"smart contracts can also do things that normal contracts can’t, such as enforce a set of rules entirely through unbreakable cryptography." Isn't "unbreakable" a bit of a dirty word in the security community? Is there really such a thing as "unbreakable cryptography"?
"Dirty" is one way to put it, yes. I'd personally use something like "false god" or "blasphemy" :). "Unbreakable" is a naive way of describing cryptographic algorithms, because it preempts conversations about intractability assertions or complexity analysis...modern cryptography accepts as a premise that "unbreakable" is not a reasonable goal, which is why we work by quantifying computational cost.
When I see someone use the phrase, "unbreakable cryptography", I mentally discount their authority to speak about cryptography and become more skeptical (this also applies to people who write things like "bank-grade security" or who tell me "AES-256" when I ask them how they're encrypting data, as if the key size is more important than the block cipher mode or confidentiality/authentication construction).
> Is there really such a thing as "unbreakable cryptography"?
A one-time pad, with the following provisos:
1. It has a truly random seed, not a pseudorandom one;
2. It is at least as long as the plaintext;
3. It is never reused (for the same plaintext, in whole or part).
It's straightforward to see why this guarantees information-theoretic security - you have no way of knowing when you've recovered the correct plaintext. There's nothing to brute-force.
In the security community we don't use one-time pads because we dare not touch the sun: they are extremely difficult to implement correctly, and even if you do implement them correctly you have traded off an inordinate amount of practical usability for a relatively small improvement in information theoretic security.
Asking people to 1) use true randomness to seed keys, 2) generate a new secret key for every message, 3) never reuse the same secret key and 4) keep each secret key securely confidential is extremely prone to error. It's often a fetishized ideal among people who first read about it, but in practice it's just not worth it.