Geoff, I think I get what you mean in your similarly worded similar conclusion! Ok, I'm assuming a linear increase of 2# per inch. With a short (but sensible) brace height let's assume the first inch pulls 10# and by the end of the draw we've pulled through a 20" power stroke and ended up at 50#. Then if we increase brace height by an inch, we start at 12# and end at 52# but the power stroke is only 19" *this doesn't quite add up, even on paper, but it's close enough for this example, I think* So, we haven't just lost 2# at the beginning and added it on again at the end, we've traded 10# at the beginning for 52# at the end, therefore the energy stored in the bow at full draw is 42# more, and on release all we're losing off the power stroke is that final inch were there is only 10# acting on the arrow anyway. I can't explain mathematically, but my gut feeling says it can't quite work as simple as that...
Consider if we plot a graph of brace height v energy of the arrow. I imagine at excessively low brace height the graph starts off low, because we aren't bending the limbs enough to store much energy. Then it rises. Then we get to excessively high brace heights and it falls again because we haven't got enough power stroke to impart the bows potential energy into the arrow. I don't think the graph goes up to a sharp point, like a triangle. I think it's a curve which fattens out in the sensible brace height range. Yes I suppose there has to be a peak, I'm just not sure it's a distinct one. We need a volunteer with a chrono, a hooter shooter, and a lot of free time, to test this out for us!
My interest in this topic (I rarely post on forums) comes from having long arms and short bows. I like short brace heights because, with a longer string, I believe I'm being kinder to my bows by not bending the limbs so much, and my back-of-a-fag-packet calculations say I'm not really losing any performance by doing so.