[twblalock]

6 years of a grad student salary is a big financial sacrifice compared to 6 years of a software industry salary, even at non-Silicon-Valley salary levels.

The opportunity cost of a PhD can be pretty high. In CS you could theoretically make back a lot of that money with a higher salary after graduation if you then enter industry, but the gap in lifetime earnings up until that point is huge.

[fooker]

Eh, not everything has to measured by how much money you accumulated.

[twblalock]

That's not the point.

Let's say you spend 6 years in grad school on a stipend of $35k per year. You think you are being paid to get a PhD, but that's just because you are ignoring opportunity cost.

If you took your talents to industry instead, let's estimate an entry-level salary. I will intentionally make it lower than average. Let's say $100k per year, which is on the low end in Silicon Valley for an entry-level engineer, and let's also make the ridiculous assumption that such an engineer would not get a raise for the entire 6-year period.

After 6 years, the hypothetical grad student was paid $210k total, and the hypothetical entry-level engineer was paid $600k total. The PhD cost $600k - $210k = $390k. Would you pay $390k for a PhD?

[fooker]

>Would you pay $390k for a PhD?

Yes, of course, even if you increased it to $800k.

Several thousand brilliant people make this decision every year. Think of it as paying $390k to buy 5-6 years of time, and it starts to sound very different.

5-6 years of time at the prime of your life where your basic needs are taken care of without any enforceable expectations from you.

You can go sit in any of the hundreds of interesting things being taught around you. You can pursue hobbies, go camping in the middle of the week.

If you are lucky, you get to work on interesting bleeding edge projects which may or may not make business sense. For this you get nearly any resource you ask for. Labs in good universities are rather well funded. Supercomputer time to run a random program, easy! Also, most companies are happy to hand out huge amounts of cloud credit, reference hardware.

Compared to making 390k, are you surprised that at least some people would make this decision? Also, 35K per year + well paid internships (~40$k for a summer) is a rather comfortable amount of money for a 25 year old person in most situations.

[2sk21]

And you get a high salary, post PhD only if you are in some hot field. I got one the earliest PhD's in neural networks - but this was in 1992, long before the world discovered machine learning. It took a while for my salary to catch up, although I can't complain, the last decade has been good.