On the subject of Phd, does anyone else think the system we have right now of BA in 3-4 yrs, masters 2 yrs, phd 3-7 yrs, is a little suboptimal for people who want to just do research, and even for those who would rather teach?
The incentives are screwed up, so few papers and studies are actually carried out in other labs, the rampant publish or die mentality that breeds mediocrity. I don't know maybe its just me, but I wish the system was a lot more open to alternative pathways which employers and govt were willing to explore.
I often have the feeling we are doing a lot of resource misallocation and deoptimization, but this is a topic I haven't looked at any data on. Has anyone?
most educated I might agree, but best at teaching the material is a different matter and is something I would really like to see some data on. Experience makes me think, that the biggest factor is probably having the most industry/applied experience. The next factor that makes a good teacher is probably passion for the subject and passion for getting others interested in it. This is all gut feeling baloney of course, which is why I asked if anyone has real metrics on some of this stuff.
> does anyone else think the system we have right now of BA in 3-4 yrs, masters 2 yrs, phd 3-7 yrs, is a little suboptimal for people who want to just do research
The US is the extreme in this regard. In the UK it's Bachelor's in 3 years, and then PhD in another 3 years, for most people. So six years in total. Some take another year to finish the PhD. I think many countries are similar, but I've even seen PhDs done in 2 years.
People in the US often think this isn't enough, but it's possible because we specialise much earlier. I started studying data structures and algorithms at 17 in what you would call high school, and we also were doing calculus at this stage. This was standard - not some kind of advanced placement. At 18 I was completely focused on computer science - no mandatory other courses except for maths. And at 21 I was writing papers and working towards the area I'd do my PhD research in. (I actually then took a really long break at this point, but you wouldn't have to.)
The UK is probably the shortest in that regard and its worth pointing out that they are actively trying to transition away from that model. See the push to setup centers for doctoral training.
Yes that adds a year, I think it's intended six months for research training (statistics and writing courses) and a six month window for an internship.
To be honest there is so much variability that it's hard to call the US extreme.
For example, in The Netherlands (at least at my previous university) for Computer Science, BSc is 3 years, MSc is 2 years, PhD is 4 years.
Non-research programs ("HBO") are 4 years for bachelors, 1 year for masters.
In Japan (at least at my university) for Computer Science, Bachelor of Engineering is 4 years, Master of Engineering is 2 years, PhD is 3 years.
This is interesting to me because despite having successfully completed my PhD, I would have had no luck if I had to specialize earlier in my academic career. I feel that my brain took longer on average to develop and that I was not capable of understanding advanced maths or computer science concepts until I was about 18-19 (first year of college). It was only then that I began to excel in course work and really enjoy math. By outside objective measures I should not have gone on to study science. In principal I think that this kind of early specialization, when students are capable of it, seems like a great idea! So much of what we learn in high school in the US is useless.
Many institutions offers research masters which are not terminal degrees meaning you can go on for a PhD. For example, Stanford CS has the Master of Science with Distinction in Research. Berkeley EECS doesn't and they only relatively recently offered a masters program you can apply to. Previously it was a bailout degree from the PhD program.
Most universities in Canada have the same thing. In my university you can do a masters "avec mémoire" (with thesis) which is four semesters of full time research.
Well, hopefully the folks doing a PhD are the ones that "want to just do research". I am doing my PhD now and I feel like 3 years (on top of a two-years masters) is not enough time for a quality dissertation. It is enough time for at least one good paper. But to avoid mediocrity, I think it can be useful for students to actually slow down (design thoughtful experiments, reproduce what others have done, thorough literature review. etc.) instead of rushing against the clock. Sometimes there is a negative attitude toward students who ask for program extensions. "Oh, you weren't able to finish in three (four, five) years, so you must not be a good scientist." Something along those lines. But there shouldn't be.
In the US for science the most common path is 4 year BS, 4-5 yrs PhD with no masters in between. This felt about right for me personally in terms of being prepared to do research on my own. However, I do not feel that a PhD really qualifies one for teaching at all. Most professors are not trained teachers. They are just experts in their field that primarily want to do research and teaching is a requirement of running research group. This is a really weird setup. Most professors that run a research group are okay teachers because of their passion for science, but they are not formally trained educators. They just have a lot of education.
I will say that I have had some discussions with PhD admissions faculty and they say that it is very hard to select which undergrad students to accept PhD programs. Essentially none of the metrics they have -- GPA, GRE scores, courses taken, etc -- have any predictive power for how "good" of a researcher a given student will be. Perhaps if they had to do a masters first it would help.
>>> In the US for science the most common path is 4 year BS, 4-5 yrs PhD with no masters in between.
That was the path I followed in physics, though I took a bit longer to finish. I had friends in fields (e.g., psychology) where they had to do an interim masters. In my view, the result was that the masters became a hoop to jump through, and one that saddled the literature with studies that weren't really very meaty.
