Slashdot Mirror
Bitcoin Protocol Vulnerability Could Lead To a Collapse
First time accepted submitter stanga writes "Cornell researchers unveiled an attack on the Bitcoin mining protocol
that enables selfish mining pools to earn more than their fair
share. In a technical
report the authors explain this
attack can be performed by a pool of any size. Rational miners
will join this pool to increase their benefits, creating a snowball
effect that may end up with a pool commanding a majority of the
system's mining power. Such a pool would be able to single-handedly
control the blockchain, violating the decentralized nature of the increasingly
successful Bitcoin.
The authors propose a patch to the protocol that would protect the
system from selfish mining pools smaller than 25% of the system. They
also show that Bitcoin can never be safe from selfish mining pools larger
than 33% of the network, whereas it was previously believed that only
groups larger than 50% of the network were a threat to the system.
The question is — can the miners operating today adopt the suggested fix and
dismantle too-large pools before a selfish mining pool arises?"
Bitcoins are the wild west...and that's why they're so exciting.
I missed the gold rush, but there's still money to be made selling shovels and pans to those who think they didn't...
Did the "selfish mining pools" us a Greedy algorithm?
much of left-wing thought is a kind of playing with fire by people who don't even know that fire is hot - George Orwell
Start with an intense desire to building your own private empire that you control.
Hiding information from others to gain a competitive advantage.
Populating other groups with spys to see what progress they are making.
Eventually giving rational people no choice but to join your team or be crushed.
I propose to call this the middle manager attack.
This attack would be very, very difficult to achieve. Doesn't seem very worrying and I'm sure it'll be fixed well before it becomes an issue. There are already some pretty good discussions on /r/Bitcoin/ covering why it's not as big a deal as the sensational headline here makes it out to be.
Error 404 - Sig Not Found
So that's what the NSA datacenter is for...
Custom electronics and digital signage for your business: www.evcircuits.com
I fairly understand that for there to be value in bitcoin there must be scarcity and that this scarcity is created via the mining mechanisms. But what I wonder is if there be any other way to create value for a virtual currency?
I ask because to me the most interesting thing about virtual currencies and specifically bitcoin is NOT the mining aspect, but rather the distributed database. The fact the hosting or provision of the database is fundamentally bound to the value-creation process seems to be the problem here. The problem seems not to necessarily be virtual currency or distributed databases themselves. The problem seems to be that value creation is based on artificial scarcity which can be manipulated through collusion.
There has to be another way to establish value for a virtual currency.
Someone trying to buy some bitcoins for cheap?
Here is the commentary from one of the Bitcoin core developers: https://bitcointalk.org/index.php?topic=324413.msg3476697#msg3476697
This is an old known attack which is boring, made a little more interesting by also assuming that the attacker has sybil attacked the network and inserted itself between every node. The result is that they can mine a disproportionally large share of coins. Academically interesting, but not terribly significant.
Mostly it's just another example that overly large pools are bad for the network, and that preventing sybil attacks (e.g. by miners setting up additional trusted peerings between each other) is useful.
Gold, salt, silver, greenbacks, plastic, bitcoin. Take your pick, None of it cures society of thieves, bank robbers, or scoundrels. And anyone who guarantees your money is secure is probably complicit in its theft. There will always be ways to steal your coin. Bitcoin just limits who might steal it.
because it's an underground currency that doesn't have any reliable defining body
What do you mean by "defining body"? I googled it, but all I got was cosmetics. Your shampoo has "defining body".
AKA Gold.
If someone has found a way to hack gold, they have had the good sense to keep quiet about it.
Maybe you don't? Proof of work is something you do that requires work and there must be an easy way to check that the work is done. Proof of work is suppose to be consistent. So if you want your work to be find a prime number larger than 1 trillion, after the number is found; then checking it is fast and easy. But finding it may have taken a long time. Finding it will take a long time if an identical machine tries the same work. So that is proof of work, two machines can confirm that finding that prime number takes work.
