Some people have expressed the opinion that a successful double-spend attack would invalidate Bitcoin as a technology, and thus undermine the value of all Bitcoins (including the attacker’s), making an attack an unwinnable situation.
That doesn’t seem reasonable to me. We have seen doublespends on other networks and those networks continuing to operate later. It would surely impact the value of Bitcoin, but it seems absurd that the new value would immediately be zero. Most people wouldn’t even catch on to the events for hours or days.
My thought is that if my node detected a large re-org and double spend within that re-org, I would not accept payment associated with the double spender. If his identity was publicly known, nobody would accept his Bitcoins because they know they are liable to be double spent. If his identity was not publicly known, his Bitcoins are still tied to the attack, and nobody ought to accept them for the same reason.
Depending on how the attacker selected transactions in their blocks, there might be something on the order of 24 000 transactions in six blocks that become unconfirmed if the attacker simply does not include any transactions in the replacement blocks. While it may be possible to find a single or few transactions with high value that resolve differently, it may be fuzzy and non-obvious for a number more. I am not convinced this is a practical approach.
But on the other side of the coin… Now, I know for a fact that all other Bitcoins not associated with the attacker cannot be double-spent, because by definition only one entity can have more than 50% of the hash power at a given time.
Assuming that the attacker only includes their own transactions, the senders of ~24 000 transactions would be in the position of being able to reissue their own transactions. It’s not clear at all to me why someone attacking the blockchain provides more confidence to the actions of other users.
Would this create a situation where the attacker has effectively burnt his own coins while simultaneously increasing the trustworthiness of all other coins?
No, not at all.
Or said another way: is it valid that we do not have to assume a 51% attack would undermine the entire network’s value to conclude that it would still be self-undermining for the attacker?
No, by all likelihood, the attacker would immediately trade out of Bitcoin if they were worried about the attack severely impacting Bitcoin’s value. They might not be able to trade some of the coins that were involved in the attack, but they could short Bitcoin in a future trade or at least trade all their other Bitcoin holdings. The attacker could then even trade back in on the dip caused by their attack’s damage to Bitcoin’s value and earn more value on the recovery. I don’t think this argument stands up to scrutiny.