Easy Ways to Fail a Ph.D.
I'd add a huge one to the top of the list:
1) Pick the wrong mentor.
Quite simply, if a mentor hasn't graduated a student in the last three years, serious red flags should go up. Professors can get a ton of work done relying on research assistants and post-docs. But students need to be nurtured and many mentors just aren't up to the task. Their track record at graduating students is the best evidence you can ask for. Look for potential role models in that history.
If the mentor is a nontenured professor, ask for their advice but don't make them your lead mentor to sign off on your work. Tenure has many stresses and some are orthogonal to your interests as a student and future researcher. The nontenured need to be focused on themselves until they get tenure. They can have a lot of energy and novel thoughts, but since your results will be tied to their future prospects, your work will get undue pressure.
The edge case is the newly crowned associate professor. Fully interview the prospective mentor and get references. Your career will literally be in this person's hands. Make sure you know that this is a person who will challenge and inspire. Associate professors can be really, really great mentors, but you just have to be careful if they haven't graduated any students. If they have, and those students are already on tenure tracks, you likely have an excellent candidate.
My personal experience* is that some of that failure rate can attributed to students not understanding what a PhD is for. It is a qualification to conduct research... it is not about creating something cool/interesting and/or useful (particularly not the last one). It's about the journey not the product.
Unless you intend to stay in academia, the qualification probably won't pay off in terms of a higher salary vs. the years you wasted. And if you do intend to be in academia, be prepared for a long (possibly indefinite) wait before you get a permanent lecturing/research position. Post-doc life is not easy.
* 3 years wasted working on a PhD, no thesis submitted in the end, walked away back to the real world.
Man, as a grad student, I can't disagree with the first two points enough.
This is OK advice if all you want to do is finish a PhD as quickly as possible (ignore course work and any studying not directly related to your dissertation), but it's terrible advice if you want to (a) become a broadly smart individual or (b) do research later on anything else besides the tiny topic of your dissertation. Grad school is the only time in your life where you'll have the freedom to study all sorts of incredible stuff. If you leave academics, you won't have access to the resources, and if you stay in academics, you'll be constantly scrambling to crank out publications as a post-doc or professors.
> In the interest of personal disclosure, I suffered from the "want to learn everything" bug when I got to Ph.D. school. I took classes all over campus for my first two years: Arabic, linguistics, economics, physics, math and even philosophy. In computer science, I took lots of classes in areas that had nothing to do with my research. The price of all this "enlightenment" was an extra year on my Ph.D.
Only a year? Sounds cheap to me! When else in your life are you going to have the opportunity for this kind of enrichment?
Yes, taking linguistics classes is probably not going to help your future physics research. (Though it can still be worthwhile). But taking chemistry, math, and physics outside of your particular niche is incredibly helpful for teaching you different ideas that you can apply later.
Also:
>Einstein's Ph.D. dissertation was a principled calculation meant to estimate Avogadro's number. He got it wrong. By a factor of 3.
Yea, well, for their PhDs, Bekenstein's discovered black hole thermodynamics, Feynman came up with the path integral, and Hawking proved the singularity theorem. Yes, you can go on to be successful after a modest thesis. But you can also kick ass young. (Also, finding a method to calculate a theretofore empirically-measured universal constant based on elementary principles--and only being wrong by a factor of 3--is pretty incredible!)
Frankly, I just don't know very many grad students who come close to "shooting too high". I'm sure this professor has met a few, but is his concern really that they are shooting themselves in the foot, or is it that grad students who take chances for big breakthroughs (and hence, usually fail) are not very useful for advancing the career of professors?
People have touched on this, including the parent article, but this needs stating in BIG SHOUTY CAPITALS:
PhDs are the last mediaeval guild apprenticeship standing, with all the ceremony and all the anachronism that implies.
Seriously. You're apprenticed to a master, you have to produce a meisterwerk, then they let you in. The entire thing is a historical accident. Remember that and everything else is... well, it's still crazy, but you can get the thread of it.
(yrs, PhD (Cantab) 2007 - http://www.lexical.org.uk/science/thesis/)
Matt, I understand that you are writing primarily to your graduate students, but I also have to object to your first two points in general. I am a counterexample, as I took 12+ unnecessary math classes (getting A's in them) on my way to a successful PhD in theoretical computer science/crypto. I've found that my background in math has given me some tools and perspective that improve my research (I recognize when we are constructing a finite field in a round-about way, or when the abstraction we really want is a group action, or when Borel-Cantelli is really needed to make an argument rigorous, etc etc). I wouldn't have obtained these tools without a huge investment in time at some point.
