User:Jadrian Miles/Thesis proposal feedback/jfh email
First of all, congratulations again. I think that your talk was a great improvement over the draft I commented on, and that you and I discussed.
That said, I want to give you some critical feedback on your proposal, in no particular order.
1. Prepare better. Two or three times during the talk, you came upon words that you needed to say and which you didn't know how to pronounce; once you came upon some reference (I think) that you couldn't remember/cite correctly.
You're supposed to be showing us that you're prepared to be one of the world experts in this field. Failing to get the little stuff right REALLY weakens your case. Failing to treat related work with respect is just bad policy --- the person who did the work, or his/her advisor, etc., might be in the audience, or might be the one reviewing your grant proposal.
2. Go ahead and number your slides. It's a lot easier for the questioner to say "can you go back to slide 28?" than "can you go back a few slides...no...a few more...no, move forward two...no...back one...THAT's it!"
3. Less marketing, more substance. You told us about things like your plan for which articles to submit when, which is a great way for you and your advisor to set benchmarks, but pretty much irrelevant for the rest of us. But you didn't give a single detailed example of a micro and macro model of a phenomenon and how coupling them could resolve ambiguity. I know you're planning to do this for the brain data, but you could, for instance, have applied the idea to an out-of-the-box optical flow computation, or something else simple like that, just to show (a) how it can be done in practice, and (b) what SORTs of tools/equations/solvers/optimizers might come into play in completing your work.
4. When you write mathematics, be precise. Your "energy" function had the appearance of being a product of terms over another term, when you really meant something like "I'll craft a function that increases as a function of these things, and decreases as a function of that thing"; addition and subtraction might have worked just as well as multiplication/division, and might many other combinations. One of the challenges I expect you to encounter is determining a principled approach for building the energy function.
5. When you give a talk, you're very rarely bringing wisdom to the people you're talking to. For some, you're telling them about a thing they don't know, but that's information, not wisdom; others may know the subject better than you; others may know some OTHER subject in which related ideas have already been well worked out. So while a talk, from the audience's point of view, is a chance to learn something new (even if it's only "other folks in my field don't know as much as I do...whew!"), from YOUR point of view, a talk is a chance to get a bunch of potentially useful thoughts from a somewhat diverse crowd. Every question you get is a chance to have a dialog in which you and the questioner both learn something. Perhaps all you learn is "I didn't say that last thing very well, because my explanation confused this guy; I'll have to say it differently next time." But USUALLY you can learn something more. Unfortunately, you had a tendency to dismiss the concerns people raised rather than engaging with them. We were reasonably polite, but if you encounter someone pushy, you'd better be damned sure that you can back up your dismissal with substance, or you'll look like a fraud. It's actually BETTER to look like someone with *some* good ideas, but perhaps not all of them (i.e., to say "I don't know and haven't thought about that; I'd like to discuss it further with you afterwards toyou’re your ideas on the topic.") See http://www.xkcd.com/803/. :)
--Spike