3 ms·

Blockchain-the-datastructure is incredibly useful and widely deployed — e.g. Git histories are blockchains (well, a slightly extended version of a blockchain th

by pdpi 8y ago

Blockchain-the-datastructure is incredibly useful and widely deployed — e.g. Git histories are blockchains (well, a slightly extended version of a blockchain that allows you to represent a DAG, same difference). Blockchain-the-trustless-consensus-protocol has a fundamental challenge that makes it completely and utterly pointless for most applications — the oracle problem. You build up a whole system around a secure trustless chain-of-custody, but then require trust in some sort of outside entity for adding resources into the system. If you're willing to trust that entity, there's many simpler ways to sort out your chain-of-custody problems. Cryptocurrencies "work" precisely because they don't need an oracle. They don't track anything that exists outside the chain.

UncleMeat 8y ago
Only in a crazy loose sense. Git histories do not have enforced immutability or trustlessness. And they are really merkle trees, which have been around for decades. If blockchains (as described in this manner) are so amazing, why wasn't there hype for merkle trees in 1995?
- pdpi 8y ago
  I think you misunderstood my point — The reason I distinguished between blockchain-the-data-structure and blockchain-the-trustless-consensus-protocol is precisely because the word used to mean specifically the former but got coopted to refer to the latter. The blockchain data structure has been around for decades, and a Merkle tree is precisely the tree analogue to a blockchain's linked list. Trustlessness is a property of Nakamoto consensus, not of the blockchain data structure, but the combination of the two (or something that bears passing resemblance to that combination) is what usually gets called a "blockchain". Unfortunately this usage of the word is borderline useless because people keep loosening the requirements of the system so as to allow their harebrained setups to count as a "blockchain".

by pdpi 8y ago

UncleMeat 8y ago

Only in a crazy loose sense. Git histories do not have enforced immutability or trustlessness. And they are really merkle trees, which have been around for decades. If blockchains (as described in this manner) are so amazing, why wasn't there hype for merkle trees in 1995?

pdpi 8y ago
I think you misunderstood my point — The reason I distinguished between blockchain-the-data-structure and blockchain-the-trustless-consensus-protocol is precisely because the word used to mean specifically the former but got coopted to refer to the latter. The blockchain data structure has been around for decades, and a Merkle tree is precisely the tree analogue to a blockchain's linked list. Trustlessness is a property of Nakamoto consensus, not of the blockchain data structure, but the combination of the two (or something that bears passing resemblance to that combination) is what usually gets called a "blockchain". Unfortunately this usage of the word is borderline useless because people keep loosening the requirements of the system so as to allow their harebrained setups to count as a "blockchain".