How to leave academia (for science PhDs)

Come on, you don't think you can shock people with Karl Marx, do you? In 2012? Try some Maistre, Carlyle, Froude. Maybe even George Fitzhugh or R.L. Dabney. Here, this'll get you started: http://books.google.com/books?id=an2LWYTh2ewC

No, I thought someone was going to fund me because my adviser is well-known and prestigious within the field.

Maybe you're not such a bad bureaucrat after all! But I meant not while you're in grad school, but after. Of course, if you stick with your mafia, and it's a good mafia, you can keep the party going. Perhaps. Definitely make sure you don't alienate your betters.

Even the successful industrial programmers I know, unless they became millionaires off a start-up, did not actually make enough money to "settle down" out of working and "found your own little university of one."

Yeah, that's what I mean. Dedicate yourself completely and unreservedly to the foul, foul art of money-grubbing, and/or money-saving. Or possibly money-stealing. Anything is better than either conning investors, or conning the government. Have you considered growing pot?

(I won't say this worked perfectly for me, but it did give me time to write my purely functional operating system. Unfortunately this took longer than expected and I'm now kinda broke. And I really can't turn my daughter's closet into a grow-room. So, we'll see if anyone thinks they need a purely functional operating system... Kickstarter, here I come.)

What is the difference? The difference is that when you invite investors to invest in your new location-based social network for morticians and coroners (or whatever the fsck), you are making an actual go at a business. Whereas when you invite investors to invest in your new programming language, you are conning investors.

Conning the government means making flagrantly bogus and absurd claims in your grant applications. Please bear in mind that all your competitors will be doing this. If you don't intend to play by the same rules, you need some other way of stabbing your competitors in the back before the finish line. Metaphorically speaking, of course.

I am descended from a long line of Beltway bandits. I recognize the patterns and practices of the tribe at once. Indeed I recognized it in my first week at Berkeley, when they sat us down and gave us the basic lecture on what it means to have a career in "science" today.

I believe the numbers given were that half of us first-year grad students would finish, half of those would find teaching jobs, and those who found teaching jobs would be dividing their time three ways: 1/3 teaching, 1/3 research, 1/3 grantwriting. I found this broadly credible and still do.

So, that's only 1/3 of your day spent as a professional liar (and/or backstabber). You might as well try to spend only 1/3 of your day as a professional crackhead, whore, etc. But hey, it's worth a try.

As for trolling - absolutely! Laugh while you can, monkey-boy...

Lesson 1: your research project is not an open-source project. Until it's done.

Lesson 2: extraordinary claims require extraordinary evidence.

That said, if you want to see some very odd-looking code:

https://github.com/cgyarvin/urbit/blob/master/jupiter/sys/20...

Scroll down to the RSA implementation around line 2500. You won't learn anything. And why does a self-compiling compiler need an RSA implementation, anyway?

There's a great anecdote about Jessica Mitford from the '50s. Mitford had spent years writing an autobiography and, during this process, had constantly been passing drafts around to all her commie-princess friends. Of whom as you'd imagine there was a great number. Finally she got an agent, whose name I forget but I'm sure was a big New York queen of some kind, who shut her down and forced her to finish the book instead. "It's like parading around all day," the agent told her, "in your underwear."

The system has only been self-hosting for 7 or 8 months. Until November the parser was still de facto impure. The packet decoder has never been tested and isn't even in the kernel. String syntax is nonexistent, the prettyprinter prints pure dog crap, Nock 6 was an aesthetic disaster so Nock 5 is actually Nock 7 but with the calling convention reversed, persistence is nonexistent, stack overflow recovery is nonexistent, profiling has succumbed again to bit rot, HTTP is completely untested, etc, etc. I could explain all these things to you, or I could finish putting my clothes on.

Obviously it's on a public repository and all the code is PD. Crap you can even build the thing if you like. You can even fork your own kernel and try to compete with me! Don't expect any documentation, however. If I don't answer my email I don't have to accept patches. If I don't read my email I don't have to answer it.