int3

It's quite odd to flog the "free software" dead horse on a software entrepreneurship site

So one page of paper can store one page of data?

Writing, speaking, community building are the obvious ones

Gilbert Strang's linalg lectures on MIT OCW are amazing.

I'm currently doing MITx's Fundamentals of Statistics MOOC, which seems fairly similar to this one. The course material doesn't require too much understanding of real analysis, though the instructor does make cursory acknowledgements of the "technical details" he's glossing over, presumably for the more advanced students. The only analysis concepts we've used are continuity, differentiability, and convergence. It isn't a proof-based course: the instructor does prove some theorems in class, but the psets are all just computation. I do agree that it'll be harder to internalize some of the concepts w/o a background in analysis, but I think you can get a reasonable amount out of the course regardless.

That said, I did study analysis (but not measure theory) in college, and I don't entirely remember what I learned from calc vs analysis classes, so I may be a bit off here.

"Basic" is ill-defined, but I think sloonz is right about it only requiring calc + linalg, which most CS / Eng majors will have taken.

It's still way more of a meritocracy than many other disciplines. In what other field can you get an upper-echelon salary via non-credentialed routes like open-source projects and blogs?

Btw, you don't need to create an OSS project from scratch. You could also get visibility by making important contributions to an existing one.

It's a backend engine; it doesn't support languages out-of-the-box. That said, it's extensively used by the open-source Android bytecode optimizer [Redex](https://github.com/facebook/redex), so there's code readily available if you want to do Android bytecode analysis with SPARTA.

Signed Constant Domain: https://github.com/facebook/redex/blob/master/service/consta...

Constant Array Domain: https://github.com/facebook/redex/blob/master/service/consta...

Pointer domain: https://github.com/facebook/redex/blob/master/service/escape...

Type domain: https://github.com/facebook/redex/blob/master/libredex/IRTyp...

For a bit more context, see this blogpost we made: https://code.fb.com/open-source/sparta/

I made something like this many years back: http://int3.github.io/metajs/

Huh, I had no idea the guy who wrote the light transport thesis also built Adwords.

> If we pay everyone enough to make a living wage, then it should cost the same for me to grab my own menu, utensils and water bottles as it would for a server to do so.

I don't see how that follows. The opportunity cost per unit time for a skilled worker is typically greater than the living wage.

> (paradoxically, if, a week later they say they need another 5 million, giving them that at that time is OK as long as you think you still have ½% chance of getting all your money back, but you shouldn’t give them 15 million up front when you thought you had a 1% chance)

No paradox here... the evaluation of the situation in the first week should be [EV of getting all your money back with the 10M investment + (negative) EV of needing to put in more money later to keep the company afloat].

Type inference is a special case of abstract interpretation: https://www.irif.fr/~mellies/mpri/mpri-ens/articles/cousot-t...

(Not that I've read that paper in detail, mind you...)

If the reviewers are giving crappy code reviews, what makes you think they would be better at mentorship?

The papers there start from 2014 and continue up to the present.

I'd like to start with formalizing my knowledge of the basics -- hypothesis testing and confidence intervals. I do some performance tuning work, and it seems like a good idea to understand if my changes are making statistically significant improvements.

What books would y'all recommend for someone who has taken a couple of college-level proof courses but never took anything on probability and statistics?

It is only a terrible answer without basic income, which was the other point I was trying to make.

Fair enough. But experience can be garnered outside companies, and the Internet is making that easier, too: OSS development, nonprofit / pro bono work, etc.

Interesting points.

> The problem is that we are eating entry-level jobs, and in turn, creating an artificial scarce supply of "experienced" workers.

On the other hand, it is easier than ever to get cheap, basic education in a variety of things, with the Internet and MOOCs making information widely available. So entry-level jobs may be less necessary these days. On the other hand, it means the people who can afford to spend time acquiring this knowledge without receiving pay will have a bigger advantage than ever. Yet another argument for basic income, I guess.

Is rustc able to run typechecking without doing codegen?

Haskell's typeclasses automatically do dictionary passing for you; OCaml modules require you to specify the dictionary yourself. See http://okmij.org/ftp/Computation/typeclass.html

... the interviewee misunderstood the question. The interviewer understood it just fine.

Another approach would be to translate the compiled output of javac into ASM method calls: https://github.com/int3/java2asm

It's fairly unpolished at the moment, though.

Nice! I did something similar a while back: http://int3.github.io/metajs/

I tried evil-mode myself and came to a similar conclusion re integration.

And for package management there's vundle, neobundle, and pathogen. And after my recent discovery of vimproc, I'm now motivated to resume writing plugins!

If two weeks is sufficient, why aren't more people finding these things? Isn't there a big incentive to make sure that e.g. popular web frameworks are secure?

Interesting approach. One criticism: The paper mentions that compile times max out at 1.1ms "for the most complex function" in the benchmark (AES), and therefore it is sufficient to just compile everything. However, those benchmarks seem too small to justify that conclusion.