Having now had a preliminary look at Levien's thesis, I see what you mean – several paragraphs in it could easily have been paraphrases of things in the above entry (or vice versa). He's taken the maths a lot further (and not been as scared as I was of solving complicated numerical problems), and he's got a much wider overview of the various problems that need balancing against each other (whereas all I saw was the one problem that had been getting in my way, and just solved that), but in spite of all that there are indeed recognisable similarities in attitude.

One notable difference in methodology between us is that he seems to like to draw his font outline the same way the eventual format specifies it, i.e. outlining the filled area so that you have to trace along both edges of a stroke. One of my first decisions was to avoid that (except, as usual, in cases where it turned out to be the right thing after all :-) in favour of using a single curve to define the centre line of the stroke and then separately specifying a varying nib. Doing it my way, it's easy to draw crossing lines (such as in the treble clef) and be confident that the portions of a stroke on each side of the other will line up as if they were two parts of the same sensible-looking curve, because they are. Doing it his way, you have to get that right by eye. Also, even in the absence of crossing strokes, I found it difficult to draw two nearly-parallel curves in a way that made them vary from parallelness just right with respect to each other.

Then again, I've looked at some of his actual font designs, and they seem to have stroke widths which vary more subtly than mine in a way that looks deliberate, so it seems entirely plausible to me that his method is better suited to his level of skill while mine is better suited to mine!

Still, all very impressive, and I'll probably finish reading his thesis at some point in the near future. Just a shame he decided to apply for patents.