[ascend to soapbox] [begin sermon] I like to think of news.yc as a social learning machine. It would be nice to see karma computed more as a function of the c

[ascend to soapbox] [begin sermon] I like to think of news.yc as a social learning machine. It would be nice to see karma computed more as a function of the contribution to the community rather than pure popularity. In terms of information content, the PG links are going to be low because almost everyone here would have read the article regardless. Seriously, is the 100 karma points gained through linking to a pg essay worth as much as the 100 earned through 10 informative (in the classic suprising sense) links? We seem to be converging to this eigenvector [a(1) a(2) ...a(n)] where a(1) == pg_essay_link, a(2)== mattmaroon_essay_link...etc. The reason I believe pg converted this forum to hacker news is that the learning machine was stuck on the search surface in a local minimum and he wanted it to 'learn' new things. We can't learn more if we reward behaviour that is not truly informative. He could put a damping coefficient on karma obtained from known links, but a better bet would be for the community to not reward the links as much. I have done my share of dumb, uniformative submissions and will try to make better contributions in the future. [end sermon] [get off soapboax]