Why Wright's proof is a fake

03 May 2016

I explained in my previous post (in italian) that the signature that Wright provided as a public proof is in fact invalid. I want to explain briefly how you could check this claim. The key in Wright's post is this:

------------------------- Signature File -------------------------
MEUCIQDBKn1Uly8m0UyzETObUSL4wYdBfd4ejvtoQfVcNCIK4AIgZmMsXNQWHvo6KDd2Tu6euEl1
3VTC3ihl6XUlhcU+fM4=
------------------------- End Signature --------------------------

Now we can use some bash utilities:

  • base64, that translates encoded ASCII text;
  • hexdump, that displays hexadecimal contents from the input;
  • cut, used to remove the binary part of the input;
  • tr, used to delete spaces and carriage return from the input;
base64 -d <<<'MEUCIQDBKn1Uly8m0UyzETObUSL4wYdBfd4ejvtoQfVcNCIK4AIgZmMsXNQWHvo6KDd2Tu6euEl13VTC3ihl6XUlhcU+fM4=' | hexdump -C| cut -b 11-60| tr -d ' \n'

3045022100c12a7d54972f26d14cb311339b5122f8c187417dde1e8efb6841f55c34220ae0022066632c5cd4161efa3a2837764eee9eb84975dd54c2de2865e9752585c53e7cce

Let's analyze the command one by one:

  • base64 -d decodes the redirected string, the output is some gibberish characters so I won't display them here;
  • hexdump -C is used with a pipe to convert to hexadecimal:
00000000  30 45 02 21 00 c1 2a 7d  54 97 2f 26 d1 4c b3 11  |0E.!..*}T./&.L..|
00000010  33 9b 51 22 f8 c1 87 41  7d de 1e 8e fb 68 41 f5  |3.Q"...A}....hA.|
00000020  5c 34 22 0a e0 02 20 66  63 2c 5c d4 16 1e fa 3a  |\4"... fc,\....:|
00000030  28 37 76 4e ee 9e b8 49  75 dd 54 c2 de 28 65 e9  |(7vN...Iu.T..(e.|
00000040  75 25 85 c5 3e 7c ce                              |u%..>|.|
  • cut -b 11-60 displays only the characters from column 11 to 60:
30 45 02 21 00 c1 2a 7d  54 97 2f 26 d1 4c b3 11  
33 9b 51 22 f8 c1 87 41  7d de 1e 8e fb 68 41 f5  
5c 34 22 0a e0 02 20 66  63 2c 5c d4 16 1e fa 3a  
28 37 76 4e ee 9e b8 49  75 dd 54 c2 de 28 65 e9  
75 25 85 c5 3e 7c ce
  • tr -d ' \n' is used to delete spaces and carriage returns from the output so that is shown in one line and it gives us the final result:
3045022100c12a7d54972f26d14cb311339b5122f8c187417dde1e8efb6841f55c34220ae0022066632c5cd4161efa3a2837764eee9eb84975dd54c2de2865e9752585c53e7cce

If you noticed, there is also another cleartext string at the beginning of Wright's post:

$ base64 -d <<<'IFdyaWdodCwgaXQgaXMgbm90IHRoZSBzYW1lIGFzIGlmIEkgc2lnbiBDcmFpZyBXcmlnaHQsIFNhdG9zaGkuCgo='
Wright, it is not the same as if I sign Craig Wright, Satoshi.

Now let's head to blockchain.info. Blockchain.info has a little utility to get hexadecimal informations out of a transaction on the blockchain, so let's use it to get the related info about this transaction:

tx/828ef3b079f9c23829c56fe86e85b4a69d9e06e5b54ea597eef5fb3ffef509fe tx/828ef3b079f9... in hexadecimal

As you can see the entire output of the first bash command, that is

3045022100c12a7d54972f26d14cb311339b5122f8c187417dde1e8efb6841f55c34220ae0022066632c5cd4161efa3a2837764eee9eb84975dd54c2de2865e9752585c53e7cce

is contained in:

"script":"483045022100c12a7d54972f26d14cb311339b5122f8c187417dde1e8efb6841f55c34220ae0022066632c5cd4161efa3a2837764eee9eb84975dd54c2de2865e9752585c53e7cce01"

except for the 48 at the beginning and the 01 at the end.

That is a signature hash: this page explains that the 48 is just a decimal prefix given to uncompressed transactions, and the 01 at the end is just a SIGHASH_ALL code that flags the end of the signature.

So, is it a fake?

Yes, indeed. At the end, I ask, why would you choose anything else than the easiest and most conclusive way to prove something?

Wright "signs" the blockchain

Source