3 ms·
In what world are you in that we have automate the selection of individuals aptitude for something, we can't even agree IQ tests are valid, or develop flawless
by Guthur 6y ago
In what world are you in that we have automate the selection of individuals aptitude for something, we can't even agree IQ tests are valid, or develop flawless visual pattern matching datasets that don't indicate that an African person is smiling or not in a passport photo.
And how do you know a VC biases are bad, is that your own bias playing out? If a VC had consistently bad biases I'm sure they wouldn't be in the game very long, their choices might not be always optimal but on average one would assume they come out ahead.
Of course maybe it's all just a coin toss and we can have an octopus choose.
- bra-ket 6y agoThe bias elimination idea is coming from financial industry where I worked for years. Investment decisions on various time horizons are mostly algorithmic these days, in the best firms. The odds of successful startup are pretty low and there is a good reason to believe that if you throw darts at the wall the success rate will be about the same. YC success has to do with large numbers of founders funded, with most of them , successful and unsuccessful, coming from particular cluster of population. Attracting the attention of that cluster is the second big reason for their success. I'm all for scientific approach, so random forest of octopi classifiers doesn't sound so bad. The difficult part is being a founder magnet like pg in the early days.