Of many, here are two of recent interest:
1) Gibbon's pointing to the increasing power of the Praetorian Guard in the Roman Empire leading to its decline. You might say this is a particularly virulent case of the kinds of special interests internally distorting the priorities of an empire (or over-reactions to those interests), like eunuchs in Ming China or the shepherd lobby in colonial era Spain; particularly virulent, because when the lobby is the people who know how to kill people, their demands are more difficult to ignore, resulting in few natural deaths for emperors.
2) The recognition by both Gibbon and Machiavelli that there may have been a connection, fortuitous though it was, between non-violent succession during the non-inherited Five Good Emperors period, and the prosperity at that time in Rome; hence the installation of regular peaceful power transfers. Good institutions are indeed important.
Relating to #2 above, I would have liked to find some quantitative work with nice graphs on the damage to the Roman economy by civil unrest or governmental inefficiency during turnover/civil wars, but I couldn't find any; there's this article by Bruce Bartlett which is mostly about the damage done by the Roman deep state (deep for that time). Suffice it to say, the Julio-Claudian dynasty marked a period of stability after the civil war between Augustus and Antony, the Five Emperors were a period of peaceful succession, and the Crisis of the Third Century was the opposite. (Note how Bartlett is gunning for the deep state and he skips the Crisis entirely.) Economies grow when credit and capital are available, which requires the world to be predictable. (In other words: risk is okay as long as it can be evaluated.) The Pax Romana and Mongolica are good examples of this; civil wars are obviously bad, but expropriation/nationalization events by otherwise stable states can do just as much damage, because they concentrate the unpredictability in the safety of capital.
The politics of crime-fighting software
3 hours ago