Don't become a scientist (1999)
physics.wustl.edu
physics.wustl.edu
I feel very lucky to be a well-paid staff researcher at such a young age, especially since I don't have a graduate degree. I'm surrounded by older people with PhDs; most of the time, I feel like there's really nothing that really differentiates us.
However, and this might be my inferiority complex kicking in, sometimes I feel like I should go back to school and get my PhD. But then I read essays and articles like these, and I wonder if a choice like that would be stupid. Granted, I'd pursue some sort of computational physics in grad school, so I'd still leave with marketable skills.
Anyway, I don't know what to do. Just thinking out loud. If anyone has any advice, I'd appreciate it.
You just have to ask yourself if the "prestige" is worth it to you. You'd be taking a salary hit for 3-5 years for the degree and 3-5 years after that for postdoc, and then after that, you'd probably not be able to find a job. From what I've heard, physics is a hard field to find work in, and you'd be over-qualified to do the job you currently have.
It's a tough call. Going back, I probably would have still gone the PhD route, but I'd have been much more selective in projects, etc... and family is a big factor.
On the other hand, I've known good staff researchers who've been told that while they'd be perfectly capable of getting a PhD, they wouldn't be able to get into the grad school where they worked.
Unless there was some kind of already established program, I couldn't seen how you'd be able to effectively do it part-time. Maybe you could get a Masters part-time, and then move to a PhD program, but it would take longer.
I've seen it done, though at a university, and in CS rather than physics. (And in the UK.)
Was it a tough call when you decided to go to graduate school? It seems like it's always such a huge decision, no matter your circumstances. Also, how would you have been more selective in projects? Did you get stuck on a particular project? Any tips on how to maneuver your way toward the good projects?
1. You are so in love with the subject matter, that you can't think of doing anything else. I'd give the same advice to someone thinking of going to music school (my Plan B when I was a kid).
2. Somebody else is paying for it. My spouse and I finished grad school with no debt.
Without wanting to be too negative about it, I think that the inferiority complex thing, and expecting to receive marketable skills, aren't good enough reasons. Are computational physicists really more employable than common, or garden, computer programmers? A lot of programming can and is done by trial and error or looking up stuff online, and a lot of people doing it are basically self taught. (I'm looking in the mirror when I say those things).
I went to grad school because my parents had science degrees, and seemed to have good careers and happy lives. And I was interested in learning more, in an academic environment. I already knew how to program, but saw it as a drudge job -- easy but ultimately boring if you were doing it for someone else. (My summer internships at a data processing center may have biased my perspective). During the time that I was in school, the people who seemed to be doing the coolest things with computers were the physicists.
Note that I have to be careful what I say about programming, in a nest of programmers. ;-) I really have a lot of respect for what folks are doing, but am describing the limited environment that I was exposed to 20+ years ago.
Hmm, that sounds like grad school, to be honest. The difference is it is _hard_ and ultimately boring because you're doing it for someone else.
>During the time that I was in school, the people who seemed to be doing the coolest things with computers were the physicists.
Hmmm. My experience with "computational physics" is
1) Run the simulation, probably in 2D a) because the codes suck and are written by subpar programmers b) your analysis scripts are dumb and only work if you output everything, so your simulation spends most of its time in IO
2) Plot the outputs in matlab. Bullshit your way through understanding as if you understand it, write papers,make presentations, enjoy your pizza parties because that's all you have in your life.
May be I just have a bad choice of groups to choose from. This is my opinion, though, so take it with a grain of salt.
I hear statement (1) quite a lot, but it seems like a really subjective criterion. When I hear this, I think, "If I'm not as passionate as Feynman, I should not go to grad school." And so I haven't. Am I looking at this the wrong way? When I compare the subjects I enjoy and the careers that allow freedom and decent pay, physics seems to be the only thing that remains. Is that not enough? I don't have the confidence to know whether I'll love the subject in ten years. I also think anyone who has that confidence is either a Feynman-type or has deluded themselves. Regarding (2), don't most hard science grad programs pay the students?
I'm also not saying that computational physicists are more employable than common programmers (or that they're better software developers!). I'm just saying I enjoy working on problems similar to those encountered in computational physics. I enjoy figuring out how things work and modelling them. In case I ended up hating physics by the end of graduate school, at least I'd have a strong background in mathematics and programming, and those skills could be transferred to domains with similar problems.
If you don't mind me asking, are you still working in physics? Based on your "about" section, you seem to be interested in a great many things, like myself. Before entering graduate school, could you truly not think of doing anything else, i.e., did you truly obey criterion (1)? I can think of many things I'd still like to do while I'm young: start a business, write songs professionally, act... but it's hard to take risks like that when you've got a good (and more practical) thing going.
(2), yes, but it's just a reminder to set a bar for yourself, that you have to get into a decent program with decent support in order for it to be worthwhile. It might not be so much the principle of getting paid to work, but having enough of a financial cushion so you don't have to deal with crises by quitting grad school.
I obeyed criterion (1) but not Feynman level. That's too strong of a criterion, like you suggest. Rather, the alternatives to grad school all looked pretty blah. I've been _extremely_ lucky. My thesis project was very much a hackathon in optics, electronics, software, etc. Today I work for a company that makes scientific instruments, and I work on assimilating new technologies on a 1 - 2 year time frame before we attempt to develop them into the next generation of products. So I do a lot of experimenting, learning theory, etc. I solve difficult problems that crop up with our products, in production and in the field.
