I've been having a kind of unspecified dread about algorithms. Even in their complex forms they seem sort of... lazy.

Create Rule X for Y content, assign % of group ß to list ω, and on and on. I mean, algorithms such as the ones from Google must necessarily ignore certain things and elevate others. But what biases are in each variable and constant? Was some number hardcoded in there without a second thought, affecting search results, elevating the 'wrong' sort of information and whitewashing the 'true' or 'correct' kind?

posted by ao: 1412 days ago