Actually, I still find myself reading and learning unnecessary math papers most evenings as a hobby. I think it makes a huge difference.
Perhaps this is unique to mathematics, where theorem proving skills are portable. I certainly wouldn't use my situation to argue that PL grad students need to study French. But maybe the PL students should take an extra distributed systems course - and TA while they do it!
An American PhD sounds tougher than here in the UK, or at the very least significantly longer. Here, you get funding for three (or sometimes four) years and I don't know of that many people who have left their programmes.
Why is there such a big difference, and is there any data about how this this reflected in later careers?
I would guess that the advice of "only do a PhD if you want to be an academic" doesn't hold as strongly over here. Does it?
(Disclaimer: I say this as someone currently debating whether or not to do a PhD, but who doesn't want to be an academic.)
Quote of the article: "I fantasize about buying an industrial-grade stapler capable of punching through three journal papers and calling it The Dissertator."
>At best a handful of chemists remember what Einstein's Ph.D. was in.
>Einstein's Ph.D. dissertation was a principled calculation meant to estimate Avogadro's number. He got it wrong. By a factor of 3.
This is interesting. A quick search can't show me the history of Avogadro's number -- what was it thought to be before? Einstein could have been off by a factor of three, or he could have been off by "only" a factor of three. Context matters.
I thought about getting a PhD at one point. It seemed like a neat thing to do. Talking to a friend of mine who is a professor disabused me of that idea quickly. Too much obedience, maybe, but the real killer is the politics. Lots of politics and politicking and stroking ego to get a PhD. No good. That's not what it should be.
The only thing I don't agree with is "Aim too high."
I can't imagine surviving at a top-ten school without having a focused research vision from the start. I'm not saying you should walk in hoping to prove P != NP, but who wants to dedicate 5+ years of their life to working on someone else's research project? Not to mention that these are typically the best years of your life.
I think the real key is to aim for a the top of a hill, not a cliff. You need to pick a goal that enables you to make gradual progress. If you start out by saying "I plan to build the best race car in the world"-- that's fine. Your first paper may be on improving the tires, then a follow-up with an even better tire material, then you discover that square tires can be used instead of round ones (thereby disrupting the entire field of tireology), and before you know it you have a PhD. You never reached the top of the hill like you expected, but you kept climbing.
Disclaimer: I don't have a PhD (yet). :)
Man, this makes an Australian PhD seem like a piece of cake comparatively.
1. No need to worry about funding: if you get an APA or an APA/I. If you don't get one of those, you'll be screwed. But they're not impossible to get either, if you're an Australian citizen. It's not $$$ but it's just enough to live on, if you are good at budgeting. If nothing else, at least you won't starve to death.
2. No courses. Just research your thesis topic for 3 years.
3. No teaching necessary. You can do it a bit on the side if you don't have an APA or the APA isn't enough. But you're limited to 8hrs/week of paid outside work if you're getting an APA anyway.
4. No defence. I guess it's too expensive to fly specialists to Australia, and those who set up the system did it in a time when video conferences didn't exist.
5. If you don't pass the first time you submit, you can just keep resubmitting until you pass. Of course, if you submit when your supervisor says it's not a good idea, that's probably going to be an unconditional fail. But you'd not do something silly like that, well, we'd hope not. The only other way to fail is to drop out. And there's a bunch of reasons to do that! The biggest ones are not being able to get a good research topic, or to not get a good supervisor.
Maybe I am just bitter because I feel like my PhD was a waste of time (and I did finish), but I feel like doing a PhD is a failure in itself.
On HN a lot of people seem interested in Doctoral study and seem to praise it as a good thing, but I am wondering, are there many people out there who found their years of a PhD helped them much on their start-up?
I feel my years of PhD study and the degree have not helped much.
I feel left out:
#11. Start a startup while doing the PhD.
This may be good advice for getting a doctorate in ?computer science? -- the author never states his discipline in the article.
It may not be particularly good advice for other disciplines. For example, as a doctoral student in Ecology you would be very lucky to have anything submitted for publication before you take your orals. It takes too long to get set up and start your research, especially if there is a seasonal component, and there are too many dead ends along the way. If you work hard, and with a little luck you may have one or two publishable papers by the time you graduate. Aim for the stars, but the important thing is clearing the trees...
Join a financial institution and get very well paid before finishing. This is the easiest way to fail your Ph.D.
I like the article. A good comprimise between pragmatic cynicism, and good advice.