Bitcoin is, for lack of better terms, pseudo-proof of work. The work is to guess a random number + some other bits of info and make a hash. Then keep trying random numbers until you find a hash with enough zeros in the front to meet the target. Two identical computers guessing numbers will end up with a different proof of work. One computer might guess the answer before the other. So how to do you gauge which machine really did the work? Well, the machine that won claims to be the winner and has a way for the other machine to check quickly. If the other machine had continued working, it might find a different answer that is also correct, but took longer. Why is its proof work any less valid than the machine that by luck found an answer first?
So again, bitcoin is not proof of work in the true sense. It is proof of luck. The paper basically shows that proof of luck is really no good when you get people involved because it is just like the lottery. You can play the billion dollar powerball all by yourself and never win. But what if you could gather everyone in the country together into one large lotto pool, the winner would share the winnings with everyone. So even if everyone only got $1 from the lotto, you still got something right? No one would play the lottery if the "mega-pool" of people are always going to win. Bitcoin by contrast suffers from the exact same human produced issue. Case closed.
Maybe you don't? Proof of work is something you do that requires work and there must be an easy way to check that the work is done.
What sha1 hash salted with ABCDEF ends in the hex: 01234? 01235? 01236? Please show your work, and explain why your average workload to solve this type of problem will not be consistent?
Bitcoin isn't intended to be a money-making investment tool.
The headline is just plain FUD. The ideas presented in that paper are merely theoretical. Not only would it be extremely difficult to achieve the right conditions to execute the attack (at the expense of losing money when you fail), but the paper makes vast assumptions about the social response to it working. Basically, the conclusion was "if this works [which it probably won't], then everyone will collectively make decisions that destroy the network because that's the rational thing to do." Obviously, it's not so rational if people don't want to see the system collapse.
This doesn't mean it should be ignored. It's an interesting "attack" that should be kept in mind as the protocol is developed further, but it's not even close to "bitcoin collapse". The headline is perhaps just wishful thinking of the submitter.
Comment removed based on user account deletion
There is already some timestamping in the Bitcoin protocol, and this kind of attack was at least considered once upon a time by Satoshi before he went and disappeared.
The time stamp process within the Bitcoin protocol is more or less an average of what most of the clients say is the current time, and that protocol could be further refined in terms of eliminating outliers (one client or a small group of clients could in theory be rejected). At the very least you can program your own packet sniffer to flag curious blocks that may indicate some sort of attack like this is going on, even if in the protocol itself doesn't directly reject these kind of blocks.
The needs of the Bitcoin protocol do not need a hyper accurate time stamp protocol like NTP (and especially not an atomic clock), but rather "good enough" (a time stamp +/- 15 minutes or so) is sufficient. I do think a minor tweak to the Bitcoin protocol could likely implement a modest protection against all but the worst offenders of this kind of attack. It does not need to be a 3rd party time stamping service but could be implemented within the existing communication protocol and remain decentralized.
Except that the time stamp is built into the hash that generated the block in the first place. In other words, if you say the block was mined just one second later (which BTW sometimes does happen.... some blocks are generated very quickly after the previous block), the time stamp much also include the hash "proof" where the time code is being used as a part of the verification that the "winning" hash has in fact been achieved.
Unfortunately the current Bitcoin protocol doesn't really care when the blocks were generated, and many clients don't even bother with properly time stamping the blocks. I guess that could change though.
Oh, the bitcoin community is perfectly OK with misconceptions (bitcoin is anonymous, you need to control 50% of the mining pool to cheat) as long as they prop up the price.
xkcd is not in the sudoers file. This incident will be reported.
It's effectively a poisson distribution, in which case the standard deviation is significant in comparison to the average.
Of course, the law of large numbers will converge a large enough number of individual poissons (or anythings) onto a normal distribution, but for bitcoin the proof of work is not for "a large enough number of" results, but for a single one. That single tweak - asking for 16 results that are 16 times easier, say - would increase the fairness, and decrease the luck aspect, significantly. Likewise it would make the generation rate far more predictable. Alas, I suspect it would have some other unwanted side-effects too
Also FatPhil on SoylentNews, id 863