To start with, I'll pick a book I read a few weeks back - The Grand Strategy of the Roman Empire from the First Century AD to the Third by Edward Luttwak.
Summary:Empires need to defend against two separate kinds of threat - raiding parties and invasions. Raiding parties are best dealt with by having numerous small garrisons on the border, while invasions are best dealt with by large, concentrated forces. In the ancient world, strategic mobility was extremely poor, particularly in an empire as large as Rome, so the natural tension of these deployment needs was exacerbated.
The core of Rome's strength was professional legions. Rome was wealthy but chronically short of manpower, so they maximized force per soldier. Elite heavy infantry, supported by allied auxiliary troops, was the best structure for this need - it gave the flexibility to use a combined-arms force, but the legions formed a necessary core for these forces. Legions plus auxiliaries could generally defeat foreign forces in pitched battles, and if the auxiliaries revolted, legions could generally defeat auxiliaries in pitched battles. Roles besides battles were secondary - it was assumed that guerrilla wars could be easily ended by reprisals or hunting down raiders if the other army was defeated.
The first system Rome used was based on client states. Subject rulers were kept in place on borders to keep out raiders, which allowed the legions to be concentrated in large blocks for maximal concentrated power against invasions. This power could then be used as a threat against any nearby enemies. Since this threat was usually sufficient to overawe enemies, the power was not used, and losses were avoided. This allowed the same forces to hold multiple threats in check simultaneously, which multiplied their effective power substantially.
Over time, having several legions in a single place became a threat to emperors - the commanding general could too easily attempt to crown himself. Management of subject kings was also a chronically troublesome task, and keeping them both powerful enough to be useful and weak enough to be subjects was sometimes beyond Rome's capacity. Subjects were thus assimilated, and in their place Roman troops were dispersed into new border infrastructure. This was still a forward defense, focused on fighting battles on enemy territory, but it lacked the efficiency of the old system.
The crisis of the third century left Rome without the power to keep this system intact. Instead, they began defending reactively, instead of being able to defend proactively. This encouraged further fortification of the border, and further inefficient dispersal of troops. Rome's weakness and the spread of their technology to neighbours also left the legions less relatively powerful than they had been, and Rome's primary advantage over their enemies became logistics more than sheer force. Fortifications, even of things like granaries, served both as defensive strongpoints and to deny attackers loot. These became integral to Rome's defences. This meant that civilians in border lands took the brunt of any attacks, and it sapped both their loyalty and their productivity.
This third system lacked any particular virtues except postponing collapse, and unless a strong Emperor reformed the system totally, Rome's decline was inevitable.
Commentary:Luttwak is fascinating, but also kind of a jerk. He's clearly a passionate believer in his ideas, he's done some excellent research, and his criticisms of those who disagree with him frequently hit home, but he is grossly overconfident. He raises many valid points, and his overall thesis does seem to fit together pretty well, but my first impression when reading a book like this is to take it as an interesting perspective, but to get another perspective before I feel like I really know anything. Sadly, I don't know many authors who write on topics like this to even get another solid perspective from. (Any good ones you know of, please share). This is basically the same impression I had when I read another of his books(Coup d'Etat: A Practical Handbook), and seems to be the same impression most other reviewers have as well.
As a side note, the impression he gives of the prevailing wisdom of the historical community at the time this book was published(1976) was quite unflattering, both in their impressions of the Romans and in my impressions of them. I tend to take as granted that people were basically the same in all recorded human cultures, and he talks as though Roman emperors were generally thought to be fools who lacked even basic tools of statecraft - apparently, the idea of mapping the empire's borders(even so much as a mental understanding of how they looked) was frequently thought to be beyond the capacity of the empire. I don't buy it. If that was actually how the historians thought, they were parochial fools, but I am not conversant enough with the history of historians to know if Luttwak is treating them fairly.
All that said, I liked it. Even if he's wrong about bits and pieces, he's a hell of an interesting writer, and he approaches things in a way that most others don't.