Twitter | Pretraživanje | |
Zane Bitter 7. pro
Odgovor korisniku/ci @paulg
I enjoyed this essay, but you should probably acknowledge that YC explicitly selects for people whose default attitude to everything is to look for a way to hack around it. It's literally a question on the application.
Reply Retweet Označi sa "sviđa mi se"
Paul Graham 7. pro
Odgovor korisniku/ci @zerobanana
See footnote 4.
Reply Retweet Označi sa "sviđa mi se"
Andrew Stepner 8. pro
Odgovor korisniku/ci @paulg
What do you think of the partial counter argument that is sometimes heard: that skill at hacking tests is an indicator of intelligence and therefore sometimes a worthy measurement?
Reply Retweet Označi sa "sviđa mi se"
Paul Graham 8. pro
Odgovor korisniku/ci @andrewstepner
That's true. But it gets in the way of learning.
Reply Retweet Označi sa "sviđa mi se"
Oisín Moran 7. pro
Odgovor korisniku/ci @paulg
Small typo: "doing working in a field" -> "working in a field"
Reply Retweet Označi sa "sviđa mi se"
Paul Graham 7. pro
Odgovor korisniku/ci @TheOisinMoran
Fixed; thanks!
Reply Retweet Označi sa "sviđa mi se"
Tristan Homsi 7. pro
Odgovor korisniku/ci @paulg @nntaleb
"Tests can be divided into two kinds: those that are imposed by authorities, and those that aren't. Tests that aren't imposed by authorities are inherently unhackable," Reminds me of this from
Reply Retweet Označi sa "sviđa mi se"
Ryan Petersen 7. pro
Odgovor korisniku/ci @paulg
I wonder how a company can do hiring interviews and performance reviews that aren’t hackable?
Reply Retweet Označi sa "sviđa mi se"
Karri Saarinen 7. pro
Odgovor korisniku/ci @typesfast @paulg
Work trials quite unhackable for interviewing, since you test the actual ability to work on a particular product and how they communicate. Performance reviews are harder since anything you measure can be gamed (most people don’t realize it though)
Reply Retweet Označi sa "sviđa mi se"
Eric Braun 7. pro
Odgovor korisniku/ci @paulg
Most schools perform what Data Scientists know is wrong — they overfit the models (students) so they can only work well with the training data. Once in the real world, they fail. Broader experiential learning works better for people and machines
Reply Retweet Označi sa "sviđa mi se"