bird idea haver

Scientist

[bird idea haver @unitroot1]

me announcing my team’s progress on our Q4 deliverables

[bird idea haver @unitroot1]

@LabGrownNeet sholto is from oz

[bird idea haver @unitroot1]

@dotnetschizo yes feels so good. i beat GTA vice city and also got more done because i didn’t give a f what anyone thought anymore

[bird idea haver @unitroot1]

@rickasaurus Drone drone = new Drone()

[bird idea haver @unitroot1]

@PosterInternet whole thread is completely bizarre, down to the assumption that you can detect cheating by finding extreme values in the test score differences

[bird idea haver @unitroot1]

@robkhenderson endogenous

[bird idea haver @unitroot1]

@Cam580851696399 @clairlemon it increases the cost of labour

[Pope Leo XIV @Pontifex]

As evidenced by the unbridled promotion and implementation of technology at the expense of human dignity, we are truly experiencing an eclipse of the sense of what it means to be human. It is imperative to recover an understanding of the true meaning and grandeur of humanity as intended by God. It is in this sense that the challenge we currently face is not technological, but anthropological, and it is my hope that the Encyclical Letter to be published within a few days will contribute to answering this challenge.

[bird idea haver @unitroot1]

@shakoistsLog W

[bird idea haver @unitroot1]

@MaxKemeny @peter_tulip @cmkusher per OP, you can USE a model. to rely on a model is to outsource your responsibility to think about the problem, to consult many sources (and there are always many) and to deny ownership when it inevitably turns out that you were badly wrong

[bird idea haver @unitroot1]

@shorewalker1 @peter_tulip @cmkusher model uncertainty is generally not quantifiable. it's vibes all the way down

[bird idea haver @unitroot1]

@peter_tulip @cmkusher the alternative is having some epistemic humility and acknowledging that estimates are highly uncertain, rather than using false precision to fool yourself and others into thinking you have the final answers on important policy questions

[bird idea haver @unitroot1]

@JonathanShedler illich was talking about this half a century ago

[bird idea haver @unitroot1]

@RizviAbul @abc730 spoken like a true bureaucrat

[bird idea haver @unitroot1]

@TheReal50972902 @quant_arb 3) idgi. if you want an interpretable model, then think about the causal graph. don't make decisions based on the data (or else you bias the model). if you want a predictive model then throw the kitchen sink at it. don't try to thread the needle

[bird idea haver @unitroot1]

@TheReal50972902 @quant_arb 1) ok if we are talking about pure OLS. otherwise regularisation can do this job. let crossvalidation decide

2) agree but two independent features can be strongly related conditional other features. bivariate orthogonality is unreliable

[bird idea haver @unitroot1]

@JeremiahDJohns *distance

[bird idea haver @unitroot1]

@geoffreylitt for the same reason i dont read docstrings. reading code is in many cases easier and more reliable

[bird idea haver @unitroot1]

@valhalla_dev it’s just a bit of fun

[bird idea haver @unitroot1]

@lurkingforev @shakoistsLog python gives you much less freedom at runtime. e.g. python will kill your program if you do something that might move a list’s elements around in memory as you are iterating over it. cpp runtime is happy to let you try it

this is partly why it’s so slow; these checks aren’t free