http://research.microsoft.com/en-us/people…

http://research.microsoft.com/en-us/people/aguilera/stumbling-chapter.pdf

"In the consensus problem, each process proposes some initial value, and processes that do not fail must reach an irrevocable decision on exactly one of the proposed values. The consensus problem captures an essential component of replication in distributed systems: the fact that replicas (processes) need to agree on the next request they handle, so that they can remain in identical states."

The consensus problem is a formal problem in dealing with distributed systems, and apparently has attracted a lot of work in the last few decades. I'm glad to find that the consensus approach in CS recognizes a difference between consensus and unanimity; there's even a paper titled "Uniform consensus is harder than consensus"!

You can read more here:

http://en.wikipedia.org/wiki/Consensus_(computer_science)
The Aguilera article linked above ends with the following line: "The consensus problem is at the heart of replicated distributed systems, which are increasingly becoming a vital part of our society in areas such as commerce, banking, ?nance, communication, critical infrastructure, and others."

Understatement of the century.

Does anyone know of any specific and deliberate attempts to apply some of the consensus-solving algorithms that are described in this literature to handle problems with organizing crowds of people, and not just for shaping networks of computers?

http://research.microsoft.com/en-us/people/aguilera/stumbling-chapter.pdf

Submit a comment