Who cares about your PhD thesis?

Question

I'm pursuing a Ph.D. in a computer science-related field (more particularly, in Artificial Intelligence). During my studies, I was involved in several projects, some of them got published in top-tier conferences. I'm ought to submit my Ph.D. thesis this year and was wondering what's the role of theses these days in CS.

Generating a polished thesis would take at least two months. However, collecting the work we've already published into a huge Latex file with a shared (newly written) introduction and discussion would take one week. I believe I can use the spare time to advance my research even further or start new collaborations.

If that matters, my career plans are to start a postdoc position, followed by seeking an academic position.

So, is there any significant downside to hastily constructing a "sandwich thesis" as opposed to carefully polishing a cohesive document? Assume that my committee would accept both.

Answer 1

Who cares about your PhD thesis?

At least your supervisor and your examiners, and potentially anybody seeking to evaluate your work in the future.

The primary reason to be careful with your PhD thesis is to avoid giving a bad impression to your PhD committee. Even if the content is solid, it would be unfortunate to be given major corrections or a lower "cum laude" honor (depending on the local usage) just because a committee member felt that you didn't do much effort with the dissertation.

It is true that most dissertations are not going to be read by many people, but your PhD dissertation is going to be a quite important part of your profile for your whole academic career. A reviewer for a hiring committee might have a cursory look at it, a student considering working with you might be curious about it, etc. Of course it's unlikely to be a major issue, but is it really worth the risk? Even simply for yourself, isn't it worth spending a bit more time in order to have a document you can be proud of in the future?

Answer 2

Nowadays some universities offer a "cumulative dissertation" (putting your papers together in the journal style with a introduction at the beginning of the dissertation) additionally to the traditional thesis publication format. I would ask your supervisor, if he allows and recommends this based on your publication track (number, extent, content, depth, siginificance, originality of your publications), as only he can likely make a good judgement here.

Apart from this, if you really think/know not much people will care about the context, background, methods of your research, this might be a reason to favor the cumulative dissertation. But I read during my PhD many PhD thesis of predecessors in the field to grasp above three points, as in papers details often have to be left out. Some thesis get also published as book by springer & co.

I don't think it can be a downer pursuing professorship and the institutions you apply for also allow a cumulative dissertation.

The PhD thesis gives you also a training effect to publish books or review articles, because of more similar format style like a journal article or letter. And to become a professor such publication contributions to the community can be very advantegeous to get tenured.

A PhD thesis also gives people an inlook how you were/are working, how did you approach a scientific question. As many papers are the work nowadays of many co-authors and collaborators and many PhD's are produced, a non-comulative PhD thesis might become very "trendy" and advantageous again to have over competitors.

Answer 3

The main issue I see with this plan is that the papers you already have worked on have been collaborative efforts with other people (your use of 'involved in' and 'we've'). The PhD needs to be your own work, to demonstrate that you personally can work to a level required to attain a Doctor of Philosophy in your field. This means producing innovative research independently of others, despite working in a team environment.

In Australia at least, there is a growing trend for PhD by Publication. I suggest you speak to your supervisor about this (or find out what the local version of the term is; you didn't specify your country of origin). The logic behind it is that if you have a number (typically three) publications which have been written by you, and independently peer reviewed by reputable reviewers, then this amounts to the PhD. This typically means publication in Q1-rated journals; conference papers typically are insufficient without a formal peer-review.

Students like the PhD by Publication approach because its hard for the examiners to refute anything which has been peer reviewed already. Some supervisors also like it, although others like mine see it as the lazy route, both for the student and for the examiners, as some of the latter may not even read those peer reviewed parts. Personally I tend to agree with the latter: in a 3 year PhD, it means getting a fully peer reviewed publication produced every 12 months. Typically the first 12 months are spent reviewing literature and deciding where your project will fit within the greater field of research, so it's hard to produce good, PhD quality research in the first year, although it sounds like you are nearing the end, so this may not be relevant.

In a traditional thesis (at least in Australia), you are required to include a List of Publications which you have worked on (this is independent from the usual bibliography section). The administrative staff at my university have recently made it a requirement that you must include the percentage of your contribution within this list, which needs to be quite detailed: you must list that you contributed 10% of the introduction, 20% of the content, 5% of the results, 10% of the conclusions, etc. The point is that they really want to make sure that what you have included in your thesis is 'your' work. This can also have consequences for the bulk of your thesis. In my case I'd written a paper on process control, and collaborated with another author who'd added some computer simulations. Although the system itself was entirely my work, I had to completely rewrite it and rearrange the model for the thesis so that it was distinct from the paper, to remove any question that I'd included research that wasn't my own in the thesis, or plagiarized a published paper (even one I'd collaborated on). The thesis certification page at the top of the submitted thesis includes the words 'This thesis is the work of authorname except where otherwise acknowledged. The work is original and has not previously been submitted for any other award, except where acknowledged.'

