This is a book that can change how you understand and interpret the world. At least, I think it did this for me.
I discern at least two insights from my reading of it. One relates to the concept of legibility, the other to metis.
Communities in the old world (Europe before enlightenment, the third world before colonisation, etc.) would often be organised around local ways of doing things, established via oral tradition, and, while tradition-bound, also constantly adjusting to changes in the environment. Even measurements, such as of length or volume, would vary by region. Land ownership would often be complicated and unformalised.
This makes taxation and conscription difficult, and to the degree that it's even possible, a strictly local affair. This explains the feudal nature of medieval Europe. A king can only indirectly levy taxes by imposing taxation on his vassals. It's then up to the vassals to tax their subordinates, and so on. Seen from a central authority, this is both inefficient and lacks control.
For any central authority wishing to cut the middlemen, a more legible land is necessary. An absolute king needs to be able to directly count the number of able-bodied men available for conscription, and the amount of grain etc. produced each year, in order to be able to tax his subjects. This leads to land reforms and censuses. It also forces a degree of uniformity over a society that may not fit local conditions, but it does introduce a (partially illusory) degree of legibility of society.
This way of understanding the modern state is illuminating, and, in my opinion, transfers well to large corporations. Particularly those corporations steeped in Taylorism suffer the same blind spots as modern states. Quantification and uniformity may work if one produces timber or model A Fords, but causes much friction in modern knowledge-based organisations.
Here, Scott's concept of metis emphasises practical know-how. This describes a kind of (often tacit) knowledge that can only be gained by practice, often via apprenticeships. This type of knowledge is contrasted to epistemic knowledge, decoupled from practical, local concerns.
The idea of metis particularly rings true for me as a programmer. While there's a strong desire (particularly among management) to see software development as a legible, controllable process, the reality seems to me to better fit the concept of metis. A good programmer has, over many years, developed a set of heuristics that may or may not apply in a given situation. There seems to be few universal, Taylorist processes that make software development more predictable or controllable.
The analysis in the book is both compelling and enlightening. I definitely feel that it's going to change the way I think about certain things. I also, however, have problems with it.
First of all, I find it too long. It's possible that I missed a level of depth in it, but what I did get out of it, I could have gained in half the pages. The point about legibility was evident to me after, say, 150 pages, and the point about metis was made over a single 30-page chapter toward the end.
At times, the book veers off into territory that seems to me to be completely irrelevant, such as the advantages of polycropping and swidden agriculture over western monocropping. I get the point, and I don't even dispute it, but the book just goes on and on about this.
Another concern I have is with the text itself. It's always academic and dense, but while it's sometimes surprisingly easy to read, at other times it becomes 'illegible'. I found Part two quite readable, particularly the chapter on the high-modernist city, while the chapters on villagisation in Tanzania and on taming nature was far less readable.
Finally, the way the book's notes are organised is really annoying. Each chapter is equipped with about a hundred notes, and while many of them are references to literature, many others constitute additions and comments on the text itself. I'd have no problem with this if these were footnotes, but alas, they're end notes. This requires the reader to flip back and forth between the chapter and the end notes, often to find that the note was nothing but a literature reference. That's just not a good reading experience.