David Dyer-Bennet wrote: > Damien Sullivan <phoenix at ugcs.caltech.edu> writes: > > > > http://www.geocities.com/Area51/Cavern/6113/t100249.txt > > > > God Stalk is 57! Woo! > > > > Except then I look at the top, and see _Armor_ at 14, and I > > remind myself Internet votes mean squat. This list had a much higher correlation to my own rankings several years ago. A lot of stuff I wouldn't touch with a ten foot pole has drifted up into the ranks of classics. I'd like to devise my own ranking system that took into account the voter's reliability*. See the following link, and skip down to the question 'Q: How are the books ranked?' for a description of how the tIT100SF/F List is generated. http://www.geocities.com/Area51/Cavern/6113/top100doc.txt > Anybody ever heard of _Krew Elfow_ by A. Sapkowski? (#5) No, but that doesn't surprise me since it is not available in English. The people who've voted on it sure liked it. About half the votes for it so far have been 10s. (Note the {F} in the ranking, and see the following link for a somewhat outdated extended listing that includes vote distribution information) http://www.geocities.com/Area51/Cavern/6113/extend13.txt At this point roughly a quarter of the top 100 are books I've read and would recommend. About ten percent are books I've read and wouldn't recommend. About a third are books that are by authors that have been recommended to me, but for one reason or another I haven't gotten around to reading. The remainder are roughly half and half between authors I avoid like the plague and one's I've never heard of. A few years back those portions would have been more like 45/5/30/10/10. *My idea of reliability would discount readers that consistently ranked a book either a 1 or a 10. Also, readers who vote only on a few books would get much less weight than ones who vote on a much wider population. I'd be tempted to also discount contraries votes, but I think that might get risky. Might drop some number of high and low votes.