So in short, simply collecting your team's work into one huge latex file is unlikely to qualify as a PhD thesis.

Answer 4

A thesis is a whole different document than a collection of papers.

Over the course of your Ph.D., you have arguably worked (and published) on a number of different projects and topics that are likely to be only sparsely related to each other. While your narrow academic community will certainly draw more benefits from your papers than from your thesis, you will learn a lot by writing a thesis.

Writing a cogent thesis that seamlessly connects the dots between the different topics on which you worked is hard. But it helps you immensely in rationalizing about your specific contributions, how it affects your field, and more importantly, what you have accomplished and how you can move on with your life. Is this field worth exploring? Were the techniques you used useful? Are there other promising approaches? Is the problem still relevant? What did you personally gain from everything you did? Are your multiple projects consistent with each other? Would you recommend people continue working on what you worked on? How? Are you contributing to society in a meaningful way? Are you advancing knowledge? How? Are there specific avenues that you didn't pursue (perhaps for the lack of time or previous knowledge) that look promising? Thinking rationally about these issues will train you in skills that will be relevant to your future, especially if you want to continue being a researcher. A great way to think about these things is by writing them on paper, which is what you will likely do for the rest of your life (as grant proposals, executive reports, white papers, etc).

Remember that working in the lab and writing papers is only a very small part of your training as a Ph.D. student. In fact, in a few years you are, statistically speaking, likely to never work in the lab, write copious amount of code or write papers in full again for the rest of your life. But you will have to continue thinking rationally about your's and everyone else's reseach and work, and that's what you learn by writing a thesis.

PS: There are also additional benefits to writing a good thesis, such as providing additional methodological details that might be useful to the community and (most importantly) to your labmates and to the people who will use the same equipment/techniques as you did in the future.

Answer 5

I would recommend the latter option, but taking the introduction a bit more seriously. One week seems awful short to do a good job. Think about it carefully and spend two or three weeks to write a good introduction for a broad audience which includes more background.

Answer 6

Depending on field, there can be substantial downsides, because academic papers and PhD theses are fundamentally different beasts.

There are certain things that people just don't publish in academic journals, but that is nevertheless useful information. Experimental or theoretical failures is the obvious one that everyone knows about - it's much harder to publish an article about "I tried this and it didn't work", than "I tried this and discovered something cool". Further, journal papers (which are not called "letters for nothing") often have stringent word, page, figure, and even citation number limits, forcing academics to pick and chose only the most critical information to include. That excluded information is often to be found in PhD theses, where those limits don't apply. If it isn't, then it's lost forever, pretty much.

Speaking for my own field (experimental atomic physics), I go looking for academic papers when I want to study advanced scientific topics - because there is nowhere else to get most of that information - but when it comes to technical details, long-form PhD theses are second only to informal, personal discussions with other people doing nearly the exact same thing as you.

"Technical details" is, itself, quite a loose term, but for me that has included things like:

• Specific part numbers - "Part X from company Y achieves the best performance of all 18 options we tried"
• Unexpected failure modes -"this type of glass suffers from thermal lensing, so you have to have windows made from this other type of glass"
• Undocumented specifications - "Part A from company B is specified to do P, but can also do Q"
• Unknown suppliers - "samples from "
• Techniques that didn't work and so were not published
• Techniques that did work, but weren't published - you'd be surprised how many of these there are.
• Alternative solutions to expensive hardware - "you could buy the all-singing-all-dancing computer controlled widget from company X (that you don't have the budget for), or you can build your own with an Arduino, this code, and half a day's soldering"
• Simple solutions to problems that nevertheless would take a considerable amount of time and error to identify yourself - "you need a 4th order high-pass filter for this signal because a 2nd order can't solve interactions with this other effect"

Off the top of my head, my own thesis cites something like 40 other PhD theses for useful information that saved me time, energy, and my supervisor's funding. That was out of 250-300 references in total, so 15%; the vast majority of the rest would have been journal papers.

