To PhD or not to PhD

[martinbutler]

http://www.cs.cmu.edu/~harchol/gradschooltalk.pdf

Interesting document for anybody considering a PhD.

 

[Daniel Duffy]

6 years to do a PhD? That's awful long.

IMO better to do a PhD in 3 years and then go into industry.

 

[martinbutler]

There must be a way to shorten it surely, but PhDs tend to be long. If you do not want to go into industry and like academia a PhD is a must. I do not know why PhDs are so long in the US compared to Europe and Latin America.

 

[Daniel Duffy]

There must be a way to shorten it surely, but PhDs tend to be long. If you do not want to go into industry and like academia a PhD is a must. I do not know why PhDs are so long in the US compared to Europe and Latin America.

AFAIK in UK/Ireland you do a 4-year dedicated Maths degree minus all the liberal arts stuff. So after 4 years you can start research.
At least, the way it was.

 

[tfors]

Afaik, in continental Europe a PhD is placed as a degree after a (research-)MSc, so you already have 2 years after your BSc.

 

[bigbadwolf]

There must be a way to shorten it surely, but PhDs tend to be long. If you do not want to go into industry and like academia a PhD is a must. I do not know why PhDs are so long in the US compared to Europe and Latin America.

The Americans specialise later (i.e., they waste years in "distribution requirements" in their undergrad degrees and even if they know what they're going to major in, they tend to start at a more basic level, with courses such as college algebra and calc 1 and 2). Secondly, the PhD program is a bureaucratic nightmare -- you have to clear the qualifying courses in the first year or two, you have to clear an oral exam by the fourth year, and only then can you start to work on a dissertation problem.

 

[bigbadwolf]

AFAIK in UK/Ireland you do a 4-year dedicated Maths degree minus all the liberal arts stuff. So after 4 years you can start research.
At least, the way it was.

You can in theory. In reality the frontiers of research in mainstream math have moved forward so at least a year or two of advanced courses beyond the math degree (BSc/MSc) is needed. The lack of such a foundation is what scuppers the prospects of many British doctoral students.

 

[Daniel Duffy]

Afaik, in continental Europe a PhD is placed as a degree after a (research-)MSc, so you already have 2 years after your BSc.

Also too long IMO.

 

[ralm]

There must be a way to shorten it surely, but PhDs tend to be long. If you do not want to go into industry and like academia a PhD is a must. I do not know why PhDs are so long in the US compared to Europe and Latin America.

6 years sounds excessive to do a PhD, even in the US. I don't know if this is particular to CMU, computer science PhDs or just the author being pessimistic, but in my experience 4-5 years is much more common for a US PhD (3 years in most of continental Europe). In Europe the standard path to get a PhD is:
Bachelor (3 years) + Masters (2 years) + PhD (3 years) = 8 years
while in the US it is:
Bachelor (4 years) + PhD (4-5 years) = 8-9 years
so not at all dissimilar and the extra 0-1 year in the US system may be due to more breadth requirements in US institutions or simply a difference in the systems. When I went to MIT to hear about their graduate program (in math) they basically said that practically everyone gets a PhD in 4-5 years (I think it was something like 0.5% are outside that interval, but I'm not sure on that). Less than 4 years would just not make you competitive in the academic market (which seems to be their primary concern) and more than 5 years and people are going to wonder why you were so slow (plus your funding tends to last for exactly 5 years). In the academic market even this seems to be too little, as evidenced by the postdoc system in academia, but if you aim for a position outside academia it may still be too long.

 

[bigbadwolf]

When I went to MIT to hear about their graduate program (in math) they basically said that practically everyone gets a PhD in 4-5 years (I think it was something like 0.5% are outside that interval, but I'm not sure on that).

Harvard also expects math grad students to get their doctorates after four years, with some exceptions taking five (assuming they finish). Maybe the same for Princeton. But outside the ranking universities I think it's the exception rather than the norm. These elite programs are attracting the high flyers, the ones who hit the ground running. Even there, exceptions exist. I know one fellow who took nine years to earn his doctorate in applied physics at Harvard.

 

[C S]

Harvard also expects math grad students to get their doctorates after four years, with some exceptions taking five (assuming they finish). Maybe the same for Princeton. But outside the ranking universities I think it's the exception rather than the norm. These elite programs are attracting the high flyers, the ones who hit the ground running. Even there, exceptions exist. I know one fellow who took nine years to earn his doctorate in applied physics at Harvard.

The AMS regularly publishes these statistics. I remember it being somewhat above 6 years for a math PhD. You're right though: the elite schools are known for graduating students very quickly. Princeton is particularly (in)famous for that.