Last updated on 5 August 2000.
If you have ever assembled a challenging jigsaw puzzle then you are fully prepared for graduate work in Computer Science. Or rather, as prepared as one can ever be for such a thing. Please allow me to explain. Your objective as a researcher is to produce an artifact that your peers will regard as interesting. The suggestion here, of course, is to think of that artifact as a jigsaw puzzle. The origin of the pieces and the process of their assembly, then, correspond to important aspects of graduate work. The analogy naturally has its flaws, but I have found this way of thinking to be very instructive at times.
Before you can begin assembling the Masters puzzle, you need a supervisor. This is because your supervisor will have all of the pieces, in a box in his or her office. Or, as is often the case, they will be scattered under tables and behind filing cabinets so that you may hunt for them to pass time while your supervisor is on the phone.
Wherever the pieces may be, the point is that your supervisor has them. Supervisors like building puzzles, and the more talented ones often work on several puzzles at once. But they don't like doing the hard parts -- the panorama of autumn forest under a vast, cloudless sky -- that's what graduate students are for. So, when you first approach potential supervisors, they will show you the puzzles they are working on. Some will be far enough along to be instantly recognizable, others might be in the initial stages and appear mysterious or even confusing. They will also have little piles of pieces here and there, roughly sorted by colour. They will show you a pile of blue pieces, for example, and say, ``I think this is sky -- how would you like to confirm it?''
It is important to remember that there are a variety of ways you can help your supervisor with a puzzle -- if sky really isn't your thing, you needn't torture yourself. There are certainly the dull bits that can be completed with only modest intellectual effort, backed by sufficient tenacity. But there is also the need to sort the pieces to begin with, which may require a bit of experimentation and the devising of a good taxonomy. Once several small sections have been assembled in isolation it will be necessary to determine how they must be positioned with respect to each other, and to join them together by locating and fitting the connecting pieces.
An important fact I have as yet neglected to mention is that these puzzles don't come in boxes, so you will have no picture of what it should look like when you're finished. Your supervisor will often be able to show you a fairly accurate sketch, but it will be lacking in many details. Improving this sketch is one of the harder aspects of puzzle making. If after examining a particular incomplete puzzle you believe you can figure out what one of its sections should ultimately contain, you should discuss this with the potential supervisor. Chances are that he or she will be immensely pleased and will buy you lunch (what those chances actually are I will leave to your imagination). Finally, it often happens that others are working on similar puzzles. So you can often gain considerably insight into your supervisor's favourite puzzle, especially in the early stages, by carefully reviewing the efforts of other researchers in that area. You may find that entire sections of the puzzle have already been worked out, or, barring that, find some sensible suggestions for sorting out and assembling the pieces you have at hand.
If you find that you really enjoy making puzzles, then you should move on to a PhD. PhD students get to make their own puzzles.
Most new PhD students have endless questions about the search for a topic, the production of a thesis proposal, the content that will ultimately be required for their thesis, and so on. The obvious response: think puzzles.
You get do this puzzle on your own, but you will need lots of help. So don't select a coral reef if your supervisor has only ever made puzzles of canoes on mountain lakes in autumn. If you're lucky, your research group will be in the long process of building a gigantic wall mural type puzzle and you will be able to claim a portion big enough to contain several interesting sections. This way you can work on a chunk the size of a normal puzzle, but have lots of people around who are good at doing exactly that sort of puzzle. It will also greatly increase the chances that your work will not be abandoned as soon as it is finished. And remember that this is research, not product development -- there are no edge pieces.
Having selected your area, you will quickly find that there are far more pieces than could possibly fit into one puzzle. Beginning with a huge assortment of pieces, you need to figure out how many different puzzles are contained within them and which ones are potentially interesting. It will be entirely unclear whether you have all the pieces necessary for any given candidate puzzle or whether the completed puzzle will contain a childish 25 pieces, or an overwhelming 2000. And of course, you don't have the final picture for any of them.
A thesis proposal must convince a supervisory committee of five things:
Ideally, a good thesis proposal should mark the end of the scary part. Considerable work will remain to flesh out the skeleton, but if you have done a good job sketching the basic outline and identifying the relevant pieces, then it only remains to proceed in a sensible fashion to fit the pieces into likely places. If you are an avid puzzle maker you will know that even when the placement of a thousand identical sky pieces seems hopeless, perseverance will prevail; and that just when it seems that your plodding has encountered eternity, the final pieces will be placed with astonishing speed.
As you near completion you will find that some pieces simply don't fit, and you may even have entire sections that belong to other puzzles. Be prepared to discard them without emotion -- if they don't fit, they don't fit. You may need to adjust significant portions of your original sketch to accommodate. There will also be pieces you simply can't find. You can only hope that these will not be numerous. And remember always that there are no edge pieces: as far as possible keep your original sketch in mind, resisting the temptation to build indefinitely in one particular direction.
When it becomes clear that you have assembled your puzzle correctly, to go further is an option that you need pursue only if you are eager to see the final product with every piece in place, thickly lacquered and hanging on your wall. It will have value as a research artifact if it is sufficiently complete to guide others in the construction of similar puzzles, or to adequately fill a hole in that large mural.