Other What's the point of a thesis?

[ergospherical]

I will play devil's advocate to some extent. I'm looking from the outside in and won't have the same insight as some of you PhDs...

If a research paper is around 20-30 pages tops, what's the point of spending all that time to write up a 100, 200, 500-page PhD thesis? I get the impression from that the actual new content could be presented on a tenth of the paper, if you cut out all of the exposition.

There's probably a historical element. Do you think the format is out-dated?

 

[pines-demon]

 

[billtodd]

Research paper might be more than 30 pages long.

Russell and Whitehead's Principia I don't think that they wrote the three books in articles format and then rearrange those articles into one big manuscript.

 

[Orodruin]

This may be different in different subjects as well as in different countries. I can mainly account for the purpose where I work and in my field. Keep that in mind when reading.

The PhD thesis is not only about presenting new research. In fact, the typical thing to do in my field is to already have several papers published and compile them into a thesis with an extensive introduction. The thesis and the thesis defense constitute the examination of the PhD. Not only should the PhD candidate be able to write papers (of which they may or may not - but usually not - be the single author), but they should also show ability to place their work in a larger context and show an ability to reason in the chosen subject. This is what the thesis introduction is for. The research part is shown by the published papers.

 

[Vanadium 50]

Before we go too far down this path, have you ever actually read a thesis? Or is the criticism based on what you think you would find if you did read one?

 

[ergospherical]

This may be different in different subjects as well as in different countries. I can mainly account for the purpose where I work and in my field. Keep that in mind when reading.

The PhD thesis is not only about presenting new research. In fact, the typical thing to do in my field is to already have several papers published and compile them into a thesis with an extensive introduction. The thesis and the thesis defense constitute the examination of the PhD. Not only should the PhD candidate be able to write papers (of which they may or may not - but usually not - be the single author), but they should also show ability to place their work in a larger context and show an ability to reason in the chosen subject. This is what the thesis introduction is for. The research part is shown by the published papers.

I guess that is the reason. If the field is so specialized that you can't reasonably set a standardized exam in it (for 1 student), then a long exposition is some measure of understanding.

The Dyson video is interesting. From my perspective of not having done / not intending to do a PhD, I think I'd agree with what he's trying to say. 4/5 years is a long time to devote to a particular problem.

It seems like a shame that there's no "modern" alternative for one or more short-term research projects beyond something like a 1-year Masters. There's opportunities for things like this in industry & internships, though - but it's slightly different context.

 

[jtbell]

It seems like a shame that there's no "modern" alternative for one or more short-term research projects beyond something like a 1-year Masters. There's opportunities for things like this in industry & internships, though - but it's slightly different context.

What purpose would this serve?

 

[Vanadium 50]

4/5 years is a long time to devote to a particular problem.

And?

I spent 15 years solving one problem.
The field spent 49 years determining whether the Higgs mechanism was the correct description of electroweak symmetry breaking.
It's been 110 years since the discovery of the nucleus and we're still learning about the forces that hold it together.
In mathematics, it took three and a half centuries to prove Fermat's Last Theorem.

Some problems take time.

 

[pines-demon]

And?

I spent 15 years solving one problem.
The field spent 49 years determining whether the Higgs mechanism was the correct description of electroweak symmetry breaking.
It's been 110 years since the discovery of the nucleus and we're still learning about the forces that hold it together.
In mathematics, it took three and a half centuries to prove Fermat's Last Theorem.

Some problems take time.

But that does not mean that somebody should spend all their time on the same problem, specially if the problem does not require it or the person does not want to. The thesis system sometimes forces that.

 

[Orodruin]

The thesis system sometimes forces that.

That’s not entirely accurate.

Edit: Well, I guess sometimes, but that’s not necessarily bad. Bad would be all the time regardless of whether it is motivated or not.

 

[CrysPhys]

But that does not mean that somebody should spend all their time on the same problem, specially if the problem does not require it or the person does not want to. The thesis system sometimes forces that.

