46

In my area (computer science) it is standard to publish several papers during the course of your PhD and then only "write" your thesis in the last few months of graduate school --- often you only start compiling the thesis once you have secured a job for after graduation.

I realize that in other areas this is different --- your thesis is the primary output of your time in graduate school. However, in my area the primary output is the papers you have published and the thesis seems redundant.

Given that all the content of my PhD thesis already exists in published papers, what is the point of writing the thesis (other than to fulfill the university's requirements)? Why would someone (other than my committee) read my thesis?

I want to know what it is that I am trying to achieve when writing my thesis (as I am currently starting this process). Thus, an alternative question would be: How do I write a good/useful thesis, given that I have published all the results it will contain?

All of my papers have full versions on arxiv, which contain discussion of motivation, background, and related work, as well as all the details and some side results. (Unlike, say, mathematics papers, which tend to be very terse.) So in terms of exposition, I am not sure what would be added in my thesis.

  • 32
    "Why would someone (other than my committee) read my thesis?" - I personally read PhD theses that describe published work because there is often room to go into more detail in the thesis. If I read a published paper that I think will be very useful to me and I want to know more, I often go look for a tech report or thesis by the author. – ff524 Apr 8 '16 at 19:38
  • @ff524 My papers have full versions on arxiv, so I don't expect to give more detail in the thesis. In fact, I might even give less detail. I have only once looked up someone's (masters) thesis, and then only because it contained unpublished results. – omar Apr 8 '16 at 19:48
  • 20
    What is the point? There is none. It's a meaningless social custom that has outlived its usefulness. – Ben Crowell Apr 9 '16 at 0:15
  • 2
    The first good review article is every bit as valuable to scholarship as the original reasearch. The catch is in that "good". If you can't add novel data to the thesis, you can still make it useful by writing it clearly and well. If you can't do that there truly is no benefit to being redundant. – The Nate Apr 9 '16 at 16:29
  • 1
    I have seen theses that consist of a short one-page introduction, together with photocopies of a few published papers. – GEdgar Apr 9 '16 at 21:23
31

Like @ff524 (comment above), I too have found PhD theses useful, and over the years I've collected quite a few of them. I've actually purchased at least 20 since 1992 from UMI/ProQuest. Unfortunately, I don't have access to their digital archive, but I have also photocopied more than double this number from various libraries during this same period. I have often found them to contain fuller explanations and motivation, more historical details, and numerous minor tangential results, all of which tend to get edited out in the typical concisely written published papers (I'm in mathematics).

To answer your question, perhaps what you can do is to write an overview of your results and how they fit into the larger scheme of similar results on your topic that others have done. Even if no one else reads this, it can help by giving a useful summary of your work that you can siphon off when you need to give talks about your work (such as in job interviews) or when you need to write introductions or abstracts for future papers based on the work. Also, if for some reason you find yourself not working on this topic for a few years and thereby forgetting some of the details, and you want to return to the topic, your summary and overview will be very helpful in remembering what it was all about.

  • 1
    Thanks. That is interesting to know. I will try to include more background and side results, although I try to do that in my published results anyway (computer science is known to have much more "wordy" papers than mathematics) – omar Apr 8 '16 at 20:17
  • I also just finished writing my computer science thesis. I think in the thesis you can do a more exhaustive related work section. typically, in conference papers there's not enough space to review related work in detail. also, you could add a background section in the thesis, which covers foundational aspects of the field your work is related to. this helps readers which are not familiar with the field. however, I did not do this in my thesis, because I think it's redundant. often, there is standard literature, which does the job (much better). – beta Apr 8 '16 at 20:25
  • 3
    moreover, in the thesis you can go into more detail regarding technical and implementation details which one would typcially not expect in published papers. what I think is also important is "setting up the stage" in your thesis in the introduction. this is where you try to show the reader how all your papers fit into one commen theme, that is your overarching research topic. hth. – beta Apr 8 '16 at 20:27
  • 1
    @omar It's not just background. If I have to spend any effort at all reading your papers to figure out exactly what you did and how to reproduce it, then there is room for more detail in your thesis. Think of the work you build off of, and how frustrating it was whenever others alluded to results and caveats and tricks without telling you explicitly what they were. Your thesis is your chance to save some poor student in the future all that frustration. – user4512 Apr 9 '16 at 0:43
  • 3
    @ChrisWhite I don't understand your comment. Are you saying the point of the thesis is to compensate for insufficient detail in papers? But my papers should have enough detail in them already. – omar Apr 9 '16 at 0:58
14

tl;dr there is no point.

