So when this question arose on Twitter:
I gave this answer:
Benefits for the Candidate
· You can, for at least one time in your life, tell the story of your scientific endeavours unconstrained by the petty limitations of a paper. No page limit; no limit on the number of figures; no savage truncation of the introduction and relegation of everything really useful to ‘supplementary information’; no limitations of ‘journal scope’ requiring you to break off and deform pieces of your multidisciplinary project to appeal to different journals. You can go as broad and as deep as you like and you don’t have to pander to the prejudices of the Editors.
· There is an expectation with a thesis that you will do all the writing yourself. When you publish a paper, this is not the case, and shouldn’t be. You do not want to miss out on all the accumulated writing talent of your co-authors, most of whom have done this many times before. The readers of the Kirghiz Journal of Analytical Chemistry deserve the best-written paper they can get. And that paper is not going to be a demonstration of your talents, unless you are rare and awesome. You should not do all the writing on a paper yourself. But, if you are like most people, you will need to practice writing a lot. And you will get a job where writing is an important skill - that on the balance of probability will not be in academia, but somewhere where nobody much cares if you have papers or not. So it is much more important for your future that you sit down and successfully tackle a big task of writing - and organising that writing - than that you contribute to a bunch of papers where the bulk of the writing, and certainly the bulk of the organisation of the writing, ought to be done by somebody else. Of course, you may be an awesome writer, and capable of writing great papers from year one of your PhD. Which brings me to the next point.
· The purpose of a thesis is to demonstrate that you are awesome: that you have mastered a field and made a valuable contribution to it. A thesis cobbled together from multi-author papers does not convincingly do this. Maybe you did design the project; maybe you did do all the work; maybe the stellar interpretation is yours. But if you did, what exactly did those three other authors contribute? Yes, you may write that Prof X contributed only 10% to paper Y; and maybe it is true; but your colleagues know Prof X, and know how she has carried weak students in the past, how she rules her group with an iron hand, how not a sparrow falls in her laboratory that she does not notice. I know of groups that will not consider applicants who have done a PhD by publication for a PostDoc, because of just this uncertainty hanging over how much of the work is really theirs. Maybe none of your PostDoc applications will end up in the inbox of a Professor who thinks like this. But maybe they will.
· If you are in a typical group, you will end up doing things that contribute to papers where the lead author is someone else. Where you are the person who – truly – has contributed 10% to a paper, you are going to look ridiculous binding that paper in your cobbled-together thesis. But that 10% may well perfectly legitimately be fit somewhere into the coherent narrative of your journey that you present as a traditional thesis.
Benefits for the User
· The theoretical purpose of a scientific paper is to communicate information, but the real purpose is to score points in the publishing game. So the false starts and dead ends are hidden, figures and procedures that should fit naturally into a flow of argument are hacked out and dumped in supplementary information, and the context and background of the work is stripped down to the bare minimum to save space. None of these things should be true of a thesis. A thesis should give the whole story, as a clear and coherent narrative with all the warts left in. And with today’s electronic repositories, it has never been easier to get hold of theses and consult them.
Benefits for the Examiner
· You can judge the candidate fairly. You don’t have to guess what is the student’s work and what isn’t. You are never in the position, as I once was, of giving a glowing report about the quality of a student’s work and then sitting through a talk at your next conference in which another student presents part of the same work as theirs. You don’t have to worry if the thing you think is important that has been left out has been left out to fit the requirements of the journal, or because the candidate is slack.
· You aren’t redundant. If you present a thesis made up of peer-reviewed papers that have been cobbled together, then for every correction or suggestion that you make, the candidate can simply point to their peer-reviewed papers and say; ‘the august reviewers at the Kirghiz Journal of Analytical Chemistry were happy with what I did, so nyah nyah.’ Presenting a thesis made up of published papers is a way to bullying you, as an examiner, into nodding and smiling.
· You can actually read the thing. Without flipping back and forth to the most important figure, which is lurking in the supplementary information. Without having to read through basically the same introduction again and again and again.
Which brings me to what is probably the real reason why I strongly discourage my students from submitting theses by publication. I sincerely believe the reasons I’ve given above about the benefits to a student of writing a traditional thesis. But ultimately, it comes down to this. A guy I greatly admire once said these words: Do unto others as you would have them do unto you. And I never, ever, ever, ever want to mark a thesis made up of papers cobbled together again. Ever. For the reasons given above. It is horrible. So I don’t want to inflict one on any other examiner.