Best blogroll ever
Or possibly “Best use of RSS in a weblog ever.” ReadingEd.com. Click any link with an exclamation point and feel your mind expanding.
Could someone with a better grasp of programming and math than me figure out how to recommend sites like that for me, please? Things like Mark’s recommended reading and various weblog neighborhoods are a nice start, but I don’t really need to know the other weblogs within my cluster: if everyone I link to links to site A, rest assured I either read it or hate it. What I need to know is what’s linked to by a fair number of people in a cluster that I’m barely linked to: Stuart and Simon seem to know all sorts of interesting people, but because their cluster is fairly loose and I’m not strongly linked to it, I don’t get recommendations from it. I suspect that doing something like throwing out the links from people whose list is too similar to mine, and then recommending the top links from the ones that are left (or maybe with the “I’m really jaded” box checked, recommending the least linked among the least similar) would give more interesting results. All I’m sure of is that I don’t need any more recommendations that tell me I might like to try this weblog by some guy named Kottke. You know, if I haven’t heard of it before.
Yeah, ”recommended reading” always tells me I should be reading Scripting News. :) ”Recommended Reading” is designed to show me the outskirts of my community, nothing too far out, but stuff I don’t read on a regular basis or stuff that other people know about that I don’t. This introduced me to unix-girl.com, tidakada.com, and a few other sites that are quite interesting but that I had never heard of before. YMMV.
To quote Leslie Orchard, ”Ideas are fucking worthless.” Complete source code is at http://diveintomark.org/projects/recommended_reading/ — if you don’t like the results, all you have to do is rewrite the algorithm. Maybe feed it your list of referrers instead of your own blogroll (I did this on my own referrers and got some, um, interesting sites). Or weight links differently. Or whatever.
Ideas are fucking worthless, but at least they are better than vague wishes: somebody has a good idea but isn’t smart enough to implement it, that’s one thing, but when all they’ve got is just a vague sort of whine about how it ought to be different… Hey! Wait a minute, we’re talking about me! Damn.
The good news is that NightmareHost finally upgraded my server, taking Python to 2.1.something rather than the 1.5ish it was before along the way, so I can stop trying and failing to install a decent version myself and start trying to wrap my head around Python. Then after a few months we’ll see how much I remember from those statistics classes twenty-odd years ago.
As long as you remember that statistics are as fucking worthless as ideas, you should be fine.
The problem might have also been in the database that Mark had to use. There are a limited number of places that have lists of blogs and where their links go (rather large databases, but then how big is the ’net?).
I have no idea if I could rewrite the algorithm, and if I did, I don’t think my server hosts Python, anyway. Sometimes, just ”wandering around” works as well as anything.
I agree!
phil ringnalda dot com. Wow, I agree, this is an incredibly cool use of CSS combined with RSS to let