Sherlock Holmes stories are bad because they glorify overfitting.
Nathaniel Bechhofer
31.8K posts
- Replying to @ATabarrokThat piece is so embarrassing for socialists. Like at least use one of the clever arguments like wastefulness of redundant logistics infrastructure rather than something that can't stay coherent within the length of a newspaper opinion piece.
- Imagine reading these arguments in a typical discrimination case and think how much credence you’d give…conditioning on post treatment variables, arguments about unobservable quality etc all here
- Replying to @jmhorpComposition effects: population growth is a lot faster on average for SSA countries with low incomes, so that offsets the growth from the countries that have done well
- Replying to @laurenbalikThat's got to be selection bias explaining which H-1B workers you encounter; probably 60%+ of H1-B holders can find x=2 just fine
- I’m very pleased to be sharing this, a @salonium and @bechhof expression of a radical vision for more effective scientific production! worksinprogress.co/issue/the-spee… It’s time to rebuild scientific institutions for the 21st century, complete with frontier tools and practices.
- If you read one thing about John Nash today, read this amazing post from @Afinetheorem afinetheorem.wordpress.com/2015/05/24/the…
- Think of Steven Pinker’s Blank Slate and Better Angels of Our Nature as analogous to the Old and New Testament.
- Replying to @fchollet
- Replying to @rubenbmathisenIt seems pretty misleading to condition on working more than 35hr/week to begin with
- Replying to @ATabarrokThat's... kind of awesome. All they needed to complete the system was a well designed matching mechanism and that would've been ideal.
- Replying to @StefanFSchubertI suspect that time invested in make-up is underreported and time invested in exercise is over reported tho
