Quote from: DeathAndTaxes on July 18, 2013, 09:20:28 PM

Quote from: poiuytr4 on July 18, 2013, 09:14:02 PM

Quote from: DeathAndTaxes on July 18, 2013, 02:21:46 AM

Given:
private keys a & b
Public keys A & B
Data to be signed d

Is it possible to create a signature S such that it can be verified given only A, B, and d?

Why not sign the data d with private key a and then sign the result with private key b to give you S.
Use public key B then public key A on S to result in data d.

Would this solve the original problem?
Even though, as pointed out, there are distinct items of data so it wouldn't work in practice anyway.

This would be for a new (incompatible) transaction format. There would be no distinct items. Transactions would simply ONLY be signed at the tx level.

I don't believe it is possible to verify a double signature the way you described. Remember is verification the entity with the public key isn't recreating the signature and comparing it to the original (if they could do that they could counterfeit the signature on any data). They entity doing the verification can only validate if the signature is valid or not (i.e. true or false).

I'm pretty sure that this would work to do what you originally described. Basically a private key is used with a random number to encrypt (sign) some data and the public key is used to decrypt (check signature). All I was suggesting was sign (encrypt) the data with private key A then sign (encrypt) the result with private key B.
To check just decrypt with public key B then decrypt with public key A. They are really just one way functions.
I'm not sure that you can do what you described above. i.e. you can't just add a lot of private keys, sign a message and then add up the corresponding public keys and use this to check the signature (if that is what you meant) even if you are using EC point addition. I'll have to have a look when I'm more awake (sober)