Personally, I'm leery of anything that makes the degree take longer. The reason is that grad students face the constant risk of things that can force them to quit without anything to show for it. For instance your advisor can spontaneously quit, or for that matter, drop dead. Your research can get "scooped." You can run into health or family issues.
>>> Most professors are not trained teachers.
Most college teachers are not professors. A few years ago, I taught a couple of college courses for a semester. If you want things like teacher training, then you have to make college teaching remotely career-worthy.
When I did my physics PhD 5 years was the advertised time to finish, but it took most people 6 or 7 years. The faculty/administrators who worry about such things knew it was a problem and were actively working to fix it, so maybe they've made some progress? The whiz kids at certain institutions got out more quickly of course, but that was not the norm. The programs were being accused of false advertising by not "updating" their time-to-complete numbers.
Yes, but we need to keep in mind that a PhD is necessary to teach some things, just not all of them. You don't need a PhD to teach Calculus, or a writing class, or intro chemistry. But if you're teaching something sufficiently advanced, such as non-intro-level complexity theory or algorithms, you probably wouldn't be qualified to teach that subject without a PhD - not that you need the PhD per se, just that the vast majority of qualified teachers would have one anyway.
I think the solution to the other problems you mentioned exists, it's just hard. This depends heavily on the field, but at least for Math and the more theoretical disciplines within CS, you don't make progress by throwing more money or people at a problem. You make progress by having the right people working on the important problems, not by having more people working on more problems. Currently in technical fields my impression is that a PhD is an immigration tool first, a job credential second, and a way of becoming a researcher (not just do research for four years, but an actual interest in a career of research) third. This isn't a rant against immigrants at all, and I'm not blaming them for the current state of research, because for them getting a PhD is just a logical way to play the immigration/career game.
But because immigrants/people looking for credentials see the PhD as an investment in a next step in their life, they've in a sense driven the "market rate" for entry-level researchers to the ground, because they're willing to sacrifice for it. There are a lot of really smart people graduating with Bachelors degrees this year who are going to work at places like Jane Street, Facebook, Goldman Sachs, and McKinsey, and they're all going to make at least $100-200k/year starting, to $250-400k after five years. It would be almost insane for these people to pursue a life of research, because even if they would be very good at it and love the research itself, they'd be working an extremely stressful $20k/year job for ~4 years for the chance of becoming a researcher. I'd bet that a lot of the greatest technical minds of our generation, who could have been great researchers in math, CS, physics, etc., are gravitating towards these jobs because the value proposition of research simply isn't there.
Ultimately what we need to do is pay PhD students more and make it more likely you will actually be able to go into academia after completing one. In other words, we need to make getting a PhD something people do because they want to do research, not because they're trying to change fields, or needed more education to reach pay grade n with an employer, or even because they want to immigrate (again, nothing against immigration, I just don't think a PhD is an appropriate channel from a policy point of view). I think a lot of problems, including publish or perish, would go away if you didn't need to be a top 0.1% researcher to get a good position in academia after graduating.
We also need to be extremely careful to not let a PhD become simply a third tier of tertiary education. I have a lot of problems with the education field offering "PhDs" and "EdDs" which are little more than a second master's program (there's some irony here). This is pretty much exclusively caused by these degrees being arbitrary barriers of entry to promotions within the education field. I hope this doesn't spread elsewhere because it really cheapens what it means to have a PhD.
The primary goal of a PhD is to provide training for independent research, and honestly it really does require 4-5 years of dedicated work to achieve that (I have a PhD in computational biology). Teaching is different altogether and absolutely does not require a PhD. Proving that you can perform independent research really does require many years of training.
Yes, it can happen for multiple reasons. I have personally seen it fail once: a PhD candidate in physics was asked a random but basic question about quantum physics, he couldn't answer, and the committee "voted" no.
From what I have seen, it is more common for people to simply quit their PhD candidacy once they form their PhD committee, because the committee members will ask for a soul crushing extra amount of work before the defense (happened to two friends of mine).
My former adviser told me that out of the 150+ committees he's been on, only one failed, and that the person who failed known for stealing research from his labmates and demonstrated in the defense itself that he didn't know what he was doing.
As far as I'm aware, defenses at most institutions are formalities. Your adviser has already decided you're done and its just the bureaucracy. Some places don't even have defenses (I don't think Berkley does, at least the chemistry department).
27 comments
[ 3.9 ms ] story [ 63.6 ms ] threadThe incentives are screwed up, so few papers and studies are actually carried out in other labs, the rampant publish or die mentality that breeds mediocrity. I don't know maybe its just me, but I wish the system was a lot more open to alternative pathways which employers and govt were willing to explore.
I often have the feeling we are doing a lot of resource misallocation and deoptimization, but this is a topic I haven't looked at any data on. Has anyone?
