5 ms·

"That's quite a difference, and one of the main features of Bitcoin. See the message in the genesis block." That is a political statement not a technical state

"That's quite a difference, and one of the main features of Bitcoin. See the message in the genesis block." That is a political statement not a technical statement, and this gets to the heart of the problem with Bitcoin and with its lack of a security definition. Bitcoin is popular, particularly among those who distrust the banking system, because there is no requirement for a central authority; that is certainly true but only in a pedantic sense. Without a security definition it is hard to even talk about what Bitcoin requires. What we have therefore is a situation where no amount of technical criticism can matter with Bitcoin. Polynomial time attacks are irrelevant, because there is no requirement that Bitcoin resist polynomial time attacks. There is similarly no particular scalability requirement, and so there can be no real criticism of Bitcoin's scalability. "True believers" in Bitcoin can easily shoot down criticism because critics cannot actually point to any requirement that Bitcoin fails to fulfill: there are no clear requirements to point to. To illustrate this point, consider this statement: There is a polynomial time attack on Bitcoin, and the whitepaper itself describes it. Now, is this a problem? Well, the answer I consistently get from Bitcoin devotees is no, that is not a problem because that is how Bitcoin works. With logic like that, who can argue? So while you call the lack of an authority a feature, I call it a logical gap until a rigorous security definition is presented. Otherwise you have a solution in search of a problem, coupled with a community of people who all have their own vague notions of what they want Bitcoin to do for them.

jafaku 13y ago
> That is a political statement not a technical statement Just because it doesn't have the particular kind of specification that you are looking for, it doesn't mean that it's not technical. Being distributed is very much a technical feature, no matter how you look at it. I much rather have a working solution that might or might not be broken in the future (if it was too damn insecure, someone would have broken it already), than be stuck forever with the old banking system just because we don't know how to create the mathematical/cryptological model that you are used to and would grant 100% theoretical security. That would be like not building a website or shutting down a project mid development just because some component is found to be difficult to unit-test.
- betterunix 13y ago
  "I much rather have a working solution that might or might not be broken in the future" What does it mean for Bitcoin to be "broken?" That is my entire point here: if we do not have a clearly stated definition of what security means, then we cannot even talk about whether or not a system is broken. To put it another way, I would call any cryptosystem that can be attacked in polynomial time "broken," yet in the case of Bitcoin there is the 51% attack which apparently does not bother you -- and like I said, despite the fact that I call it broken, I cannot actually point to anything that would have required Bitcoin to be secure against such an attack. To put things in scientific terms, we are talking about falsifiable hypotheses. As an example, here is a commonly assumed hypothesis: the block cipher AES is a pseudorandom permutation. This is a claim that could be disproved i.e. we can falsify the claim by presenting an algorithm that efficiently distinguishes AES input/output pairs (with a secret key that the algorithm does not receive as input) from the input/output pairs of a random permutation. Consequently we can speak about the security of AES in a meaningful way, even though have no theoretical proof that it meets its security definition (only heuristic evidence and a lack of known attacks). My point about Bitcoin is that we do not have such a hypothesis, so we cannot even be sure that we mean the same thing when we say, "Bitcoin is secure" or "Bitcoin is broken." That is the point of having a security definition. Without such definitions, you can always make the claim that Bitcoin is secure, not matter what sort of attacks are carried out, because you can always just say that Bitcoin is not supposed to defend against those attacks.
  - jafaku 13y ago
    > What does it mean for Bitcoin to be "broken?" It's pretty intuitive and I'm sure you know it: People being able to break the rules. Eg: create coins out of nowhere, move coins that weren't theirs, slow down transactions, unstabilize the market, etc. To say that it's not falsifiable is a bit of a stretch. It is falsifiable, we just don't know how to create the model you are looking for. Maybe it can't be done [with crypto]. Maybe it wasn't a competence of Cryptography to begin with, and we need an entirely new field (I know cryptographers have tried in the past, but to me this seems different). Don't you agree that Bitcoin involves too many disciplines to try to simplify it into a cryptographic formula? Something that breaks Bitcoin could even simply come from Economics, and your crypto model (if you ever find one) would be useless. Or, there is a cryptographic model, has been proven secure, and we just don't know it because its creator[s] wanted it to stay secret. Let's try with an analogy (you can attack it or say why it's bad). Your ship is sinking for some reason, there are no emergency boats, but you find out something that might be used to stay afloat. You hesitate, because you don't have physical proof of it like you did with the ship (an expert created the drawings, calculated the forces involved, etc.). But then you see a lot of people using it and it's working fine. The more time it passes, the more certain you are that it works. That's empirical evidence, and it drives a big part of scientific advancements. So maybe once each field is validated individually as thoroughly as possible (crypto components are being used correctly, the correct distributed computing techniques are used for scaling to the moon, the economic variables are chosen correctly -Satoshi wanted Bitcoin to mimic gold-, the right incentives are given to miners and hindrances to attackers -Game Theory-), only empirical evidence can be used to finish the test. If by breaking Bitcoin you can silently steal millions and safely cash out (one hell of an incentive in my book), yet no one does it, I say it's secure enough. Not in a cryptographic or <individual traditional field> way, but in a pragmatic way. It just works.

