Comment by benchaney | Hacker News Reader

Comment by benchaney | Hacker News Reader

Parent Full thread

> I just wonder what goes on in your mind when you try to ponder the phrase "simplest possible blockchain"

What constitutes the "simplest possible blockchain" depends on what definition of the word blockchain you use. I don't see why that is particularly difficult to understand.

> because you don't think there's a specific single characteristic of a "blockchain", but there actually is.

I'm not sure what this is supposed to mean. The defining characteristic of a blockchain is a Sybil tolerant distributed consensus algorithm. At least that is the definition I (along with most people) use.

> https://medium.com/@lhartikk/a-blockchain-in-200-lines-of-co....

I'm not sure why you bothered to reference this. You have already established that there is one person who agrees with you. It isn't particularly relevant.

> Regarding Merkle: Even the Wikipedia page on "blockchain" explains the Merkle connection and it's precisely what I said, and the Merkle Wiki page says: "Hash trees are a generalization of hash lists and hash chains.", which was also my point.

The Merkle connection is that each block in the blockchain contains a Merkle Tree. That is not what you said at all. Also, if you clicked on the link for hash list or hash chain, you would see they are both very different from the linked list structure used in a blockchain. A hash chain is the repeated application of a hash function to a piece of data [0]. A hash list is a list of hashes, which is again hashed [1].

[0] https://en.wikipedia.org/wiki/Hash_chain

[1] https://en.wikipedia.org/wiki/Hash_list

A very literal reading of the Merkle sentence on wikipedia could conclude that it's stating each block of a blockchain must contain a valid Merkle tree, and so technically that sentence on the Wiki is incorrect; because, aside from re-hashing the prior hash (along with each node data), the blockchain algorithm doesn't care what the 'data' payload is. I was giving the Wiki editor the benefit of the doubt, assuming he just worded that sentence incorrectly.

The only difference between an order-one (degree-one) merkle tree (wired like a 'linked list' graph) and a blockchain, is that merkle doesn't require new data to hash on each non-leaf node, but has data only on leaf nodes. In merkle, each level up the tree is rehashing ONLY the actual child hashes below it, so that all hashes are ultimately determined only by "hashes of hashes" (recursively) of LEAF nodes.

So this means from a coding standpoint every order-one merkel is a single line of code away from being a blockchain. That is: The line of code that adds the 'node data' byte array to the SHA256 Hash Digester input stream, before outputting the node's hash. However this may also be a case where the Merkle wiki is not fully complete and accurate, because I cannot fathom that Merkle never thought of a case where each node contains data of it's own to add into the hashing. I bet the original source material from Merkle does include node data as an option, and would make him technically the actual first inventor of blockchain.

> A very literal reading of the Merkle sentence on wikipedia could conclude that it's stating each block of a blockchain must contain a valid Merkle tree, and so technically that sentence on the Wiki is incorrect;

That is not incorrect. That is exactly what is happening.

> because, aside from re-hashing the prior hash (along with each node data), the blockchain algorithm doesn't care what the 'data' payload is. I was giving the Wiki editor the benefit of the doubt, assuming he just worded that sentence incorrectly.

At the blockchain level, sure. But the Merkle tree exists at a higher level of abstraction.

> The only difference between an order-one (degree-one) merkle tree (wired like a 'linked list' graph) and a blockchain, is that merkle doesn't require new data to hash on each non-leaf node, but has data only on leaf nodes.

Right. This means that a Merkle tree of degree one can only contain one element, because it always have one leaf. That is why I called it useless in an earlier post.

> So this means from a coding standpoint every order-one merkel is a single line of code away from being a blockchain. That is: The line of code that adds the 'node data' byte array to the SHA256 Hash Digester input stream, before outputting the node's hash. However this may also be a case where the Merkle wiki is not fully complete and accurate, because I cannot fathom that Merkle never thought of a case where each node contains data of it's own to add into the hashing. I bet the original source material from Merkle does include node data as an option, and would make him technically the actual first inventor of blockchain.

It is only a one line of code difference, but it is a very important line of code. Changing the Merkle Tree in that way complete breaks the two main algorithms that make it useful [0]. Data structures aren't just about the way data is organized. They are also about ways to manipulate that data in that organization that allow for algorithms that wouldn't be possible otherwise. This is why Merkle didn't patent the idea of including data in non-leaf nodes, and that is also why most people don't consider any linked list using cryptographic hashes instead of pointers to be a blockchain. Because the way data is manipulated is just as important as how it is stored.

[0] https://www.certificate-transparency.org/log-proofs-work

Reply on news.ycombinator.com