Hacker News Reader

Comment by klochner | Hacker News Reader

A computer science Ph.D. is nothing like what is described in this article. For various reasons (other opportunities, generous industry grants, minimal financial overhead to run a lab), students in computer science are comparatively well paid and free to work at their own pace. YMMV based on advisor.

It's been exhaustively shown that stopping after undergrad or masters degrees will be more financially lucrative, so some reasons to go on:

  - you want one of the specialized researchy positions that
      are typically filled by Ph.D. holders
  - you have a genuine interest in some specialized field and
      want to work purely on learning/investigating that field
  - you want to be a professor

View on HN

Unfortunately the "generous grants" often conflict with "free to work at their own pace". It's getting more common to find arrangements where PhD students essentially have a part-time job working for their RA funding. In the best case this overlaps with their own research, but sometimes doesn't, and may involve considerable additional work that doesn't really "count" as research. Particularly the case if you're funded on a DARPA grant, since those have very specific timelines and deliverables, plus days where you have to be ready for a "site visit" to demo things. Some industry grants are fairly hands-on as well, expecting the lab to basically lend out a PhD student to work part-time on a project the company wants completed. Though others are more hands-off and just donate a pile of money without demanding specific things in return.

I would limit that to just #2 in your list. The other two carrots are so out of your reach or subject to factors so far beyond your control that a love for learning and a burning interest in your research is the only thing that you can count on sustaining you.

#1 is also pretty achievable. For example, if you get a Ph.D. studying how to implement compilers and virtual machines, you basically have your choice of which place that implements compiler or VMs to work at (source: I am about to defend my Ph.D. in implementing compilers and VMs).

But how many places are there that do that? You have google, Microsoft, maybe intel, a few companies in the valley. PhDs are very specialized, and so they really can be at the mercy of their hot skill.

And even google...might tell you to go work on ads with your hot VM/compiler skills. Disclosure: I'm a PL PhD who got out of the vm/compiler area to focus on PL design, which is even more esoteric.

I know of quite a few data analytics companies that hire mainly PhD's for their research positions. Many traditional companies do so too, to get on the big data trend.

Yes, but they are hiring PhDs in systems, PL (for some reason, we are good at MapReduce), database, and some data mining fields. Its not like you want to hire an HCI researcher to do big data.

I don't know about google - all the PL people I know who went to google no longer work on PL stuff, though I don't know if that's because of working at offices where PL work is not supported rather than a lack of company-wide freedom.

From conversations with hiring managers at each, Apple, Microsoft, Intel, NVidia, and Adobe all have far more positions than they can fill for people with PhD-level VM/compiler writing skills. Heck, we can't turn out experienced _undergrads_ fast enough for their tastes. I assume there would also be stuff in embedded systems, particularly FPGA compiler-related, though I don't personally know hiring managers at the big firms in that area.

Google has found PL PhDs very versatile I guess. Many of their top people are in PL, so it could also just be a top down hiring bias.

I'd be weary taking just any job in VM/compilers, much of it is quite tedious and boring for a PhD. Intel used to have compiler groups filled with PhDs, Sun also. Them there are smaller companies like coverity.

Reply on news.ycombinator.com