<<Emphasis added>> I can understand an advisor requiring a student to spend more time on a problem than the student wants to. But could you give examples of an advisor requiring a student to spend more time on a problem if the problem does not require it? It's in the advisor's self-interest to move on to something else if the problem is indeed satisfactorily solved, correct?

 

[CrysPhys]

If a research paper is around 20-30 pages tops, what's the point of spending all that time to write up a 100, 200, 500-page PhD thesis? I get the impression from that the actual new content could be presented on a tenth of the paper, if you cut out all of the exposition.

As others have noted, the protocol of what's in a thesis will vary among different universities. But one reason why a thesis is often considerably longer than a publication is because the thesis contains more nitty-gritty detail. This is especially true in experimental research. The thesis, e.g., will include details of sample preparation, apparatus design, and apparatus operation that are either not included or briefly summarized in a publication.

 

[Orodruin]

Theory theses are long, but for different reasons. Let's take a completely random example:
https://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-4415
This is a typical compilation thesis as mentioned above. The "full text" pdf contains the introduction, while the printed version also contained the seven papers. The introduction puts those papers in context as well as introduce readers who may be not directly familiar with the subject to details not covered in the papers. The full length of the thesis including both papers and introduction was about 250 pages so that's half-half between introduction and the actual research papers. I would not say this particular PhD student focused on a singular problem only as much as looked at different problems within a particular subfield.

 

[CrysPhys]

The Dyson video is interesting. From my perspective of not having done / not intending to do a PhD, I think I'd agree with what he's trying to say. 4/5 years is a long time to devote to a particular problem.

It seems like a shame that there's no "modern" alternative for one or more short-term research projects beyond something like a 1-year Masters. There's opportunities for things like this in industry & internships, though - but it's slightly different context.

I won't address outliers such as Dyson. But most PhD candidates (especially in US programs) do not enter a PhD program qualified to perform original research. There is an extended learning period, at the end of which, if successful, they are qualified to perform original research.

Compare the medieval guild system, in which a person progressed from apprentice to journeyman to master craftsman. Or compare the US medical system, in which a person progresses from med student to MD to intern to resident to board-certified physician.

 

[ergospherical]

I don't doubt that it takes a considerable amount of time and effort to reach a level where you can contribute to solving problems that nobody else has. That's obviously why PhDs take so long.

I was more asking about the reasons for every candidate to write what is - essentially - a book. There have been some good reasons suggested, which are probably true, but to me (a relative outsider to academia) it seems like a bit of an old-fashioned requirement.

I'm sure it takes a lot of time to write a document that long - how many people will read it...?

 

[billtodd]

I don't doubt that it takes a considerable amount of time and effort to reach a level where you can contribute to solving problems that nobody else has. That's obviously why PhDs take so long.

I was more asking about the reasons for every candidate to write what is - essentially - a book. There have been some good reasons suggested, which are probably true, but to me (a relative outsider to academia) it seems like a bit of an old-fashioned requirement.

I'm sure it takes a lot of time to write a document that long - how many people will read it...?

Some people like myself read books from cover to cover... technical ones.

It indeed takes time to read those books.

 

[billtodd]

@ergospherical obviously not a lot of people will read some technical book.

 

[Choppy]

Just to throw in my own two cents on the matter... one of the big reasons for the thesis (or more accurately, a dissertation) is documentation.

As you work on a PhD, you do a lot of research. You start with a tremendous amount of background reading and skill development in your own particular sub-field. You systematically develop tools, specialized techniques, validate ideas and prove that your equipment works the way it's supposed to. Then you systematically investigate usually not just one, but a series of related problems, and make novel advances on them.

A dissertation is a concise, coherent record of all of that work. It explains not only what you did, but why you did it, what the results mean, how multiple projects fit together, how they should be interpreted, and how your work ultimately fits into the bigger picture of the advancement of human knowledge in your field. Not only is it a demonstration that you as the author know what you've done, but it becomes an accessible, objective record of you having done it.

 

[Vanadium 50]

- how many people will read it...?

One data point. I printed 100 copies of my dissertation. I ran out (well, I kept two for myself). Fifteen or more years later I was in a talk by a theorist (not someone in the 98) and he showed one of the plots in my thesis.

