Comment by Animats | Hacker News Reader

Comment by Animats | Hacker News Reader

All you get from the Bitcoin blockchain is the ability to store some public data that's hard for anybody to alter once the data has settled. That's useful for a few special problems, but not widely useful.

The cost of maintaining that blockchain is insanely high - all those giant "mining" operations. Large numbers of machines have to have a copy of the blockchain and track all updates to keep everybody honest. It takes about half an hour before a data item can be considered solidly committed. Even for Bitcoin itself, this is is an operational headache.

If all you need is a certified timestamp on something, several European certificate authorities run free trusted timestamping services. You send in a secure hash of something, they add a timestamp, sign that, and send back a reply.

"All you get" is a distributed & anonymous consensus algorithm that's resilient to malicious actors.

That is the value. If you don't need these properties, sure, you don't need blockchain.

As for the maintenance cost: Proof-of-Stake is making interesting progress. There are initial ideas around not having an entire copy of the blockchain.

I'm not aware of commit-time reduction work, but would welcome pointers. (Either way, you'll always have a commit-time issue in a distributed system. You might reduce it somewhat, but it won't go away.)

Tendermint has the strongest guarantees against double-spend attacks.

http://tendermint.com/docs/tendermint.pdf

>All you get from the Bitcoin blockchain is the ability to store some public data that's hard for anybody to alter once the data has settled. That's useful for a few special problems, but not widely useful.

You get more than that, you get a new social trust and consensus model that has never before existed. Eg, the ability for an indefinite number of anonymous, mutually untrusting participants to nevertheless agree on the state of a continually updating communal data source.

The tech is still in alpha stages, but is potentially powerful, and hence isn't cheap in terms of resources, and it shouldn't be. In fact, the big question and ultimate objective now is, is it possible to make the consensus system even more secure at the cost of even less efficiency?

Reply on news.ycombinator.com