Bitcoin Forum

I have a question. Recently there was a flood and a notebook containing a offline wallet was damage and it destroyed part of a WIF private key, so now I basically have:

Kw**********(I have the next 40 characters, just not posting for obvious reasons), so I am missing 10 characters in all.

I also have the public key. Is it possible to use this software to start a search at Kw... and iterate over the missing 10 characters with the known 40 characters also in the key.

For example : 1B2Q8vPm5E5b8yxaNWUW5ZfdCR5Zu1KMJn - is the public address

and for the WIF private key I'd have Kw**********jcQmKRPNTF8CU9H1thzC981DCrZgmS4m8ygXFCEk (this is just an example, no funds are here)

basically I want to always have search for private key start at 'Kw', then search for missing 10, and end with 'jcQmKRPNTF8CU9H1thzC981DCrZgmS4m8ygXFCEk'

Any advice is appreciated and examples are even more appreciated. I do have a programming background, but I don't have much experience programming to use GPUs, which I imagine I am going to need, so I am open to any and all suggestions, other than just giving the key over to someone. I will offer a reward to anyone who successfully helps me with this too.

Or is 10 characters just too many possible combos. All advice is much appreciated!

Thank you,
S.

What do you mean ^{"it destroyed part of a WIF private key"}?
Is the private key you talking about is on a piece of paper? I think I misunderstood the notebook is it the laptop?

That's too many missing characters actually some tools out there can only recover if the private key has missing 1 up to 5 characters.

However, would you mind to try the tool developed by Coding Enthusiast (https://bitcointalk.org/index.php?action=profile;u=879277)
Let's hope his tool will work to find the missing character you can find the tool from this thread below

- https://bitcointalk.org/index.php?topic=5214021.0

Or go directly on Github https://github.com/Coding-Enthusiast/FinderOuter

It was a paper wallet, so yes the water destroyed the ink on the paper. I have already done a CPU based search and it was going to take like 2.5 centuries or something to solve it, so I am wanting to attempt with GPUs to see if I can solve this any faster. I know how FinderOuter works and with an i7 it can do around 74 million combos a second, but I am going to need something in the billions to solve this soon, and I know GPUs can definitely reach that height.

Thank you though for your answer :)

Thanks,
S.

Chemical analysis of the paper  :)

There is nothing left. It mushed up the paper and ink. basically it has a big chunk missing out of the top left of it. :(

do you have the seed phrase or only saved the PK?

Is the paper missing after Kw or is there enough paper for some letters, so that the chemical analysis would help you to recover some more characters to make the calculation easier?

For example:
Kw**********jcQ...

** -> paper / chemical analysis / recover characters
****** -> hole / no paper

-> so you would have 6 missing characters

Its just a PK from many many years ago.

the paper turned to mush. There is no hope of any kind of analysis.

Do you have the compressed/uncompressed public key or outgoing transactions from that address?

I have the public address which will have transactions on it yes. Does that help any?
I basically only have the damage WIF key and the public address like in the example in original post.

See my answer on Jean Luc's Kangaroo thread:


Note: it's important to place your public address at the end of the command (after the stride), although I did not write that in the other post.

If there is an outgoing transaction, then with the un/compressed public key and kangaroo or pollard, also possible.

'

I thought WIF keys that began with K or L were compressed keys and WIF keys that began with 5 were the uncompressed ones? Also I am a bit confused how I get those big numbers with only the public address and a WIF key that is missing characters. Could you elaborate anymore? I apologize I am fairly new to learning about all this.

Sadly this was an offline wallet, so it only ever had incoming transaction. It has never sent out.

So that would be very easy. ~60 missing bits, we would have the private key within minutes.

Really? I thought with 10 missing characters that was 58^10th or a large number and it was going to take centuries to solve this. How exactly do I solve this in minutes? What tool should I use, or is there something custom? I am willing to pay a bounty for someone assisting me to set this up on machine.

We don't have an outgoing transaction, so we can't do it with pollard or kangaroo.

You said above that we could do it in minutes. Is that not possible now since we don't have any outgoing TXs?

Thanks,
S.

Yeah you're right - my memory was a bit rusty. In any case just replace the -u flag with -c.

