[LarrySDonald]

I think it's partly because so many of us grew up basically hating him for the software we had to work under (and before it's brought up, no, no gun to the head, but usually there were few options and for the most part you work with what the boss tells you to work with), rightly or wrongly.

It's also partly because there are some flaws in his methods here and there and it isn't perfect. That's more of a reason stemming from a complete disconnect from how charity projects usually work. It's very rare to see any metric, useful or otherwise, nor any consideration toward "Ok, so looking at the data, is there a better approach?". Most take a kind of "field of dreams" approach and at best confirm that they're doing more good than harm and leave it at that, rather than heavily crunch into how much good it's doing or isn't doing, how well it works compared to other approaches, what could be learned and applied to future efforts including area differences, and so forth. That's terrible - just because something is better than nothing is no excuse to not attempt to optimize it. Yet it never seems to happen outside of medicine (a field used to living and dying on it's data) and even then often taking a rather naive approach and considering improvements in cases rather than a more bulk-medicine approach similar to military medicine (people are dying constantly at a high rate, so the problem isn't so much how well an individual treatment works as how well a treatment, applied shotgun over a group, works considering it's cost).