 

[ergospherical]

That’s pretty cool. I remember that de Broglie’s famous result was in his thesis too. I also didn’t know until recently that a lot of the formalism of astrophysical disks was worked out by Pringle and Rees during their PhD projects.

Sure there are many other examples. But I guess most people’s first big publication isn’t quite that good.

 

[CrysPhys]

I'm sure it takes a lot of time to write a document that long - how many people will read it...?

People will likely not read it cover-to-cover, but those portions that are relevant to their work. To re-emphasize what Choppy wrote, the thesis serves as an archive to preserve a record of what the candidate has done. Archives in any field are generally not read cover-to-cover. But for the benefit of future researchers, it's important for the information to be retained. The thesis also serves as an existence proof of what the candidate is capable of. Much as in centuries past a master craftsman would fabricate a special object to demonstrate what he was capable of, regardless of the number of people who actually used the special object.

 

[Vanadium 50]

But I guess most people’s first big publication isn’t quite that good.

On what grounds do you base this claim? How many have you written, supervised, or even read?

 

[mathwonk]

I agree especially with points made by Choppy and CrysPhys: first of all, the purpose of writing the thesis, for most of us, is to strengthen the author's research chops to the point of being ready to do, and write up, competitive research - i.e. it is a learning experience for the author of the thesis; hence the more detailed and longer, the better. Indeed it may well give opportunity to record background facts that are hard to find in print, although widely assumed.
Secondly, with regard to any original discoveries made by the author, it serves as a historical record that the work was actually done by the young person, who otherwise has no reputation that would protect him/her from intellectual poaching.
One benefit that in a sense combines the two purposes, is also that a thesis allows more scope to give convincing detail as to the correctness of new results, than does a typical published article where space is at a premium.
I am of course speaking from a perspective of pure mathematics theses, but it seems the principle is more general.

 

[symbolipoint]

@ergospherical obviously not a lot of people will read some technical book.

I want to say something about that comment but just do not know how to put it.

Here's a try: The comment is far too general and understandably true; but the technical book or even the thesis was not written for a wide varied audience.

 

[hutchphd]

it seems like a bit of an old-fashioned requirement.

Better you should just copy it onto a 3.5 inch floppy disc? I think there is something to be said for a technology that has withstood millenia (well there was that nasty fire in Alexandria....), is perhaps a dealimiter of civilization, and it serves as a touchstone if nothing else. My thesis was not a huge success on the Best Seller list and was a mild pain the ass logistically but I am pleased it exists, and I am not usually sentimental.
I think it is also a holdover from a time when doctoral work was not immediately (pre)published in a journal. A bound copy implied permanence and weight.

 

[Vanadium 50]

Better you should just copy it onto a 3.5 inch floppy disc?

Hey! My thesis was on a 3.5" floppy disk!

I did have to turn in 3 printed copies, however, printed on special low-acid paper.

 

[hutchphd]

That’s pretty cool. I remember that de Broglie’s famous result was in his thesis too

And Feynman's was the path integral approach to QM which I find a remarkable piece of work......

 

[billtodd]

I want to say something about that comment but just do not know how to put it.

Here's a try: The comment is far too general and understandably true; but the technical book or even the thesis was not written for a wide varied audience.

Obviously.
I always encourage my folks to read some of the books I have in my personal library, but somehow they don't fancy reading a book on Infinitesimal Calculus or Linear Algebra... I guess I am one of a kind.

 

[ergospherical]

Theory theses are long, but for different reasons. Let's take a completely random example:
https://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-4415

On an unrelated note, this is actually really interesting. My project is related to sphalerons. Give your completely random example my compliments.

 

[Orodruin]

f the field is so specialized that you can't reasonably set a standardized exam in it (for 1 student), then a long exposition is some measure of understanding.

This has been nagging my mind a bit. Standardized written exams are among the worst ways of approaching testing understanding when it comes to the quality of the testing itself. They are however extremely efficient in terms of scaling to a large number of people, which is why they are used. For quality of showing understanding, nothing really beats personal exposition, both oral and written. Even if it were possible to set standardized exams at the highest level of academic studies, I would argue that this should not be done simply because of quality reasons.

