Re: [bitcoin-dev] Transition to post-quantum

[Tim Ruffing <tim.ruffing@mmci.uni-saarland.de>]

On Thu, 2018-02-15 at 21:27 +0100, Natanael wrote:
> I addressed this partially before, and this is unfortunately
> incomplete.
> 
> Situation A: Regardless of expiration of commitments, we allow
> doubles. (Or no doubles allowed, but commitments expire.) 
> 
> If I can block your transaction from confirming (censorship), then I
> can make my own commitment + transaction. The miners will see two
> commitments referencing the same UTXO - but can see only one
> transaction which match a valid challenge and spends them, which is
> mine. You gained nothing from the commitment.

Yes, I assume situation A: 
  * commitments never expire
  * and there is no limit on the number of commitment for the same UTXO

As I understand, you mean "decommitment" when you say "transaction".
Please correct me if I'm wrong. I'll stick with "decommitment".

Let's assume the attacker blocks the decommitment by the honest user,
inserts his own malicious commitment and his own decommitment, which
should be valid for the malicious commitment. Then the miners will see
two commitments (the earlier commitment by the honest user and the
later one by the attacker).

Also, the miners will indeed see one valid decommitment. This
decommitment may have been sent by the attacker but it's the preimage
chal of the address, because otherwise it's not valid for the malicious
commitment. But if the decommitment is chal, then this decommitment is
also valid for the commitment of the honest user, which is earliest
additionally. So the honest commitment wins. The attacker does not
succeed and everything is fine.

The reason why this works:
There is only one unique decommitment for the UTXO (assuming H_addr is
collision-resistant). The decommitment does not depend on the
commitment. The attacker cannot send a different decommitment, just
because there is none.

Maybe I'm wrong and I just don't understand your attack. In this case,
please explain it more detail.

Regards,
Tim