The principal idea that the author is attempting to communicate here, is cleverly communicated by the title. Or rather, by the fact that the title itself does not mean what you think it does. By "Edge of Chaos", Ramalingam does not mean only a regrettable and disordered situation. Rather, he is referring to the mathematical concept of unpredictable nonlinearity, called "chaos". One of the most fundamental characteristics of that kind of (somewhat) precisely defined "chaos" is that it is inherently unpredictable. You cannot do a little of something, see how that worked, and then predict with confidence that doing more of that same something will work better (or work at all).
This is because human society has many, many forms of nonlinear feedback. If you deliver aid (of whatever form) to a society in need of it, you are not outside of the system, delivering an impulse of some sort, the result of which can be measured with Newtonian objectivity. Rather, you are becoming part of the system; the other parts notice that you delivered aid (of whatever form), consider that you might deliver more of it, and therefore (in the way of humans) consider whether or not that is something that would work out well for them.
If it seems to be so, they will likely consider how to bring about the delivery of more of it. If it seems that it may jeopardize their position in some way, they may take action to prevent this. In either case, the response may be quite a bit different than expected, different than the last time you did it, different from one society to another, etc. etc. Yet, because chaos (in the mathematical sense) is difficult to think about, aid agencies are inherently biased against admitting that they cannot measure in some way the benefit which their aid has brought about.
Ramalingam does a good job of illustrating all of this with real world examples, some from his own experience and some from well-documented aid programs around the world. He does a decent job, also, of surveying the work that has been done in academia, of trying to develop a theoretical framework for thinking about complexity (a closely related concept to chaos). His writing is good, his enthusiasm for the subject(s) is palpable, and he does try to tie the two threads together.
There are also a lot of cartoony illustrations, that help to shine the light of (occasionally harsh) humor on whatever topic is being discussed. Taking an example at random: it shows a series of people, each talking to the next in a series from left to right:
[man at computer] "Food prices are rising!"
[man with report in hand] "Okay, what do we do?"
[woman with headdress of some kind] "Humanitarian aid is on the way!"
[man with tie] "But the campaign releases aren't ready"
[next man with tie, and phone] "What campaign? We need proper data."
[third man with tie] "Stop! Wait! What are the donors saying?"
[fourth man with tie, looking down at tombstone that says "RIP"] "Oops, too late."
If I have any criticism of Ramalingam, it is that he does a better job of saying why the old ways are not competent to handle the messy realities of the areas which we are trying to help, but he isn't very good at suggesting what should be done instead. This may be because the very idea of using external, aid-dispensing bureaucracies to help people in crisis in a part of the world they know little about, is what's flawed, and if you want to make a living in the aid industry this is more or less the only place to do it. Or, it may be that saying what to do to make aid actually effective is inherently difficult, in a messy world.
It may also be that we know enough to understand why our current methods aren't working very well (at all?), but we simply don't know enough to understand what to do instead. There are some estimates that the international aid industry is, on balance, not only not as good as we might wish it to be, but perhaps not even a net positive influence (although such things are difficult to measure, which is part of the problem).
Sometimes, though, one must admit that the current crop of ideas is not working, before better methods can be found. Ramalingam's book is a bit frustrating at times, but definitely worth reading, and worth thinking about afterwards.