The US is the extreme in this regard. In the UK it's Bachelor's in 3 years, and then PhD in another 3 years, for most people. So six years in total. Some take another year to finish the PhD. I think many countries are similar, but I've even seen PhDs done in 2 years.
People in the US often think this isn't enough, but it's possible because we specialise much earlier. I started studying data structures and algorithms at 17 in what you would call high school, and we also were doing calculus at this stage. This was standard - not some kind of advanced placement. At 18 I was completely focused on computer science - no mandatory other courses except for maths. And at 21 I was writing papers and working towards the area I'd do my PhD research in. (I actually then took a really long break at this point, but you wouldn't have to.)
For example, in The Netherlands (at least at my previous university) for Computer Science, BSc is 3 years, MSc is 2 years, PhD is 4 years. Non-research programs ("HBO") are 4 years for bachelors, 1 year for masters.
In Japan (at least at my university) for Computer Science, Bachelor of Engineering is 4 years, Master of Engineering is 2 years, PhD is 3 years.
Well, apart from Scotland where 4 years undergraduate courses are standard.
https://exploredegrees-nextyear.stanford.edu/schoolofenginee...
You graduate with an MSc or MASc, whereas for the non-research masters, you graduate with an MEng (for the engineering programs).
I will say that I have had some discussions with PhD admissions faculty and they say that it is very hard to select which undergrad students to accept PhD programs. Essentially none of the metrics they have -- GPA, GRE scores, courses taken, etc -- have any predictive power for how "good" of a researcher a given student will be. Perhaps if they had to do a masters first it would help.
That was the path I followed in physics, though I took a bit longer to finish. I had friends in fields (e.g., psychology) where they had to do an interim masters. In my view, the result was that the masters became a hoop to jump through, and one that saddled the literature with studies that weren't really very meaty.
Personally, I'm leery of anything that makes the degree take longer. The reason is that grad students face the constant risk of things that can force them to quit without anything to show for it. For instance your advisor can spontaneously quit, or for that matter, drop dead. Your research can get "scooped." You can run into health or family issues.
>>> Most professors are not trained teachers.
Most college teachers are not professors. A few years ago, I taught a couple of college courses for a semester. If you want things like teacher training, then you have to make college teaching remotely career-worthy.
I think the solution to the other problems you mentioned exists, it's just hard. This depends heavily on the field, but at least for Math and the more theoretical disciplines within CS, you don't make progress by throwing more money or people at a problem. You make progress by having the right people working on the important problems, not by having more people working on more problems. Currently in technical fields my impression is that a PhD is an immigration tool first, a job credential second, and a way of becoming a researcher (not just do research for four years, but an actual interest in a career of research) third. This isn't a rant against immigrants at all, and I'm not blaming them for the current state of research, because for them getting a PhD is just a logical way to play the immigration/career game.
But because immigrants/people looking for credentials see the PhD as an investment in a next step in their life, they've in a sense driven the "market rate" for entry-level researchers to the ground, because they're willing to sacrifice for it. There are a lot of really smart people graduating with Bachelors degrees this year who are going to work at places like Jane Street, Facebook, Goldman Sachs, and McKinsey, and they're all going to make at least $100-200k/year starting, to $250-400k after five years. It would be almost insane for these people to pursue a life of research, because even if they would be very good at it and love the research itself, they'd be working an extremely stressful $20k/year job for ~4 years for the chance of becoming a researcher. I'd bet that a lot of the greatest technical minds of our generation, who could have been great researchers in math, CS, physics, etc., are gravitating towards these jobs because the value proposition of research simply isn't there.
Ultimately what we need to do is pay PhD students more and make it more likely you will actually be able to go into academia after completing one. In other words, we need to make getting a PhD something people do because they want to do research, not because they're trying to change fields, or needed more education to reach pay grade n with an employer, or even because they want to immigrate (again, nothing against immigration, I just don't think a PhD is an appropriate channel from a policy point of view). I think a lot of problems, including publish or perish, would go away if you didn't need to be a top 0.1% researcher to get a good position in academia after graduating.
We also need to be extremely careful to not let a PhD become simply a third tier of tertiary education. I have a lot of problems with the education field offering "PhDs" and "EdDs" which are little more than a second master's program (there's some irony here). This is pretty much exclusively caused by these degrees being arbitrary barriers of entry to promotions within the education field. I hope this doesn't spread elsewhere because it really cheapens what it means to have a PhD.
I didn't get the nervousness of the defense. I was nervous about doing something embarrassing, but not actually of passing.
From what I have seen, it is more common for people to simply quit their PhD candidacy once they form their PhD committee, because the committee members will ask for a soul crushing extra amount of work before the defense (happened to two friends of mine).
As far as I'm aware, defenses at most institutions are formalities. Your adviser has already decided you're done and its just the bureaucracy. Some places don't even have defenses (I don't think Berkley does, at least the chemistry department).