I simply converted the base58 of the lower characters to decimal (and hex). First I went to this page: https://www.dcode.fr/base-58-cipher

And then I pasted the characters after the lost 10 chars inside the page. Before the characters, I pated the 'w' (since you know you have that), followed the 10 characters lowest possible private keys that still base-58 encode into w........JzXaqU2rcFSoaLaehAQHqoQX1cWCo92tAA3ihLJ7 - replace the dots with 10 "1" (the number one) characters. Because 1 is the first digit of base8 number system. The resulting hex gives the starting range.

Then to get the end range, you repeat the process but instead of ten 1 characters, you insert 10 'z' characters (the highest character in base58 is lowercase 'z').

To get the stride, I simply converted the lower part of the base58 you had (the one after the dots).

To determine the start and end ranges and the stride, you only need part of the WIF, not the public address.

These steps will create a range and strie that is suitable to input inside Bitcrack.

No problem, because Bitcrack is more efficient than Kangaroo for your problem (also, Kangaroo will only work if you have the public key, not the address).



58**10 is a very large number, however to estimate the difficulty, you need the equivalent power in base 2, so what we do is we take the log2 of the result: log2(58**10). Then if gives us the difficulty in bits: such that 2**bits == 58**10 (here, bits equals 58.a_fractional_part).

Difficulty can also be written as "10 base 8 characters", but its common for programs to estimate it in terms of bits as well.

This will probably not take minutes unless you have a large GPU farm, but a few weeks is a more accurate estimate. Since Bitcrack can only talk to 1 GPU as far as I know.

I thought, you have the public key too.


Without an outgoing transaction, we don't have the public key, it will take longer.

I wish you good luck.

so I am plugging the value into that site: w1111111111JzXaqU2rcFSoaLaehAQHqoQX1cWCo92tAA3ihLJ7 and for hex I am getting:

64 B3 82 BA 7E 44 5B 0F 02 B1 26 EE 95 04 62 B9 E3 A8 8D 67 7C 5D E1 74 E6 44 88 6D 5B 20 F8 8F F0 10 E5 FF A5 C0. I feel like I am doing something wrong on the site.

Any insight? I am not getting the hex values you got in your original post.

Pollard's Kangaroo method also uses a start and end range. However, it does not support strides (AFAIK) because Pollard's Kangaroo algo is using it's own kind of stride while making the tame and wild kangaroos hop. There's no way to tell Kangaroo: "OK, there is a part of the private key at the end that's already known, don't search those bits". The consequence is that for this particular situation, Kangaroo will take longer to find the PK (way too long actually).

Bitcrack, on the other hand, lets us specify an arbitrarily large stride and it converts it to a 256-bit int. Since in this case, 80% of the private key is already known, we can make the stride equal to that huge PK chunk at the end and effectively, only search keys between Kw11111111...... and Kwzzzzzzzzzz.......


My hex for w1111111111 came from the following input: w1111111111JzXaqU2rcFSoaLaehAQHqoQX1cWCo92tAA3ihLJ7oQX1cWCo92tAA3ihLJ7 (I accidentally duplicated the last part).

Ok, so I get how to get the range for the start and finish in hex to then use in the --keyspace. What I don't know is how did you get the decimal number that starts with 244 that goes after --stride?

Also do I plug the public address in anywhere?

Thank you,
S.

Keep in mind that this method is significantly slower than to simply test each character permutation.
You see each character at the start of the string (from left) converts to a much bigger integer than any character from the end of the string. So even the difference between 1 char missing becomes huge.
Take the following example:
Ky**DfuvLpt8eSb8EQzhZwDCQeCaycKeAoxJMY8pfPZXmn3uB38R
Even though only 2 characters are missing the difference between Ky11Df... and KyzzDf... as integer is
While the permutations are only 3364.

Do you have a faster way I can do this with GPUs? I have used your tool, but with just CPU its going to take far too long for 10 missing characters.

Unfortunately I haven't been able to add GPU support to FinderOuter yet.

The program I'm using is able to skip-count so I think that can mitigate the speed penalty. So I can jump through each next value, adding the huge difference to the previous each time to get the next.

1 year ago, user PawGo calculated exactly your case (10 missing WIF characters) but with the public key in less than 11 minutes.


They started with less missing characters but then ...


Thread: Using Kangaroo for WIF solving http://bitcointalk.org/index.php?topic=5315607.0

---

Edit: PawGo posted

Normal Kangaroo yes, but I have already patched it for solving WIFs, custom stride is supported. Link in post above

I will check it out. I have two 1070s in a rig I can dedicate this too, so hopefully it can do some fast solving. I will message back if I need help. Thank you!