jafaku 13y ago

> That is a political statement not a technical statement Just because it doesn't have the particular kind of specification that you are looking for, it doesn't mean that it's not technical. Being distributed is very much a technical feature, no matter how you look at it. I much rather have a working solution that might or might not be broken in the future (if it was too damn insecure, someone would have broken it already), than be stuck forever with the old banking system just because we don't know how to create the mathematical/cryptological model that you are used to and would grant 100% theoretical security. That would be like not building a website or shutting down a project mid development just because some component is found to be difficult to unit-test.

betterunix 13y ago
"I much rather have a working solution that might or might not be broken in the future" What does it mean for Bitcoin to be "broken?" That is my entire point here: if we do not have a clearly stated definition of what security means, then we cannot even talk about whether or not a system is broken. To put it another way, I would call any cryptosystem that can be attacked in polynomial time "broken," yet in the case of Bitcoin there is the 51% attack which apparently does not bother you -- and like I said, despite the fact that I call it broken, I cannot actually point to anything that would have required Bitcoin to be secure against such an attack. To put things in scientific terms, we are talking about falsifiable hypotheses. As an example, here is a commonly assumed hypothesis: the block cipher AES is a pseudorandom permutation. This is a claim that could be disproved i.e. we can falsify the claim by presenting an algorithm that efficiently distinguishes AES input/output pairs (with a secret key that the algorithm does not receive as input) from the input/output pairs of a random permutation. Consequently we can speak about the security of AES in a meaningful way, even though have no theoretical proof that it meets its security definition (only heuristic evidence and a lack of known attacks). My point about Bitcoin is that we do not have such a hypothesis, so we cannot even be sure that we mean the same thing when we say, "Bitcoin is secure" or "Bitcoin is broken." That is the point of having a security definition. Without such definitions, you can always make the claim that Bitcoin is secure, not matter what sort of attacks are carried out, because you can always just say that Bitcoin is not supposed to defend against those attacks.
- jafaku 13y ago
  > What does it mean for Bitcoin to be "broken?" It's pretty intuitive and I'm sure you know it: People being able to break the rules. Eg: create coins out of nowhere, move coins that weren't theirs, slow down transactions, unstabilize the market, etc. To say that it's not falsifiable is a bit of a stretch. It is falsifiable, we just don't know how to create the model you are looking for. Maybe it can't be done [with crypto]. Maybe it wasn't a competence of Cryptography to begin with, and we need an entirely new field (I know cryptographers have tried in the past, but to me this seems different). Don't you agree that Bitcoin involves too many disciplines to try to simplify it into a cryptographic formula? Something that breaks Bitcoin could even simply come from Economics, and your crypto model (if you ever find one) would be useless. Or, there is a cryptographic model, has been proven secure, and we just don't know it because its creator[s] wanted it to stay secret. Let's try with an analogy (you can attack it or say why it's bad). Your ship is sinking for some reason, there are no emergency boats, but you find out something that might be used to stay afloat. You hesitate, because you don't have physical proof of it like you did with the ship (an expert created the drawings, calculated the forces involved, etc.). But then you see a lot of people using it and it's working fine. The more time it passes, the more certain you are that it works. That's empirical evidence, and it drives a big part of scientific advancements. So maybe once each field is validated individually as thoroughly as possible (crypto components are being used correctly, the correct distributed computing techniques are used for scaling to the moon, the economic variables are chosen correctly -Satoshi wanted Bitcoin to mimic gold-, the right incentives are given to miners and hindrances to attackers -Game Theory-), only empirical evidence can be used to finish the test. If by breaking Bitcoin you can silently steal millions and safely cash out (one hell of an incentive in my book), yet no one does it, I say it's secure enough. Not in a cryptographic or <individual traditional field> way, but in a pragmatic way. It just works.