Quote
Take a look at Table 2 in Meni's paper: "The maximal safe transaction value, in BTC, as a function of the attacker's hashrate
q and the number of con firmations n."
Lets go with a 33% hashpower attacker, at 6 confirmations you can assume any transaction up to about 300 bitcoins is safe.
(disclaimer: I haven't taken the time to digest Meni's analysis, I'm going to assume his numbers are correct).
If you're worried about that, then don't make multi-thousand-dollar bitcoin transactions with people you think might try to double-spend and rip you off OR wait for more confirmations.
Also: don't forget that "33% hashpower" means you have half as many (asics/fpgas) as the rest of the network combined:
Before attack: lets say network has 100 Thash
You add 50 Thash, so during attack you have 50 of 150 Thash (== 33%)
q and the number of con firmations n."
Lets go with a 33% hashpower attacker, at 6 confirmations you can assume any transaction up to about 300 bitcoins is safe.
(disclaimer: I haven't taken the time to digest Meni's analysis, I'm going to assume his numbers are correct).
If you're worried about that, then don't make multi-thousand-dollar bitcoin transactions with people you think might try to double-spend and rip you off OR wait for more confirmations.
Also: don't forget that "33% hashpower" means you have half as many (asics/fpgas) as the rest of the network combined:
Before attack: lets say network has 100 Thash
You add 50 Thash, so during attack you have 50 of 150 Thash (== 33%)
I don't worry much right now about economically irrational, "I'm going to spend millions of dollars to disrupt the bitcoin network" attacks because I don't think anybody is going to spend millions of dollars to disrupt our tiny payment network.
I have no idea what bitcoin payments will look like in 5-10 years; I expect all sorts of trust mechanisms and relationships to develop that are independent of the bitcoin network, and I suspect some of those will make 51% attacks irrelevant.
And I have no idea what the mining landscape will look like in 5-10 years; if thousands of companies around the world are installing bitcoin mining hardware for free in every house built in cold climates (generate bitcoins and get a discount on your heating bill) then it may be completely inconceivable for even a government to amass enough hashing power to mount a 51% attack.
So while I encourage y'all to keep thinking about it as an interesting theoretical problem, it is only slightly higher on my personal priority list than worrying about quantum computers breaking ECDSA.