 

[pines-demon]

I agree with most of you. Thesis are spectacularly useful in our field. However when the thread started, I answered with the Dyson video because I think that the PhD system is not perfect. It has is benefits and problems and I just think we should have both sides of the coin in an honest discussion.

The advantages depend a lot on the format. Limited-time PhD contracts exist which lead to many problems like rushed publications, unlimited stress to find something interesting before finishing and not having enough time to steer your research into another topic if it is not working. Near-unlimited PhD studies are nicer in that aspect but suffer from the problem that not-so-good candidates can take too much time to graduate reducing and hurting their opportunities to find a job latter if the PhD does not do so well. Note that most PhD do not get tenure.

Many have given examples of excellent theses, but there are a lot of mediocre ones out there. Not every successful physicist has a specially noteworthy thesis (many change topics afterwards). And for career track it does not matter that much.

Also I think that we should leave the door open for improving the system. For example what would be so wrong of having special Phd-like contracts that are long but finite and you are only required to help your team publish? This kind of contracts will allow the same training and to build a book (thesis) if your advisor/team think it is of specific value, but if not you will dedicate your time to solving problems and publishing like if you were in a research job already. You could even do multiple of these contracts before deciding to go for a postdoc or get a tenure track. This will allow you to train yourself and do research in less constrained manner.

The current Phd system does not seem fit for a publish-or-perish culture (maybe is this culture that we have to change).

 

[Vanadium 50]

This has been nagging my mind a bit.

Me too.

I am coming at from a different direction. A PhD is a research degree. "Can't you just take a test?" doesn't address whether someone has learned that or not. The same argument can be made for senior recitals in music degrees, or seeing patients for medical degrees, and so on.

The question really is "should there be a doctoral-level degree that only requires regurgitating past knowledge, and not creating new knowledge?" I don't see the point. There is one in engineering, the "Mechanical (Civil, Electrical, etc.) Engineer" degree. It's rare, and that's another way to say unpopular. And there is probably more of a need for it in engineering than there is in physics. So again, why do we need such a thing.

 

[ergospherical]

Landau had a famous `theoretical minimum' exam to admit students to his theoretical school. He must have considered it to be a pretty robust measure of understanding...

 

[sbrothy]

Someone may have pointed this out but it's not unusual for, for instance, medical students to perform experiments over several years as part of their degrees (I heard of a 2-person joint thesis which took 6 years!).

In these cases, although I'm sure journals, permits and so on is a lot of pages (and I'm not even sure that counts) compared to the amount of time used I don't think several hundred pages is a lot.

[Should that be "are a lot?" It just sounds wrong to me.]

 

[Orodruin]

Someone may have pointed this out but it's not unusual for, for instance, medical students to perform experiments over several years as part of their degrees (I heard of a 2-person joint thesis which took 6 years!).

There are plenty of LHC students who can easily up that …

Landau had a famous `theoretical minimum' exam to admit students to his theoretical school. He must have considered it to be a pretty robust measure of understanding...

So what? That will test understanding of things that are known, not the ability to do research to produce new knowledge. The latter may benefit from the former, but is in no way a guarantee.

 

[sbrothy]

There are plenty of LHC students who can easily up that …[...]

Oh, I don't doubt that. Not trying to belittle anyone's profession. That was just an example I personally knew of.

 

[ergospherical]

So what? That will test understanding of things that are known, not the ability to do research to produce new knowledge. The latter may benefit from the former, but is in no way a guarantee.

Right, maybe. But Landau's school produced remarkably successful research. And it appears like the main criterion for admission was whether you could solve these famously difficult problems.

It seems like a different route to, say, admitting faculty based on their graduate research experience and dissertation.

Not saying one method is "better" than the other! Just makes you wonder if the classic PhD & dissertation is too restrictive.

 

[ohwilleke]

I will play devil's advocate to some extent. I'm looking from the outside in and won't have the same insight as some of you PhDs...

