[TD-Linux]

The introduction in the actual code actually explains it. It's not yet another rebadged Bitcoin Core codebase.

http://intelledger.github.io/introduction.html

The security model is totally different than Bitcoin - it's based on trusted nodes, leading to much different tradeoffs.

[cloudjacker]

> The security model is totally different than Bitcoin - it's based on trusted nodes, leading to much different tradeoffs.

Tradeoffs such as?

I read that and it is hard to conceptualize the pros and cons. Sometimes I think about why blockchains weren't considered as a solution to the byzantine generals problem before 2008. Was it fucking stupid? Was the consensus idea improperly dismissed? Never considered? I would say from 2000 - 2008 internet latency was 'good enough' in some places for this. Maybe 2004 - 2008 the storage space improvements were also good enough.

Sometimes I really wonder.

[TD-Linux]

One obvious one: rather than consuming tons of power hashing, miners buy Intel chips. Hashrate in Bitcoin is equivalent to number of Intel CPUs owned in this scheme. The Intel CPUs are mostly idle, so use very little power.

There are a lot of other interesting things that they could do with SGX, though just from reading the introduction it seems that they didn't attempt any other innovations.

[jaekwon]

You're trusting Intel, essentially.

It would be interesting to see Intel open their SGX platform so as for it to be configurable, and use it to attest Tendermint nodes. Accountable BFT consensus benefit from hardware attestation.

[beaner]

So they reinvented Ripple?

[cloudjacker]

Yes. They modified consensus algorithms from both Ripple and Stellar, they say it.