Dissertation advice

August 4th, 2010  |  Tags: , , ,  |  Leave a comment

I’ve been blissfully out of the grad school game for so long that I might have to start teaching boxing to bright-eyed masters students to keep them from entering a life of research. But I was recently reminded of the best advice for anyone embarking on a substantial applied research effort (especially a dissertation): build your work around a thesis statement. That is, your main point should be an objective, defensible point that your experiments and prose will support. It should also fit on a single slide in 64-point type.1

This point seems obvious (after all, schoolchildren are instructed to begin constructing a five-paragraph essay with a thesis statement; why wouldn’t adults do the same for a 200-page monograph?), but it is surprising how many junior and senior grads can’t express the central claim that their dissertation is meant to argue in a single sentence. I first encountered this advice from Olin Shivers’ web site early in my graduate career; it was a big help in focusing my work, even as my own research veered more into analysis than into runtime support as I had originally intended. (Shivers is also responsible for the greatest acknowledgements section of all time, which I’ve linked to earlier.) I was reminded of Shivers’ advice this morning when I read a nice article by his student Matt Might (now at Utah) elaborating on the benefits of a thesis statement, in which Shivers’ influence is palpable.

1 This last point is where the thesis statement that appears in the text of my dissertation falls short. Although my thesis statement is one sentence, it includes a great deal of unnecessary detail about specific mechanisms rather than their essential properties (e.g. “type-based analysis” instead of “scalable analysis”).


August 2nd, 2008  |  Tags: , , , ,  |  Leave a comment


I recommend to everyone — but especially to my friends finishing dissertations, and doubly especially to those in Computer Science — Olin Shivers’ amazing acknowledgments section from the scsh manual, which I first encountered as a young Scheme nerd a long time ago. (Philip Greenspun’s gloss on Prof. Shivers’ acknowledgments is pretty delightful as well; scroll ahead to the second block quotation and prepare to be amazed.)


Speaking of acknowledgments, I make brief and jocular reference to the “preface paradox” in the draft preface of my dissertation. This is one of my favorite paradoxes (originally due to David C. Makinson). The basic idea is that a writer believes every individual claim in a manuscript is true (or else he or she would not have committed them to paper); however, some writers claim that their work inevitably will be found to contain some errors. As a consequence, writers are in the curious position of believing the conjunction of every claim in a book and believing the negation of the conjunction of every claim in a book. Whether or not this is irrational is — I guess — an open question with a few plausible solutions.