EDIT: But sometimes it is helpful for the scientific community, especially in case you are dealing with an emerging area, so that your work can become a reference for many people wanting to join in.


That's a good question. And I'll be unconventional (as most my answers here seem to be unconventional), and claim the following:

Indeed, there is not much benefit for writing the thesis in areas, like some parts of CS, where you already published the full versions of your results before submitting the thesis. (A thesis is needed for historical reasons only.)

But this is not a problem, on the contrary: you should be happy, because your job is very easy now. Simply wrap up the papers, put them in some overarching perspective using the introduction, make sure the notation is consistent and you don't repeat the same information too many times, and viola, you have a thesis! Not only a thesis, in fact a thesis that will be successfully defended (with very high probability), since it was already peer-reviewed.

Comment: It is utterly crucial that you already have the full versions of your papers. They should be submitted or better accepted to a journal. Otherwise, I would claim that writing the thesis is equivalent to writing and submitting for a peer-review the journal versions of your work.

  • 5
    @ChrisWhite, the assumption is that the work has already been published in a reasonably-good conference, and the full version is submitted. I stand by my statement: if these assumptions are correct, the PhD will be granted with very very high probability. (for you, I added "with very high probability" in the answer). – Dilworth Apr 9 '16 at 0:34
  • 2
    Also, peer review is not only about correctness, but about merit. Peer review in a good journal is more serious than a PhD committee (from experience). – Dilworth Apr 9 '16 at 0:36
  • 3
    Also, note that the OP is talking about CS. Not about other areas. You might not know the conventions of this area, but they are different than other areas. – Dilworth Apr 9 '16 at 0:46
  • 1
    Thanks. This "pro stapler thesis" view is quite common. (Although it is somewhat disheartening to think that the thesis is a waste of time!) – omar Apr 9 '16 at 1:12
  • 2
    @omar - In lots of other fields, such "stapler thesis" are unacceptable. In addition to being field specific, this might also be university specific. Then, the problem gets down to rewriting the peer-reviewed articles into a thesis, with lots of other details also peppered in, and then, thesis writing is quite a task. That's probably why Dilworth added "_ I would claim that writing the thesis is equivalent to writing and submitting for a peer-review the journal versions of your work._" – 299792458 Apr 9 '16 at 5:15
6

Often the answer is, as @DaveLRenfro has pointed out, more information and context. What you're describing is often called a "sandwich thesis", and may contain chapters at the beginning and the end ("the bread") that are not actually published papers. An introduction linking the papers together for example, or examining prior work and what has come before (and why your contribution is needed).

In the case of my dissertation, which was on a mathematical model of a public health problem, while the major results are all published, my thesis contains a pretty exhaustive account of how each and every parameter in the model came about - the logic behind it, calculations if it needed to be rescaled somehow, assessing whether some distributional assumptions I made were true, etc.

  • Thanks. In a sense my question is "why shouldn't I just create a stapler thesis?" Providing better exposition is definitely good, but I am not sure what I could add above what is already in my published papers. – omar Apr 8 '16 at 21:43
  • @omar Context and more detail, exactly as the answer says. Three freestanding papers aren't necessarily a coherent body of work all on their own. – Fomite Apr 8 '16 at 22:52
  • @omar, even a stapler thesis requires some connective tissue, and there's often some meat there, too. – Bill Barth Apr 8 '16 at 23:18
  • 1
    A summary or overview to tie the papers is the idea behind the "compilation theses" that comprise around two-thirds of doctoral theses in Sweden – carnendil Apr 8 '16 at 23:38
4

‘I want to know what it is that I am trying to achieve when writing my thesis.’