Also I should note I only have the WIF key and public address. I don't have the private address or public key. THe WIF is missing 10 character exactly due to damage of paper wallet. Does this make any difference?

What kind of hardware did he have to solve 10 characters in 11 minutes! Thats awesome and is exactly what I need. going to look into this as well now!
**EDIT** I just saw he did it with just a CPU. Thats crazy! But I don't have public key, so thats not gonna help me much right? And since this address has never sent anything I can't get it right?

You can't do it in that short time, as you don't have the public key.

Ok that makes sense. Seems like bitcrack is my only hope at this point. Thank you for your continued help. Should I solve this I would love to pay everyone a bounty who has helped :)

Thanks,
S.

SO I may have gotten lucky and found an old screenshot of wallet that contains a Hash160 address. That is the public key right? Please tell me I just got lucky and can use the kangaroo stuff above?

Thanks,
S.

No, with that Hash160 address, you calculate your Bitcoin address.

examples of public keys:

public key (uncompressed)
04b4632d08485ff1df2db55b9dafd23347d1c47a457072a1e87be26896549a87378ec38ff91d43e 8c2092ebda601780485263da089465619e0358a5c1be7ac91f4

public key (compressed)
02b4632d08485ff1df2db55b9dafd23347d1c47a457072a1e87be26896549a8737

or
03b4632d08485ff1df2db55b9dafd23347d1c47a457072a1e87be26896549a8737

they all start with 02, 03 or 04
length is 66 or 130 characters (0,1,2,3,4,5,6,7,8,9,a,b,c,d,e,f) letters can be upper case (A,B,C,D,E,F)

Blah thats right. Man really wish I'd have done at least 1 tx on this thing.

I know it's a long shot, but still: Any chance you ever signed a message to prove ownership of that address?

Thats a good idea, but ssadly this was a offline wallet from the start and never used, only ever deposited into.

Thanks tho :)
S.

@soferox

I have made test too 4 digit recovery take 75 to 80sec on single core CPU. 10 digit is way out of range for CPU.
but it's possible to code something for GPU just for WIF solving.
but it's depend on how much your address is holding and how much reward we are talking.

if it's worth it, hire coder who can make cuda kernel for bf WIF and its take few minutes to recover that.
but cuda coding is not cheap.

Update:

The OP has since identified two of the unknown WIF characters at the end. We managed to get Bitcrack running on his two GTX 1070's searching for the remaining 8 characters at a combined speed of 400Mkeys/s, over Discord.  The estimated runtime is 16 days. Hopefully, it turns up with his lost WIF.

Indeed. He contacted me and we have some estimations. With 10 missing characters it is really doubtful, but as he wants to risk with 2 extra characters, it is more realistic. My estimation for his cards was 400MKeys but with settings he sent me it will be a bit less. Estimation was 11 days, BUT unfortunately he said he is not sure about the beginning, so maybe it will be around 40 days total (11 days for Kz, Ky, Kx and around 7 days for Kw).
We will see…

Very good! But not that way? Do you know, how?

I'm curious: how did they do this?
Was it something like mynonce's scenario?

It was through taking a hard look at the piece of paper, not any chemical analysis  :)

These numbers don't look correct.
With 8 missing characters it should take less than 4 days (1/4 the reported runtime) to check all permutations.

I had different calculations... Anyway, the problem is that OP is unsure about one of known characters, so work must be multiplied *4.

Check your pm for a python script

That's because he wasn't sure that the "w" at the beginning of "Kw" was correct so he wanted to check Kw, Kx, Ky and Kz just to be sure.

No python script...

As to everyone else who has been helping thank you so much. I will 100% give a bounty out to everyone that has added some help to this effort. I am really hoping the 2 newly identified characters are correct. Based on my estimation with a missing 8 characters and unsure if the second if its w, x, y, z it will take 3.9 days per run (w, x, y, z are the runs) so 3.9 x 4 is roughly ~16 days. I hope to come back with really good news and if not well lesson learned to get water proof paper lol...

Thank you all again,
S.

its more than 16 days now... I hope you could find something bro.

Hey Bro, I read through everything. I Know John Cantrell Wrote a program in Rust for seed phrases. We could tweak that and make it work for PKEYS. Let me try tweaking it and if it works Ill update you all and go from there. I have 8 Tesla V100s that may also help if need be. If anyone here is a Rust language Master please PM me as I am only Learning it.

