PMJ: I have always thought that a reasonable strategy is to get to halfway at a slightly faster pace than you think you can run for the whole race. Last year I was obviously a lot fitter than I thought; I wasn't sure if I could manage sub 3 (thought I was likely to run 3:02) but wanted to keep my options open, so planned to get to halfway in 1:29, and that would allow for a slight fade in the second half to get the 3:02, or if I managed to pace it evenly I could get my sub 3.
This year I thought I would run 2:52, and therefore that it would not be insane to get to halfway in 1:25. So exactly the same strategy, but with faster paces. I wasn't expecting to run 2:50 as I didn't think recent training or races indicated that, but I still believed that running at a slightly too fast - but not suicidal - pace for the first half was a good strategy (I did slightly overcook this, as you point out).
As it turned out, I think I was right: I ran more or less exactly the time I expected, but with the planned-for fade near the end. I didn't 'hit the wall', but I did get tired and slow down. If I'd been more cautious in the first half, I suspect I might have run maybe 30 seconds faster overall at the most, but maybe not. I'd have had a pleasanter race, but I would have finished wondering if I could have squeezed any more out of it.
So in conclusion, I think this confirms what I have always thought: when people say 'one second per mile too fast at the start = one minute slower at the end', I think that is bollocks. I think there's a range that's not stupid, and if you don't try to be a little bit (but not crazily) ambitious you might not meet your potential.