P2P Web Search on Randomland

[zourtney]

(We briefly discussed this elsewhere, but I thought I'd give it its own thread.)

I recently came across YaCy. It is a decentralized peer-to-peer web search protocol. You can try it out at peer-search.net, but in pracitce it runs a client/server setup on your own machine to help spread the disease (aka expand the network). Honestly, the search results are aren't that great right now (and are possibly inappropriate). But I like the idea! (Warning: you'll get no results with NoScript enabled because it connects to dozens of clients via javascript) Since the Randomland server does almost nothing all day long, I thought it might be fun to set this up on it.

I installed it, just to see how it'd work. For whatever reason, I was getting even less relevant search results than from above-linked site. Maybe disabled Javascript, maybe firewalling, maybe just the peers I was connected to at the time. I'll try it again tonight. I did notice it being a huge memory hog as it started indexing the internet. Go figure. Supposedly there are ways to throttle it, though I actually find those settings yet.

A cool feature is that you can give it one (or more?) sites to index. We would, of course, want to index Randomland and make ourselves known. If it does a decent enough job, that could override our site-wide search, I suppose (then we could search drupal, forum, and wiki all from the same box and do it Googlelessly)p We should also index cool places, like...um...well, surely you can think of something on the internet worth indexing.

Just a thought.

[zourtney]

I did end up playing with this last night. It's actually cool. But participation is low, so search results aren't terribly relevant. It can also be really slow. But it's fun downloading the internet I think I had it set up to share my indexed results properly (through Virtualbox, through router's virtual server).

Yeah.