Contra Thiel, the key distinction between stats/calculus is not indeterminate/determinate, but discrete/continuous blakemasters.com/post/234357439
Conversation
Replying to
Stats is intuitive in discrete (dice, coins, balls in urns), unintuitive in continuous. Calculus is opposite. This has deep effects.
1
1
6
Replying to
Calculus being more natural on continuum means discrete realities feel like approximations. You try to go more fine-grained for better truth
1
2
Replying to
Calculus is more fundamentally indeterminate view of the world: zoom in enuf (and you can do so infinitely in continuous), more bits appear
1
1
3
Replying to
Stats on the other hand, leads naturally to determinacy through finiteness. Like discrete set of futures with countable branching structure
2
Replying to
Of course, at limit, this gets to philosophical imponderables like quantum scale, or digital vs. regular physics.
1
Replying to
You get misled when you cherry pick example like predictability of space orbits as "proof" of precise predictability of calculus world.
1
Replying to
Try Navier-Stokes equations (also calculus!) for predicting turbulent fluid flow. Opposite of predictable.
1
1
Replying to
Equally, stats can lead to highly predictable results, as in dominant game theoretic strategies over long iteration horizons.
1
1
Replying to
Not nerd-quibbling with Thiel's model for no good reason. This has serious implications for mental models on the Thiel 2x2.
1
Replying to
The motivation for that 2x2 is to talk about luck and success as constructed by society, but the account is simplistic.
Replying to
ε/δ thinking gets at a more fundamental question: when do small changes lead to small effects, versus huge, rapid, snowball effects?
1
3
