Hacker News new | ask | show | jobs
by geofft 3725 days ago
Here's the interesting part over existing blockchains / the reason why Intel is involved, from http://intelledger.github.io/introduction.html :

"This project includes a consensus algorithm, PoET (Proof of Elapsed Time), which is intended to run in a Trusted Execution Environment (TEE), such as Intel® Software Guard Extensions (SGX)."
2 comments
troymc 3725 days ago
Yes, I think that's the most interesting part of the Intel proposal. Rather than using lots of energy to compute a gazillion hashes (the work in Proof of Work), they rely on special processors to prove that they've waited for some amount of time (Proof of Elapsed Time). (The time is a random variable with known distribution.) That takes way less energy (cost). It's clever.

It got me thinking of other ways for computers to prove that a certain amount of time has passed, and I came up with this idea that takes advantage of the finite speed of light:

https://github.com/hyperledger/hyperledger/issues/18

link
jhallenworld 3725 days ago
Byzantine fault tolerance becomes simple if you have trusted hardware.

I've been trying to understand the base of trust in this: I think it's Intel EPID: it boils down to a secret key burned into the hardware and managed by the ME. What happens if Intel's signing machine is exposed?

link
geofft 3725 days ago
That's an interesting point. If the threat model assumes that Intel's secure computing infrastructure will not be compromised, can't you just use remote attestation to prove you're running an unmodified Foocoin client on physical Intel hardware, and have a traditional consensus protocol? Then you don't need proof-of-time or proof-of-anything-else, just proof-of-physical-Intel-hardware.

Either way, whoever controls 50% of the genuine (where "genuine" = "signed by Intel's master key") Intel hardware gets to control consensus, right?

link