Truth be told, I also spend a reasonable amount of time on mainstream work, e.g., project engineering. I've tried my hand at both project management and people management. Those were OK. I don't have enough mental organization to do the project management thing when I'm actually competing with other project managers for resources.
Over the duration of grad school, I kinda lost interest in basic research. It's cool, but I'm not a _great_ physicist, and I've simply never had an idea for my own line of research, which would kinda put a damper on an academic career. I prefer to work on shorter term projects where I can add value by developing an understanding of how things work. I also feel that the industry I'm in genuinely creates good solid jobs for people. I don't face the dilemma of wondering if I have to exploit people in order for my own work to progress.
There may be a cautionary tale in there for computer science: once the VC carousel stops the situation in computer science will be quite as bad.
That's why I said the unemployment crisis isn't caused by an oversupply of scientists - up to about 2004 most everyone with a PhD degree could find employment in industry. But this fellow Katz doesn't even mention industry. He is either ignorant or intentionally dishonest. There is no excuse for either.
Other professors and PhDs I've talked to have basically said the same things as he says here. There are definitely jobs for PhDs in CS in industry, but whether you would be considered a "scientist" for those jobs is debatable, as most of the time you don't have the same level of freedom you would have in academia (or at least that's what I've been told).
I graduated with an B.S. in Physics about the time that he wrote, and his analysis pretty much matched what I saw in the lab I worked in. Of my physics classmates who I know what happened to them, all got jobs doing no physics (including the newly minted Ph.D. who was supervising me). I talked with the postdoc in our lab (her second postdoc position), looked at the number of Ph.D.s and postdocs at Ohio State (lots), looked at the number of open positions in the Physics Department at Ohio State (none), and decided I did not want to postdoc until a professor died and then hope that I get picked. So I said physics was a fun, stimulating, foundational undergraduate choice, felt like I understood the entire world, and promptly pursued software. (Fortunately, I was aided by being tired of school and not really enjoying physics research)
One of my physics classmates did software for a while, then decided to do a Ph.D. in (rocket) propulsion, which is physics-related. "Industry" for him is pretty much limited to NASA (who else is going to employ someone looking into new forms of propulsion?) Turns out he doesn't want to work for NASA bureaucracy, so he switched back to software again.
I think the author is right on, at least for his field of physics.
The difference is that CS is something that every company increasingly needs, not just tech companies. Most companies have never needed and will never need science.
However, there just aren't the jobs available for the number of people that are being trained. I'm somewhat lucky that I was a programmer first, molecular biologist second, so computational work was a no-brainer. However, it is still extremely difficult to find jobs past the post-doc stage, and that stage does last about 5 years longer than it used to. And post-docs get paid like crap for each of those years.
There are lots of career options available, but really, academia is still the place where science gets done. I know lots of people work in industry, startups, biotechs, pharma, etc. But unless you're developing a drug, you're probably trying to sell to academia. There are some great projects and good work being done, but with the lull in Federal funding, the long-tail of grant funding isn't there to support existing labs, let alone new ones.
Debating the reasons for all of this could go on for hours... and there are lots of reasons and enough blame to go around.
It's a great time for science, but a bad time to be a scientist. It was never lucrative, but it just isn't a viable profession for many.
Dropping out of academia was amongst the more difficult decisions I've made. Also amongst the best.
There are huge opportunities for scientists today. I'm a physicist. I've worked in computer-assisted surgery, medical physics, image processing, computer vision, genomics, computational physics and a few other things besides. It has paid well and let me do all kinds of interesting research that I would never have been able to touch as an academic.
Things to do as a student: study modern statistics hard. Being a card-carrying Bayesian is like the secret handshake to an unlimited world of really cool stuff these days. There's a childhood friend of mine who's (finally) finishing his PhD in psychology, and when we reconnected a few years ago we were amazed that we could meaningfully talk to each other about our work because we both spoke the same statistical language. It really is a universal grammar, and anyone in any field who speaks it fluently has a world of opportunity waiting for them, from finance to fisheries.
This is something I've been thinking about doing. My work is in computational materials science, and I often feel there is a complete lack of statistical rigor in analyzing the results. (Peter Gill joked at a conference that the only difference between applied mathematicians and quantum chemists is that the mathematicians calculate the error once they get their answers).
I know basic statistics: distributions, standard error, correlations, etc. But I don't know anything about the terms Bayesian or Frequentist. In your opinion, what would studying these topics provide for me?
Would you recommend some texts that cover this?
A friend suggested:
* Data Analysis: A Bayesian Tutorial by Devinderjit Sivia, John Skilling
* Bayesian Data Analysis, by Andrew Gelman, John B. Carlin, Hal S. Stern
* Introduction to Statistical Inference by Jack C. Kiefer, Gary Lorden
Currently I am working my way through this.
Pattern Recognition and Machine Learning by Christopher M. Bishop
These books are expensive and very time-consuming so it is important to pick the right ones.
or, increase funding for science (academic and otherwise)