Nobody likes writing a PhD thesis - and it's true, very few people will ever read them compared to, say, Lord of the Rings or Harry Potter - but it is a valuable source of information which never makes it into published papers. Think about all the things that you have learned over the course of your PhD - if you never write it down, you may as well have never learned them.

Answer 7

Short version : no one. How much you care for it is something, but the supervisor or the examiners couldn't care less. I'm a math and computer science major from Romania if this info is worth anything to you.

I say all of this because I've experienced it . So , I've been working on my application for the thesis for a year , a hole year. The application wasn't some website like the majority of my schoolmates did, it was an app that could read a barcode with the help of a usb scanner. I've built everything from scratch with libraries, made it so there were almost no bugs, worked like a charm and did more than half the app before finding a supervisor. He didn't look through the code, not even a glance. I other words he didn't care about the code.

On the written part I've worked about 2 months . About this one too i was pretty proud. All the parts fitted together nicely and didn't have filler info. The supervisor read through this, several times.

On the presentation day, the examiners flipped through the written part just to see if i got any interesting images , asked something about the part of the app that involved that USB scanner and the code of the reports and that was it. Next. All the work i've done over that year was reduced to them to just two chunks of code. Nobody cared that the app was nearly perfect, nobody cared about the written part.

So my advice is don't put too much work in it if you intend to leave the project just as the thesis, ultimately nobody will appreciate for what it truly is

Answer 8

I agree with other commenters that, in general, not a lot of people will care about your thesis, per se. However, you need to also consider what job you are wanting in the future.

If you are applying for academic jobs, it will likely be under much greater scrutiny, particularly by your potential postdoc advisor (I always read the thesis of postdoc applicants to my group and I know many other professors do the same). For one, it will demonstrate whether you have the tenacity to conduct long, well thought out research. Second, it will demonstrate whether you can effectively communicate that research. Third, it will demonstrate if you know how to write a thesis and, therefore, have a chance of mentoring your future students through the process. Note that even postdocs are asked to mentor students through the process, particularly in larger research groups. If you have not gone through the long, meticulous process of writing out the thesis, you will not be qualified to do this.

Your statement "Generating a polished thesis would take at least two months," gave me pause. Yes, that is true. Probably even a bit longer than that. But this is what a PhD actually requires. Getting the degree demonstrates that you had the tenacity to go through it. And there is nothing wrong with NOT wanting to go through it! But if you don't want to put in those few months, then maybe you might want to consider going to work in industry with the master's degree?

You might find universities that accept stacking several research papers together into one thesis to be acceptable. My university does not, however where I got my PhD did. At the same time, while it was an accepted format, I knew someone who used it and their committee tore it apart and they basically had to rewrite the entire thing after their defense. The reason for this is that most professors wrote their dissertations in a time before universities considered it an acceptable format. So they didn't consider it a worth while option, even if the university did choose to accept it. I would make time with your PI and committee to discuss the format, if it is an option for you, before you make this decision.

Answer 9

Let me answer by focusing on the purpose of the thesis, derived by the purpose of a PhD.

Disclaimer: This is an opinionated answer, I haven't done research in CS, and I don't have supervised PhD students — so take my word with caution.

I believe (p. xiii) that:

The purpose of a PhD is to train a research student as an autonomous scientist and a good researcher — i.e. as someone deserving the grade of Doctor.

The objective of a PhD thesis is hence to demonstrate to one’s peers that its author can be considered in this way by reporting the successful completion of a high-quality piece of research.

_{Side note: don't get me wrong. “Successful” is here about the completion of the project, and not the research itself: I do believe that if the research question was relevant regarding the literature review and the experiment well designed and implemented, the project is successful even despite null/non-significant results.}

So is a sandwich thesis OK?

If it consists in multiple papers about research you have done (i.e. you are not simply the co-author that have crunched the numbers, or written the paper without doing the experiments — the point here is to demonstrate that you master all the various aspects of being a researcher: crunching the numbers is one of them, but writing is another, as well as formulating a research question, designing an experiment, etc.), I would consider so.

In such case, I would stress the importance of the global introduction and conclusion, which should highlight how your separate papers contribute to an homogeneous research body (i.e. you're not only grabbing low-hanging fruits in minor papers, but you are able to have a coherent research agenda on the medium/long term).

But as others have answered, your thesis will be one of the key elements used to assess your qualities/skills as researcher in the first years of your career. So delivering something half-baked might not give the best impression.