Do you mean those software?
https://github.com/johncantrell97/bip39-solver-gpu (https://github.com/johncantrell97/bip39-solver-gpu)
https://github.com/johncantrell97/bip39-solver-server (https://github.com/johncantrell97/bip39-solver-server)


Since BitCrack already utlize GPU, do you think it's possible to achieve higher hashrate (with same device) by modifying his software?

For some cases my program is a least 2x faster than BitCrack.
https://github.com/PawelGorny/WifSolverCuda | https://bitcointalk.org/index.php?topic=5382190.0

About the subject - OP is working on the problem, but apparently the decoded characters were not correct, so there is much more work than expected.

Since BitCrack already utlize GPU, do you think it's possible to achieve higher hashrate (with same device) by modifying his software?
[/quote]

I guess so because His program can utilise Multiple GPUs where as bitcrack can only use 1 at a time. And for the guy posing the WIFsolver CUDA does it work with multiple GPUs. I have 8 Tesla V100s on a 2x Xeon 256 GB ram server. Let me know if we wanna try out something before this Sunday. As i am flying out to a remote place for work and won't be available for 2 weeks unless there is network here and there sometimes.

You may always split the work. The same with BitCrack. There is even option to use the specified part of given range - so you may launch 8 instances with the same configuration and then specify for each device to process 1/8, 2/8, 7/8 of the range... I think I will add something like that to my program, good idea.

how this thing ended ? OP found it ?

Here they're searching it:

https://bitcointalk.org/index.php?topic=5404406.0 (https://bitcointalk.org/index.php?topic=5404406.0)

It has nothing to do with original question (at least not based on what I know). OP claims he is still searching for his WIF, as he wrongly decoded some characters, and work continues. Nothing new is known.
It is true WIF500 is a very similar problem (similar combination of known characters and missing ones).

Find the WIF challenge https://github.com/phrutis/wif500 (https://github.com/phrutis/wif500)
Some others think it too  ???

No, you were the first who said that two problems are connected, while they are not:

@NotATether is helping @soferox
And then he answers, after I had said, that they are searching it there https://bitcointalk.org/index.php?topic=5404406.0 (https://bitcointalk.org/index.php?topic=5404406.0), where is a link to the challenge https://github.com/phrutis/wif500 (https://github.com/phrutis/wif500) and there you can see the damaged paper:
I thought that he would know it.


Are you sure that they are not?

@soferox
Can you give us an answer?

The damaged paper in the picture is burned not water damaged.

We never got anywhere near a solution, and besides, OP deleted his telegram account 3 months ago (or maybe it was autodeleted by TG) - That is where we were working from, hardly at all on Discord. Note that we were trying to recover an ETH private key from a screenshot which had some purple marker scribbles from Paint on it.

It just made no sense on the Cracked site that if the key was recovered from the characters he put up that this would end up being a key that would not work that "they" would have to decode it,impossible to me,key found would be THE KEY,if I'm wrong about anything please let me know,I wouldnt waste my time on the eth key,I like to play with my program for partial keys but it's near impossible to actually find a real one.

Or if you can confirm the @soferox partial key is this posted key that would be great,you worked with him and would know the answer to that question,I'd like to try it and would let my machine run 2 years but not without some confirmation from you or @sorerox.

Are you still trying?, cause I have my own programms for that, I can join to the search ASAP!  :D

Don't waste anytime on this,I found this during Covid lockdown when I was looking into my old wallets and keys I had and needed some help,I finished with them and thought this would be a cool thing to try,it's a scam,the OP put in a clearly fake picture,all burned instead of water damaged LOL,the Address is real the key is fake for that address,no one is so stupid to post the the real key,sure we can't break it now but in 10 years who knows,so the way he had it set up was you would join his group and be given a "range" and when you got 90% through the range you would contact him and then he would finish it up,this guy was on crack or something thinking people would actually do it,the range he sent you with HIS program is probably real but it was not searching for the key in the picture,only the real key is known to him and a few others maybe,garbage don't waste your time,if the OP is out there and reads this come back on here and talk,prove me wrong please others think your an idiot too,he won't.

You mean the 500 WIF? of the user Phrutis? or (something like that), this is not another Key, I remember that 500wif is 12 chars missing.

Yes that's the one I meant the 500 WIF by Phrutis,which one are you talking about ?

Looks to me there the same and the @soferox one is legit,the other one by phrutis is a scam,my opinion only though.