4 ms·

I've been following Siacoin development very closely for the last couple of months. Their devs seem like one of the few remaining people in crypto that actually

I've been following Siacoin development very closely for the last couple of months. Their devs seem like one of the few remaining people in crypto that actually care about developing a working product/coin, vs just abusing the unregulated market and getting filthy rich. Kudos to the team for getting on the front page of HN. I'm sure this will raise awareness of their product at least a little bit. Just looking at the coin itself, looks like it jumped 10 points since this blog appeared.

placeybordeaux 9y ago
For those that aren't following it seems that siacoin is looking to provide an efficient distributed storage network made up of individuals that offer a contact to store data. To offer the contract the host puts up collateral that will be given to the renter if the host fails to provide a proof of storage (my guess before reading the paper is that the contract gives a hash and a file size and the host is responsible for giving a file of that size and that hash if there is a dispute). Seems cool. I have an always on computer & ~100 GB free, I might experiment with running a node.
- nemo1618 9y ago
  >my guess before reading the paper is that the contract gives a hash and a file size and the host is responsible for giving a file of that size and that hash if there is a dispute Close! Sia uses Merkle trees to construct the storage proof. The storage contract specifies the Merkle tree root hash. The host must then provide one of the "leaves" of the tree (64 bytes of actual data), along with the other subtree roots in the tree necessary to construct the root hash. The consensus algorithm can then verify that by hashing the leaf and combining it with the other subtree roots, the top-level root can be obtained. This is a lot more compact and efficient than sending the entire file contents. The only trick is selecting which leaf to demand from the host -- if it could be predicted in advance, the host could cheat and store less data. But it can't be purely random either, since all nodes on the network need to pick the same leaf to arrive at consensus. So the leaf is chosen using existing entropy in the blockchain (e.g. the hash of the last block), so that it's both deterministic and difficult for an attacker to control.
  - placeybordeaux 9y ago
    Neat! When I first learned about bitcoin I spent a fair bit of time navel gazing about how to make something like siacoin. Excellent to see an ostensibly well made implementation. Any experience actually using it for storage? Also any other coins actually doing some innovative work? A friend of mine got excited about Golem, but somehow I don't have a good feeling about that, wish I could easily short it...

placeybordeaux 9y ago

For those that aren't following it seems that siacoin is looking to provide an efficient distributed storage network made up of individuals that offer a contact to store data. To offer the contract the host puts up collateral that will be given to the renter if the host fails to provide a proof of storage (my guess before reading the paper is that the contract gives a hash and a file size and the host is responsible for giving a file of that size and that hash if there is a dispute). Seems cool. I have an always on computer & ~100 GB free, I might experiment with running a node.

nemo1618 9y ago
>my guess before reading the paper is that the contract gives a hash and a file size and the host is responsible for giving a file of that size and that hash if there is a dispute Close! Sia uses Merkle trees to construct the storage proof. The storage contract specifies the Merkle tree root hash. The host must then provide one of the "leaves" of the tree (64 bytes of actual data), along with the other subtree roots in the tree necessary to construct the root hash. The consensus algorithm can then verify that by hashing the leaf and combining it with the other subtree roots, the top-level root can be obtained. This is a lot more compact and efficient than sending the entire file contents. The only trick is selecting which leaf to demand from the host -- if it could be predicted in advance, the host could cheat and store less data. But it can't be purely random either, since all nodes on the network need to pick the same leaf to arrive at consensus. So the leaf is chosen using existing entropy in the blockchain (e.g. the hash of the last block), so that it's both deterministic and difficult for an attacker to control.
- placeybordeaux 9y ago
  Neat! When I first learned about bitcoin I spent a fair bit of time navel gazing about how to make something like siacoin. Excellent to see an ostensibly well made implementation. Any experience actually using it for storage? Also any other coins actually doing some innovative work? A friend of mine got excited about Golem, but somehow I don't have a good feeling about that, wish I could easily short it...