If a research paper is around 20-30 pages tops, what's the point of spending all that time to write up a 100, 200, 500-page PhD thesis? I get the impression from that the actual new content could be presented on a tenth of the paper, if you cut out all of the exposition.

There's probably a historical element. Do you think the format is out-dated?

A small nitpick: The long paper produced in connection with earning a PhD is usually called a "dissertation", while the term "thesis" is more often used to refer to the long paper produced as a capstone requirement for a Master's Degree (although not all master's degrees require a thesis), at least in academia in the U.S.

Another point worth noting is that there is significant variation in the length of a dissertation by discipline. In mathematics, dissertations are frequently shorter than in almost all other disciplines, for example. Some mathematics dissertations are 40 pages or less.

And, one of the reasons that dissertations are so long is that unlike research papers, which apart from a brief introduction, take all prior work in the field as a given, a dissertation is expected to have a comprehensive review of the existing literature in the field to put the new work into context. This is an important pedagogical feature of this project that is intended to force the author to self-study the field, even though it may be of limited benefit to others. The review of the literature in the field is typically 1/5th to 1/3rd of the entire dissertation.

Another issue related to the length is that maybe the problem is not that dissertations are too long, but that research papers are too short.

The length of research papers is driven by academic promotion incentives that favor publishing as many papers as possible without much regard to the length or quality of each paper, which creates an incentive to publish the "minimum publishable unit" rather than the optimal amount of material in a single presentation.

The length of a research paper also driven by the length expectations of scientific journals which typically have an expected range of number of pages that is quite short. But a significant share of research papers also come with "supplemental materials" and appendixes that are often several times as long as the core research papers, with the total number of pages in the main paper and the supplements approximating the length of a dissertation in a big paper covering a broad topic. So, while the paper itself may be 15-30 pages, there may be 60-100 pages of supplemental materials. It is getting to the point in some sub-disciplines where reading only the paper itself, without the supplements, is almost like reading a long abstract, rather than a full paper.

Finally, it is worth noting that the dissertation is a much smaller barrier to getting a PhD in STEM than it is in non-STEM fields. The average length of time it takes to earn a PhD in STEM is somewhere around 5 years, while it is closer to 8-10 years in many non-STEM fields, and that is mostly due to the longer period of time it takes to produce a dissertation in non-STEM fields. So, the downside of a dissertation reducing the flow of PhD graduates isn't that great in STEM.

 

[Vanadium 50]

But Landau's school produced remarkably successful research.

Remarkably?

The Soviet Union produced perhaps a half-dozen Nobel prizes then and after. The US alone produced around 100 in that period. The entire USSR was comparable to maybe Canada.

Yes, some good solid work was done in that time and place. But "remarkable"? The contribution of tiny countries like Denmark is remarkable. The fact that a superpower could barely keep up with a single US institution like Berkeley is hardly evidence that their model should be emulated.

 

[ergospherical]

It’s not a competition…!

The Landau institute was a relatively small outfit and did a lot of influential work. What’s interesting about it is the non-standard admissions procedure. You only had to convince Landau that you could pass the Theoretical Minimum - he didn’t care about your research credentials.

 

[martinbn]

It’s not a competition…!

The Landau institute was a relatively small outfit and did a lot of influential work. What’s interesting about it is the non-standard admissions procedure. You only had to convince Landau that you could pass the Theoretical Minimum - he didn’t care about your research credentials.

Why is that non-standard? It is always the case that to get the gig you need to convince the commitee to choose you.

 

[pines-demon]

It’s not a competition…!

The Landau institute was a relatively small outfit and did a lot of influential work. What’s interesting about it is the non-standard admissions procedure. You only had to convince Landau that you could pass the Theoretical Minimum - he didn’t care about your research credentials.

As others have said that's a really unfair comparison. Not every excellent physicists came from this kind of selection. In the Soviet Union it was probably was good way to find students but that means that it was extremely selective for those students that were not good at this kind of examinations. I do not think Landau allowed for taking the exams twice. Also many fields of physics were not invented yet and many people did not carry university students at the time.

Now, undergraduate studies do not suffice, there is a very large number of people going to university, the physics field is almost twice as large as before, not everything is theory, and the increasing number of professors means also that they need more graduate students.

