On 22 Sep 2005, at 10:16, Alex R. Mosteo wrote: > I'm going to read too the "Using the Small-World Model to Improve > Freenet Performance" paper.
I wouldn't bother, it is flawed. That paper proposes to replace LRU cache replacement with a mechanism that artificially enforces a Kleinberg link distribution, but have failed to realise that LRU cache replacement already achieves this in a much more natural and elegant manner (we have found this experimentally, and Oskar has been working on explaining this mathematically, with positive early results). IIRC Oskar suspects that the authors of this paper's observed improvement in performance is much more likely to be due to the fact that their algorithm was better at ensuring the presence of short links in their experiments than LRU - but not that LRU wasn't achieving a Kleinberg link distribution. Ian.
