You may ask whom or what I'm subtweeting...
Quite a few examples, obviously. The math in poker is not complicated (though actual GTO is hard, and frankly poker harder than the other problems, math-wise). Most hedge fund math isn't hard, though much more than math to making money
Conversation
Baseball/sports analytics math isn't hard. Neither is the math in biology/genomics.
1
5
The math in machine learning... mostly isn't hard. Though sometimes it can be.
You do kind of need to know how gradient descent works. Though understanding measurement (stats) and loss functions is probably more value add tbh.
1
7
Say your model (actually someone else's model) is a black box. How do you optimize training, and measure your progress.
You'd be surprised how many ML graduates can try to answer this Q but aren't really comfortable around it.
Like they throw out suggestions, not really method
1
3
When I was interviewing candidates for my team at the hedge fund, I started all applicants with a few written questions. At some point might write a post about that... it's been a while after all.
But one Q was "do you have experience with custom loss functions"
1
5
Another was asking the candidate to model a (specific) ML problem with non-normal distribution of outputs. Most got it right, but I was surprised how many had no feeling for why this matters.
1
3
Let me also add a few fields where math is hard, and really good math does matter.
* math theory, obviously
* cryptography
* anything to do with proofs, and adversarial thinking
1
8
Having come from this background -- pure math -> cryptography -> poker and games AI -- I do think this kind of thinking helps in other fields. But not as much as knowing the subject area, caring about the problem, and good measurement.
1
5
Have fun out there. Your initial intuition might be correct -- and yes you, a meh math professor type, could with some effort have the best math in your applied sub-domain. Possibly much better.
Just don't think that math is a product.
1
4
Unless, again, you're doing poker AI, zero knowledge proofs, theoretical cryptography, etc.
Really glad I did that stuff. Not sure I could do it now, but could probably get back into it. Want to, some day...
1
4
Math do matter tho
"biggest predictor of whether a company will successfully adopt machine intelligence is whether they have a C-Suite executive with an advanced math degree. These executives understand it isn’t magic—it is just (hard) math." --
Replying to
Also why I've found it impossible, in practice, to work for a direct boss who's not a math (or advanced CS) PhD.
And now, having driven them all crazy... I work for myself 🤷♂️
1
3
But seriously, I've had some great bosses. Not just as people and managers, but just about all were exceptional mathematicians or strong CS PhDs.
My father's thesis advisor was a Nobel laureate (physics) so I grew up like that. The genial older men in my town were at that level.
1