In the end it depends on what route we want to take. Do we want to have only Nobel Prize material and the rest should quit? Let us do that sooner, let us rise the difficulty even more starting in high-school. Let every student know that physics is hyper-competitive from the start and only let the Landau-level students go for it. That way there is funding and permanent positions for everybody.

The other option is to be more kind to those students that love science. Find ways to let a large number of students in physics go through their studies without much suffering, adjust better to the students aptitudes, allow for more adequate training for the current market and for alternative paths, not just academia. Do not let them (a) start a thesis if this is going to be a bad experience (see previous comment) (b) stay indefinitely in a PhD or a postdoc track (c) work in precarious professional conditions and underpaid.

 

[martinbn]

In the end it depends on what route we want to take. Do we want to have only Nobel Prize material and the rest should quit? Let us do that sooner, let us rise the difficulty even more starting in high-school. Let every student know that physics is hyper-competitive from the start and only let the Landau-level students go for it. That way there is funding and permanent positions for everybody.

That's not how it was at the time of Landau. There were many, the majority, that did not work with him.

 

[pines-demon]

That's not how it was at the time of Landau. There were many, the majority, that did not work with him.

You mean in Kharkiv on in the rest of the Soviet Union? anyway that is far from the point.

 

[martinbn]

You mean in Kharkiv on in the rest of the Soviet Union? anyway that is far from the point.

May i missed it, what was your point?

 

[pines-demon]

May i missed it, what was your point?

In that paragraph I was suggesting a hypothetical. That if we only want the best and the field is going to be hard on resources and difficulty, ONE of the options is that we can start the filters sooner in the physics students lives.

 

[FactChecker]

I can only refer to my own experience in pure mathematics. My thesis skipped no steps in the proofs, some of which were tedious (but IMO, not very routine). That was the only way that my adviser and the professors could really be sure that my proofs were valid and that I knew what I was talking about. All the necessary definitions and references were given (possibly to prove that I had done the necessary research).

When the same proofs were written for publication, much of the detail was omitted. It was assumed that if readers really wanted to verify the results they would put in the time and effort to go through the details on their own. It was also assumed that a reader was already reasonably familiar with all the background material and references. Experts in the field do not want to read my summary of the basics. All that reduced the page count considerably.

 

[ohwilleke]

I can only refer to my own experience in pure mathematics. My thesis skipped no steps in the proofs, some of which were tedious (but IMO, not very routine). That was the only way that my adviser and the professors could really be sure that my proofs were valid and that I knew what I was talking about. All the necessary definitions and references were given (possibly to prove that I had done the necessary research).

When the same proofs were written for publication, much of the detail was omitted. It was assumed that if readers really wanted to verify the results they would put in the time and effort to go through the details on their own. It was also assumed that a reader was already reasonably familiar with all the background material and references. Experts in the field do not want to read my summary of the basics. All that reduced the page count considerably.

Good point. Imposing a higher standard of rigor in a thesis really serves two purposes (which make it longer than a research paper on the same topic):

1. A higher standard of rigor is necessary because unlike someone who has earned a PhD (and generally been hired as a professor or research fellow), one can't take for granted that a graduate student "filled in the gaps" between what would usually be presented in a published paper correctly; and

2. It serves an educational purpose of forcing you to learn all of the necessary steps and apply them correctly, looking up and citing to definitions and theories that are often taken for granted without knowing their exact source.

 

[Orodruin]

unlike someone who has earned a PhD (and generally been hired as a professor or research fellow), one can't take for granted that a graduate student "filled in the gaps"

Unfortunately, one cannot take that for granted with someone who has earned a PhD either ... I know of a number of instances just from the case studies in a research ethics module given to PhD students in the program I direct.

 

[FactChecker]

Unfortunately, one cannot take that for granted with someone who has earned a PhD either ... I know of a number of instances just from the case studies in a research ethics module given to PhD students in the program I direct.

Yes. But I did my thesis in the old days when a person who lied on a published paper was thrown into the T-rex cage to be eaten. ;-)