I'm not sure that is possible. If a 51% attack for 1-week costs $1-million USD to pull off (hypothetical round numbers against a hypothetical coin)... then how many confirms do you wait for?
If you wait for 1008 confirms (for a transaction of $1 Million USD), then that doesn't stop the attack at all. After all, the attacker simply has to perform $2 Million USD worth of double-spends during that timeframe to make a profit. (Not necessarily with you, but across the entire blockchain).
After waiting 1008 confirms (1-week for a 10 minute/block coin), the attacker simply replaces those 1008 blocks with his 1049 secret-blocks and makes his chain the longest. That means that ALL spends for the past week are nullified, allowing the attacker to double-spend his coins on the new blockchain.
----------
In effect, waiting for 1000-confirms (or whatever) is closing the barn door after the horse has bolted. It doesn't protect your transaction... it protects the NEXT GUY's transaction.
In any case, the attacker gets to double-spend his money, which probably includes a large exchange (where you can turn coins into US Dollars easily).
If the attacker has 50+% of mining power for the entire duration then yes it wouldn’t work as the longer things stretch out the further ahead the attacker gets (statistically).
However it the attacker has less than 50% they can still wage an attack. With 10% of mining power I’ve got a 1% chance of mining consecutive blocks. It dwindles quickly even for larger percentages and a large number of confirms makes that angle impractical.
> However it the attacker has less than 50% they can still wage an attack. With 10% of mining power I’ve got a 1% chance of mining consecutive blocks. It dwindles quickly even for larger percentages and a large number of confirms makes that angle impractical.
ETC's blockchain just had over 100+ blocks change, according to the tweet.
Even with 40% of the mining power, the probability of such an event (40% mining power, 100 consecutive blocks) is 0.000000000000000000000000000000000000016%. That's a very unlikely hypothesis you're presenting.