It seems to me that the first person you should be asking is your PhD supervisor.

Since you are here, however, I would point out that one significant function of your thesis is to show that you are expert in the wider discourse.

It should also demonstrate that you are able to communicate your work in a fully structured way (not simply as a sequence of little papers), and that you fully understand where it fits in the wider world.

Any professional thinker (including me, although ostensibly qualified mainly in humanities) can write clever code. A Ph.D candidate, however, should be demonstrating a unified grasp of the field. If you cannot do that, then your external examiner might well fail you. (A friend and colleague of mine recently had to do that to someone.)

Your thesis is an opportunity to draw-together the bits and bobs that you have published (and probably some other material as well) as a cohesive contribution to the discipline. At the very least it ought to be possible for you to write a Conclusion (and therefore also an Introduction) showing how your combined work adds something thematically useful to the field.

I published quite a few bits of my Ph.D research before I submitted, but the point of the thesis was twofold: looking inward at my own original work to tie it all together; and also looking outwards to locate it in relation to other historical and cutting-edge work.

To put it another way...

Anyone can (in principle) come up with novel code, or make a clever film, or translate a German poem.

A real expert, however, can point to the full range of other approaches and techniques, and explain why this one contributes uniquely to the discipline. As a consequence, your Conclusion should also be able to point ahead and suggest what new avenues your (accumulated) work has opened, and what further developments of it might achieve.

The fact that you are undertaking a Ph.D in the first place is indication of cleverness. The thesis is about being professional, and delivering.

2

The purpose of your research is to produce something that people outside your field find interesting or useful. Obviously you cannot always achieve that, but that is why those people are funding your research.

In this grand scheme of things, research articles are rather pointless, as they are written for other people in your field. They are only the first steps in the publication process. They are the individual trees in a forest. Sooner or later, someone has to see the forest from the trees and describe it to a wider audience. That description may take the form of a thesis, a survey article, or a textbook.

This gives you one possible purpose for a "stapler thesis", in addition to the purposes other answers have already told about. While the individual papers describe specific results to a specialist audience, the thesis itself may tell about the big picture to a wider audience.

2

Writing papers is a good part of the job, and it seems you already know how to do it.

But there is another very important thing in research, which is the ability to conduct a research program over many years. It requires very specific skills to envision a topic than will span over multiple articles coherently, establish usefull collaborations, drive students, reorient the project if needed, etc. Beyond the quality of the articles themselves, it is also those skills that the jury will evaluate.

Of course your supervisor has to take some part in this process. It's normal since this is something difficult to learn. But if you didn't took the chance to manage your long-term project yourself during your PhD, you really missed something.

Then, there is often an important part of what you did that simply didn't work, and therefore that you haven't been able to publish. Your thesis is the right place to put these attempts and to explain why it didn't worked and how much you have learned from these errors.

  • 3
    Yes, I agree that a long term vision is inherent to good research. I disagree that the thesis writing will take any substantive part in this, or that even the dissertation committee will seriously consider this towards rejecting the thesis. I claim that there is almost no example of a thesis rejected due to "the researcher failing to establish a long-term vision, or carry out a research direction by him/herself" (assuming that the papers have been accepted to reasonably good conferences/journals). – Dilworth Apr 10 '16 at 14:03
  • I think the supervisor is supposed to be responsible for the vision part before graduation to get the student in a well-planned spot but sometimes students accidentally get their own vision before graduation... – mathreadler Aug 3 '17 at 14:29
0

One of the main purposes of a research degree such as a PhD is getting a stamp of approval that one is able to conduct academic research independently. The thesis is presented and publicly defended against criticism from other scholars showing the candidates abilities to conduct independent research in ways that the papers themselves might not always show. Maybe especially in the cases where the papers could be having multiple authors, the thesis only has one and only one is to defend it.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.