Not a lot of time now, but a few quick thoughts:
1) Chapters 1-5 explore the motives behind a lot of do-gooderism, and especially a desire for "packaged interventions." Everyone is looking for a "best practice" that is affordable, scalable, and can be implemented by idiots. That's great if you're trying to open as many burger restaurants as possible, but not so great if your goal is to help people do smarter things and do them better. There seems to be a desire to find an intervention that you can pay for once and them BOOM! World fixed!
2) This is the excuse of every person who fails to deliver, but it's also true: Many of the most important aspects of human behavior and human learning are not amenable to easy metrics.
3) On page 161, Toyama makes an important point about the craze for "nudges" and (as Cass Sunstein calls it) "libertarian paternalism." It's the idea that very small, minimally intrusive interventions can have huge effects if you are smart about human psychology. On one level it's unobjectionable: If changing a couple of words on a form helps people make better choices, well, great. But, on another level, I suspect that nudges are most powerful when people don't know they're being nudged, and if nudges became more common we'd get a response.
More importantly, and in keeping with the previous book that I blogged about, the whole theory of nudges is based on the idea that people are full of irrational little switches that are easy to flip because they don't perceive what's going on. I'm not sure that that's as true as people think.
4) Finally, about the book itself: The second half is more about anecdotes, but the anecdotes are used well. Toyama doesn't try to say "See, here's this anecdote, where these people just did this one thing, and it was all great!" It's much more "Here's an example of how success came down to a lot more than just pouring in the right technology and pre-packaged tools." Nonetheless, a lot of it is stuff that right-thinking people have heard or read before, so I've been skimming.