On Sun, 30 Mar 2003 palomitas at hushmail.com wrote: > > On Fri, 28 Mar 2003 16:08:31 -0800 Ian Clarke <ian at locut.us> wrote: > >We need to get probabilistic caching in the code ASAP - the current > >level of load most nodes are getting is totally disrupting the > >clustering effect from what I have seen in my node's diagnostics. > >[snip] > > > >I propose making this the highest priority at this time. > > I still vote for considering the "Node learning contact probability" > thingy... > (http://hawk.freenetproject.org:8080/pipermail/devl/2003-March/004842.html).
The node-learning-contact-probability scheme is too simple. Node uptime cycles are based on a combination of many many different natural waveforms: Is it the weekend? Is it near payday and the node operator has nothing else to do? Has some whiz-bang new freesite just gone up? Is it computer maintenance time? An hour by hour analysis wouldn't catch the majority of this irregularity. To actually implement some learning algorithm that had any sort of usefulness, and wasn't just scraping a couple more node-hours out of the network, would probably be pretty difficult. I would guess the payoff wouldn't be worth the effort. -todd _______________________________________________ devl mailing list devl at freenetproject.org http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl
