My random musings from three years in CS grad school at UNC:
Of all the graduate students I met, you could more or less divide them into two classes. The first class wanted the PhD as a means to an end, such as doing research, or getting a professorship. The second class had more abstract and less well-defined goals. Sometimes that meant generally "doing good for the world", or "seemed like the right thing to do", or "the working world sucks, this is the lesser of two evils".
I fell into the second class.
I spent three years doing computer science grad school at UNC. I passed my dissertation proposal and then decided to leave.
I don't resent those three years and did learn a few things along the way.
Getting a PhD is ultimately about getting 5 other PhD's to sign off on your dissertation. At least that's the requirement in computer science at UNC. I wish that had been more clear to me at the beginning.
Graduate school as a means of stalling to find something you love to work on is not a horrible way to go. In the meantime you might solve a few problems. I also consider it noble that we (graduate students) as a whole are willing to do the nation's research for such little pay.
Had I thought long and hard about why I was doing the PhD to being with, I may not have started. But life always charges its own tuition.
From the article "As long as you have an answer that you believe in passionately, then that's enough." My three years in graduate school taught me that the most. Find something you love, or least have passion and do that. It may be a silly belief, but it's worth believing in.
The first class wanted the PhD as a means to an end, such as doing research, or getting a professorship. The second class had more abstract and less well-defined goals. Sometimes that meant generally "doing good for the world", or "seemed like the right thing to do", or "the working world sucks, this is the lesser of two evils".
I agree with your subdivision, but I'd add another class: people who want to understand/solve/think-about some hard problem they find genuinly interesting and who don't particularly care about career as an academic, or the good of the world or anything much else, they just want to think about this topic because they enjoy it and grad school gives them a relatively stable environment where they can do it.
> Graduate school as a means of stalling to find something you love to work on is not a horrible way to go.
I agree with this (despite being someone who is finishing soon, god willing), as long as you can have the mental toughness to not buy into the "man, failing out of grad school, what a shame" mentality that many people will direct your way. It's not that people are all jerks (though some are), but some people are just completionists (never start anything you won't finish), and other people are there because academia really is the best home for them, and they genuinely feel that it would be a waste for anyone smart and good at research not to finish and go into academia.
If you can get over that, though, a few years of grad school without finishing is not a bad deal. If you're in a PhD program (rather than masters) you generally get free tuition, plus a modest stipend that should cover at least rent+groceries, so you end up basically getting to study something for free for a few years. You could make a lot more doing something else, but if your goal is "learn some stuff about [subject]", it's just about the cheapest way you'll find to be able to do that full-time.
> Of all the graduate students I met, you could more or less divide them into two classes
Agree. I almost went for it as well, though for a challenge and as a competition - publish the best paper in the best conference. Seemed like not too many people viewed it as that.
Usually, this is very hard to achieve. Not only is it hard to find the best conference in your field, getting the best paper award (nb. that there are conferences that don't have a one!) has (at times) something to do with sense of fashion and/or the valuation of your achievements by the programme committe. (I recently stumbled upon an article [unfortunately I cannot remember when and where, 20min of Googling did not help either] that detailed how Albert Einstein did not believe in one of his own equations, which was in the larger context that is difficult for scientists to properly judge their own contributions -- assuming this is correct, having the intention of writing "best paper award" papers seems very difficult...)
I've interviewed a fair number of people that claim that they completed 90% of their PhD, but didn't quite get to the finishing line. While I sympathize, they haven't really grasped that the last 10% of the distance takes at least 50% of the effort.
From the outside, a PhD sounds like it's all about 'smarts'. Having been through the process I'd estimate the requirements of smarts:tenacity = 50:50. (YMMV)
It depends what you're looking for, though. The last 10% is heavily on the writing up and justifying independent research part, with some academic politics thrown in (convincing five professors to sign off on your thesis is not as objective a process as one might think). If what you want is someone really good at applying research, perhaps with clever tweaks, that part might not be as important. That's one reason places like Palantir explicitly recruit for "all-but-dissertation" PhD students. They need the kind of knowledge that someone will get from being in the research world for 3-5 years, but their needs aren't quite the same as someone hiring for a professorship.
"The last 10% is heavily on the writing up and justifying independent research part, with some academic politics thrown in....places like Palantir explicitly recruit for "all-but-dissertation" PhD students. They need the kind of knowledge that someone will get from being in the research world for 3-5 years"
That's called a Master's degree. You take some coursework, learn how to read papers and implement the ideas therein. It's a fine degree, but it's not even remotely the important part of a PhD.
People who actually go through the "writing up and justifying independent research part" of a PhD tend to find that the writing and justifying are the hardest part of the technical process. It's easy to brush it off as minor, but only once you've begun writing and organizing do you realize the places where your work falls short. The writing takes a long time not because it's especially hard to write a dissertation, but because you usually have to go back and deal with a lot of unanticipated technical problems before you can say the work is done.
Well, I'm writing up currently, and I guess I don't consider it the most important part, really. I'm finishing it more because I want the Ph.D., not because I'm learning anything at this point.
Perhaps you could get the deep research engagement via an unusually research-focused master's, but there usually isn't scope for that. An M.S. is normally only two years, while 3-5 years is imo the amount of engagement you need with a field to really understand what's going on in research. You need to attend the major conferences in a field more than once, ideally present papers at several of them, present a few more papers after the initial time you do (your first presentation is never the greatest engagement with a research community), and so on. You can get to that as an ABD PhD student, but rarely as a master's student. An M.S. student publishes between 0 and 1 papers typically, and maybe attends 1 or 2 conferences, while a 5th-year PhD student, in CS at least, will more typically have published at least 3 papers in major venues (often 5+) and attended 5+ conferences.
That's all still writing-and-justifying independent research (getting a few conference publications and journal articles), but not quite the same as writing the equivalent of a book, which is a fairly different skill, and one you're not actually likely to use again after getting a Ph.D., since most research is done in conference and journal papers. I wouldn't say it's a complete waste of my time to be writing a thesis, but I wouldn't say it's the most important thing I've done in grad school either, by a long shot--- I value my peer-reviewed publications in the scientific literature much more than I value my institution's thesis/defense process.
I don't want to appear too critical, but you've just outlined the reasons not to persevere with the write-up. The point is that you are right in every respect : except for the part that doing the last 10% is only a modest amount of work.
With luck, you've got enough momentum to push through to the end smoothly. Otherwise, you might find yourself within 'just a short distance to the summit' and being someone that climbed 90% of the way up Everest.
Got a job at the university, talked to professors about possible topics but ultimately decided to get a job in the industry.
While teaching classes was fun and the job gave me plenty of free time to read things.
What was really annoying me is the fact that for every little thing you want to do (e.g. buy another smartphone for the study, go to a conference, ...) you have to run through a horribly inefficient process of filling out a dozen forms and getting signatures from all sorts of people.
I also felt bad about not really "creating" something. In my master's thesis, I created an asynchronous web-crawler, evaluated lots of NoSQL backends, hacked for about 8 hours a day and had a nice project in the end.
At the university, I created slides, did lectures but never really did something "cool". There are already dozens of lectures on all sorts of topics and I pretty much only transformed text into pretty slides.
Starting next month, I'll be working at a small company (9 people) and really look forward to hack on something with likeminded people :)
Holy shit, this article could read like all the characteristics that are needed to be a good leader and co-worker. Thanks for this. It made the morning cup of tea taste that much better.
I always find an interesting contrast with articles like this on getting a CS Ph.D., and my experience in engineering.
I got a Ph.D. in Chemical Engineering at Carnegie Mellon. It was a really interesting four years (and three months) - I loved it. I actually brought my thesis topic with me to grad school from a summer job as an undergrad. I wrote three good papers, spent six weeks turning those papers into a thesis (mostly a matter of doing a nice literature review in the introduction), and I was done. I had a excellent advisor, and we had a great relationship. I was part of a solid research group, and we helped each other. The process seemed very straightforward and linear. Everyone in my department seemed to be roughly the same. We all worked hard, did research and got our degree in 4-5 years. Very few of use would have felt on our own.
I get the feeling that this happy experience is less common in CS. The relationship between grad student and advisor seems more distant, and less helpful. The happy hours we had with the SCS at CMU reinforced that. They had "black fridays" where projects were terminated and grad students left. That "how will I get my committee to sign off" thing was there, unlike in our department.
Is it something about the nature of "research" in CS that leads to this, or is it a cultural thing?
I really appreciate guides like these, particularly because I'm preparing for graduate school myself.
Another really helpful read was "Getting What You Came For" by Robert Peters. He walks through the entire Ph.D. process and describes a pipeline that can knock a 7 year program to 4 or 5 and a 5 year program to 3 or 4. That time saved would be incredibly valuable.
I'm always on the hunt for good material like this. What other Ph.D./grad school/research guides have people found helpful?
I found this to be a good overview of the Ph.D. system from a CS perspective. I'm doing grad school in a slightly different field, but a lot of the advice was applicable.
Yet summaries without that term often sky-rocket: http://news.ycombinator.com/item?id=2002255 And more than one HNer publicly stated that they read comments first, to see if the article is worth their time.
This only make sense if "tl;dr" is considered as a swear word here at HN. Is it?
Of all the graduate students I met, you could more or less divide them into two classes. The first class wanted the PhD as a means to an end, such as doing research, or getting a professorship. The second class had more abstract and less well-defined goals. Sometimes that meant generally "doing good for the world", or "seemed like the right thing to do", or "the working world sucks, this is the lesser of two evils".
I fell into the second class.
I spent three years doing computer science grad school at UNC. I passed my dissertation proposal and then decided to leave.
I don't resent those three years and did learn a few things along the way.
Getting a PhD is ultimately about getting 5 other PhD's to sign off on your dissertation. At least that's the requirement in computer science at UNC. I wish that had been more clear to me at the beginning.
Graduate school as a means of stalling to find something you love to work on is not a horrible way to go. In the meantime you might solve a few problems. I also consider it noble that we (graduate students) as a whole are willing to do the nation's research for such little pay.
Had I thought long and hard about why I was doing the PhD to being with, I may not have started. But life always charges its own tuition.
From the article "As long as you have an answer that you believe in passionately, then that's enough." My three years in graduate school taught me that the most. Find something you love, or least have passion and do that. It may be a silly belief, but it's worth believing in.