightGreen
• R - radar
• A - Aspirin #UPPERCASE lightGreen
• E - Eiffel Tower
• n - Nobel
• i - ice skate #lowercase BLUE
• x - XX chromosome
• y - XY chromosome
• i - igloo #lowercase BLUE
• F - film
• g - Galileo
• x - Xerox
• q - Question
• k - Nakamoto #lowercase BLUE
• V - vaporiser #UPPERCASE lightGreen
• h - Satoshi
• g - Gramophone
• Z - zipper #UPPERCASE lightGreen
• X - alphabet X
• v - viagra #lowercase BLUE
• 9 - Euro - 1996
• b - Braile for B
• W - WiFi #UPPERCASE lightGreen
• W - Windows
• k - krypton (periodic table) #lowercase BLUE
• n - DNA #lowercase BLUE
• B - balloon #UPPERCASE lightGreen
• e - Einstein
• x - 2009 MMIX #lowercase BLUE
• e - Etch a Sketch
• g - Gameboy
• x - Xeon (periodic table) #lowercase BLUE
• 6 - 1962 LED invention date
• t - Tesla
• M - Mouse

For this second big run, I decided to make some assumptions on upper/lowercase based on tile colour... the first clue being the "i" characters. Due to the Base58 alphabet only having lowercase i, and the 2 "i" clues that I was fairly confident on (iron and igloo) both being blue tiles, I decided to try making all the blue tiles lowercase. The 2nd character is always an uppercase in a WIF address. It was light green, so I decided to set all the lightgreen tiles to uppercase. For all the other tile colours, I was including both the lower and uppercase for my best guess at what the character was which increased the keyspace quite considerably.

However, setting the blues and light greens to a single case helped cut the final keyspace I was searching down to "only" 8,589,934,592 key combinations, which at 25,000/s would take around 90ish hours (just under 4 days) to complete.

Analysis after finding the key shows that in addition to my (thankfully correct) assumptions on upper/lower case, that black tiles were all uppercase, while the red and darker green tiles were all lowercase.

In the end, the script only got through just over 1,103,000,000 keys (in just under 12 hours) before I got lucky:


I sent a note to the puzzle creator to let them know...  and I have created a PasteBin with my source code in case anyone wants to see my pretty ugly and untidy python code

 That is very disappointing. 
If we look at it from another perspective, it's actually good news because if we can't solve this puzzle even with the clues, imagine how safe our Bitcoins are from brute force cracking

well... that's a little disappointing...


Found 3 valid private keys out of 9,663,676,416 combinations... and none of them were the right one (or had any BTCs ) So, obviously, some of my assumptions were wrong 

We really need to solve the mystery of whether or not there is any way to determine upper or lowercase based on the information we have... I have not been able to discern any pattern as yet...

So after nearly 20 hours...


Almost 1.9 billion keys searched... so the average still seems to be around 1mil per 35-40 seconds... even with me doing other things on the machine and playing Overwatch etc... I guess that isn't toooo surprising given that it should be running single threaded... and I'm on a 4 core machine...

Maybe I could force multiple copies of the script to run on different cores... and have the keyspace divided amongst the copies... I could check them faster?

Ok... so I've done some rewriting... and am actually now just doing the WIF checksum checking... ie.

• 1. convert WIF to Byte Array
• 2. drop last 4 checksum bytes from 1.
• 3. SHA256 hash result from 2.
• 4. SHA256 hash result from 3.
• 5. First 4 bytes from 4. is calculated checksum
• 6. Compare with Last 4 bytes from 1.
• If checksums are the same, then we're good and have a 'valid' WIF private key (not necessarily the winning WIF), else ignore that WIF

I used this site as a way to generate some test WIFs and to get it straight in my head the correct steps needed to make sure the checksum was good. I am fairly confident (around 99.9%) that if my script generates a valid WIF private key, it will output it for me to check against the public address.


As for speed... I've set it up to output a datetime every 1,000,000 addresses... it seems to be doing this about every 40 seconds... which is around 25,000 keys per second... According to my math, with my current 'guesses' for the possible chars in each position, I have 9,663,676,416 keys to check. At the current rate, this run should have checked the entire keyspace that I am checking in around 4 days.

EDIT:

don't get me wrong. i'm totally for more but what is ion? an altcoin? then it is maybe the wrong place. and for a new one i would suggest to open a new thread anyway. 

Gonna cross link another game here:

https://www.iontalk.net/t/100-ion-competition-find-the-missing-privkey-characters/218/

Hope the OP doesn't mind

replied!


thanks, do you have any estimate of how many pk are you checking per second with that cpu?


i say this from my own experience but i may be wrong: throwing exception and catching/handling it will slow down your loop drastically


that would be the best idea.
i have zero knowledge of python but check this out:
https://github.com/jgarzik/python-bitcoinlib/blob/master/bitcoin/base58.py
or
https://github.com/petertodd/python-bitcoinlib/blob/master/bitcoin/base58.py

and whatever you do first check your code with some correct private keys to see if it works

Base58check is a lot faster, since it doesn't need to use ECDSA. That can improve the speed by over a 1000x haha.

I am using an old i5-3570k... 8 Gigs of RAM... Windows 10, Python 3.6... nothing really special... and it turns out I was waaaaaay off with my estimates... my code is still running after I got back home tonight from work... so it probably has another day or so to run

As for checking the privkey against the address... I'm using the privkey_to_addr() function from the pybitcointools python library... basically you pass in a privkey and it returns the address (or generates an exception for a 'bad' privkey)... I either catch the exception (and basically ignore it) or compare any returned address against "1qkCBiEjY4GAUFBcrsDXqyM6EPbZKTqCW" which is the published address.

I should probably do the Base58CheckSum thing, and looked into it briefly, but got confused by all the math and couldn't see any obvious "Base58Checksum()" type function, so I went with the first function I found that looked like it could tell me if the generated privkey matched the bitcoin address

Please check your PM.

how do you check it against public key? are you trying to generate the pubkey with different combinations using the code? i don't know how can that even work but i feel like you are taking the long route if my guess is right.

i used a simple Base58CheckSum on each combination which i think it is faster


i would like to know about your CPU power and also the time it takes to complete it for science you can see mine here:

Yeah... I saw that c# code and how you were building your char arrays... I've basically done the same thing, but with some slightly different assumptions for a couple of the clues... I've written it in Python (was fun learning the syntax and structure for python this afternoon) and it is currently grinding away testing my privkeys against the public address that was published.

My spreadsheet tells me that with my assumptions (uppercase for Surname and company name clues etc) I have 1,811,939,328 keys to test... will be interesting to see how long it takes to complete the run...

How to use you gist in C# I haven't got much experience with coding.

that would be my picture
and here is the gist and to be honest i worked on it for a while and moved on to new projects like the wallet in my signature and This OHLC Chart one.
i am planning on doing a fresh run working from scratch though

i would love to hear any idea.

and as for brute force i made my own code and checked a couple of 10s of million private key variation and even found one but never the answer. you can see the build that use in the gist to get an idea about what i am doing.