Hi, First of all, I'm still very very n00b on networks, this is just a quick excursion so don't take this too seriously and/or consider me a crank :) Anyway, today in our university's exercise class for networking course we were taught the famous "count-to-ínfinity" problem in distributed distance-vector shortest route finding. I.e., if one link breaks down, other links can go bizarre with their distance vectors. Then we were presented a few solutions that attempted to allieviate the problem, but they were hard and none of them worked in general. So I made up a following algorithm: Assume a link between routers A and B breaks down. Now, router A's shortest distance to B is set to infinity, and router A will *stop* taking *any* distance information from its neighbors until it has sent "scandal messages" to all its neighbors (let's refer this to as "scandal state"). The semantics of the scandal message is, "I, router A, cannot provide routing to B anymore". Now, this "scandal message" is sent to all neighbors of A, let's denote each of them C. Now, if C's routing table says that "the shortest route to B is through A", C *itself* is set to the "scandal state", and it sends scandal message "I, router C, cannot provide routing to B anymore". Then this process is just repeated. Now, my intuition says this fixes the count-to-infinity problem completely. That is, I can't come up with any counterexamples =) There might be some technical restrictions too of which I'm not aware. In every case, I'd be pleased to hear your comments. Thanks for your time, -- Mikko