So, now that PARSEC is out everybody is very pleased with the progress and the new consensus mechanism. But with regards to the consensus mechanism, I was wondering what the downsides are? I’ve heard all the benefits but not a lot on the doubts.
I don’t want to spread FUD, only want to learn more on the mechanism. I’m not extremely tech savvy so let’s try to keep it in layman’s terms.
Example: PARSEC is currently not fully asynchronous. What kind of issues does that raise?
One aspect that has recent discussions is that of bad actor nodes.
As far as async is concerned this is more to do with ensuring consensus comes to completion. For SAFE this is not really a problem. The ideal is though to have fully async. Although there has to be limit to how long waits for consensus last.
PARSEC is the replacement for the current consensus mechanism and as such as long as it works it is a major improvement on the current mechanism. Faster and proven to come to consensus with a higher resistance to bad actors. In that sense its an improvement
As far as a stand alone consensus mechanism the downsides will depend on what system it is slotted into. For SAFE it is an improvement. For other systems there maybe downsides depending on application.
Nothing while the algorithm works and and comply with the properties of termination, validity, integrity and agreement.
This little synchrony, and the use of a concrete coin, are tricks that are used to generate indeterminism that allows circumvent the FLP impossibility.
The FLP impossibility tells us that can not get a consensus, in a completely asynchronous system, when one of the nodes fails. This is a basic principle in distributed computing and much of the research in this field is to find ways to achieve consensus despite this principle and it’s what Maidsafe tries to achieve with Parsec.