Message: 2
     Date: Mon, 11 Aug 2014 08:42:20 +1000
     From: "James A. Donald" <[1]jamesd@echeque.com>
     To: [2]cypherpunks@cpunks.org
     Subject: Ripple's consensus algorithm.
     Message-ID: <[3]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 [4]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.

References

   1. mailto:jamesd@echeque.com
   2. mailto:cypherpunks@cpunks.org
   3. mailto:53E7F54C.80805@echeque.com
   4. http://rippleinvestmentguide.com/