Re: cypherpunks Digest, Vol 14, Issue 11

Message: 2 Date: Mon, 11 Aug 2014 08:42:20 +1000 From: "James A. Donald" To: cypherpunks@cpunks.org Subject: Ripple's consensus algorithm. Message-ID: <53E7F54C.80805@echeque.com> Content-Type: text/plain; charset=UTF-8; format=flowed

Bitcoin's consensus algorithm is weight of computing power, which is OK as long as weight of computing power aligns with interest in bitcoin being a useful currency.

Weight of stake would be better, but so far I am unaware of any satisfactory proposals for weight of stake.

This is called "proof of stake" not weight of stake and is different from Ripple's consensus process.

Ripples consensus algorithm is weight of club members, and the process for getting into the club is opaque, as are the interests and incentives of the existing club members.

This doesn't really describe how Ripple works. Ripple relies on the agreement of 80% or more of validator nodes per gateway to verify if a transaction took place or not. I write more about this in http://rippleinvestmentguide.com/

I would suppose one gets into the club if no existing member blackballs you, which would be fine if there is already sufficient diversity of interests within the club.

Only if 80% of the networks validator nodes do that which is in practice not likely to happen on a large enough scale.

It is not obvious to me how well the ripple consensus algorithm would work in the event of substantial conflicts between club members, or bad behavior by club members, or bad things happening to the network.

Has it been analyzed for performance in the event of bad behavior by some club members?

I'll ping David Schwartz one of the co-inventors about this question.