The converse:
"How to Succeed as a Graduate Student in the Sciences"
http://www.chemistry-blog.com/wp-content/uploads/2010/06/Gas...
huh. I wish my advisor had been told about this. Before I dropped out, I'm pretty sure the problems I was having in my Ph.D. were mostly because of not doing 1, 2, 3, and 10. And my prof. pretty much pushed us towards 5 (the too soon one) exclusively. Although, to be fair, I was also an expert at 4 and 8. :)
My way:
#1, pick a supervisor who's going to be absent the entire first year and who decides to change your research topic significantly from its proposal to fit in with her own research (yes, I know this is common)
then
#2, decide the real world is more exciting and start working on a startup idea. Whoops. Still, dropping out to do the startup makes me cool like Larry and Sergey, so I'm down with that.
I think Einstein's dissertation is a bad example of a dissertation that was not an immediate superstar. It was. Einstein could have taken this author's advice of stapling together his first three papers. Einstein's considerations about more precise calculations of Avogadro's Number lead directly to his novel technique(at the time!) of using Brownian motion to deduce that photons exist as particles, and not just waves. Then it only took a couple of years for that idea to morph into his famous Annus Miribilis paper On the electrodynamics of moving bodies, from which Special Relativity sprung forth. And the rest was history.
Professor Might's advice is on the whole good.
I would add:
Academic Jobs. Professor Might suggested that a Ph.D. is good only for an academic career, and I would in part disagree: I got a Ph.D. in engineering where it was solidly in my mind before, during, and since that never did I want an academic career. Actually I did take an academic job for a while as a way to have time better to care for my wife in a long illness, but I regarded the job as a waste of time for all concerned.
The best of my Ph.D. coursework was terrific stuff. And the Ph.D. did confirm to me that I knew how to do research. To me, both of these are the two main pillars of my current attempts to start a successful business.
Long one of the best approaches to progress is to do field crossing: For a career in computer science, either in practice or in research, I would suggest (1) avoiding taking any courses at all in computer science unless just want to waste some time and (2) taking all the best courses could find in the mathematical sciences.
In particular, my view is that now, for the future of computing, computer science has a fatal disease and is nearly dead -- the field is missing any powerful intellectual methodology. The problems in computing remain important, but by a very wide margin the most powerful tools for progress in those problems are just the mathematical sciences.
For how to get a Ph.D., I would recommend: Start with what is fairly clearly an important, apparently not well solved, real problem from outside academics. Then attack the problem with some new work in the form of theorems and proofs with prerequisites in the mathematical sciences.
The usual criteria for publication are "new, correct, and significant": Okay, given where you got the problem and that you did some original work just for that problem, your work is likely "new". For "correct", it is fairly easy to know that theorems and proofs are correct, and it is difficult to argue with them. For "significant", since the real problem was, likely so is your solution.
I brought my own research problem to graduate school. The best coursework in my first year was a BIG help. I did all the actual research independently in my first summer. All my advisors ever did was approve my final work. I recommend this approach.
If a student has done some good research and still has a problem with his advisors, then I'd recommend just publishing the work. Nearly no one in academics wants to argue with the significance of published paper.
I do recommend doing some publishable research while a graduate student. Once in a course I saw a problem that should have been solved but was not solved in the literature. I took out a week, found a crude solution, got the problem approved for a reading course, in the next week found a better solution, and wrote up the work. It was clear that the work was publishable, and, thus, much better than needed for a reading course, and later I did publish it. That work gave me good research credibility and helped me get the rest of my way through graduate school.
For getting a paper published, it can also help if the paper has more prerequisites in the mathematical sciences than any of the reviewers have; this situation can be relatively easy for someone bringing to computer science original work based on the mathematical sciences.
Can this approach to research work? Here's my evidence: I've published several papers in computer science jointly with others, and I've published two papers on my own. One of these two, and the best paper of the lot, really is in computer science. I've never had a paper rejected, and I've never had to make any significant revisions. The two papers where I was the sole author were in relatively good journals.
I was encouraged to publish my dissertation but wanted to sell it and refused to publish it! Again, I've never had any interest in an academic career.
I would take issue with the path suggested for research that a student should start with advanced courses in computer science, read 50-150 papers, and then do some research. That approach is too narrow -- it's nose to the grindstone, shoulder to the wheel, and ear to the ground and then trying to do good work in that position. Moreover, in a plowing analogy, will likely break a plow just where the last 50 people did.
For big success in academics, need to do some broadly powerful work. For that, I would suggest picking a direction with a much wider field of view, also starting with a field of computing important outside academics.