I like the insight that bureaucracies are composed of agents who are too incompetent or misaligned to be trusted with actual power. Starting with the stipulation that they’re frustratingly obstructionist BY DESIGN is very helpful.
I think the next section vastly oversimplifies by assigning a few boolean dimensions (owned/abandoned, effective/ineffective), when in fact all organizations have elements in all 4 quadrants, and both dimensions have degrees that matter, not just on/off.
I think this would benefit by recognizing that this is mostly an specific case of the principal-agent problem, and it doesn’t matter too much whether the agent is an individual, an automaton, or a collective bureaucracy.
I like the point about the principal-agent problem. I disagree only slightly, and think it would be worth distinguishing between individuals and automatons on one side, and a collective bureaucracy on the other: this is because the bureaucracy is in fact a chain of principal-agent links. It seems to me a big reason that any kind of organizational management is hard is because whatever anyone tries to do, it will be mediated by many principal-agent steps, with the predictably accumulated error.
I have just realized I never considered the possibility of an automaton (software, say) as an independent link in the chain. I don’t know how correct it is, but it sure makes a lot of the problems I encounter on a daily basis more understandable.
I like the insight that bureaucracies are composed of agents who are too incompetent or misaligned to be trusted with actual power. Starting with the stipulation that they’re frustratingly obstructionist BY DESIGN is very helpful.
I think the next section vastly oversimplifies by assigning a few boolean dimensions (owned/abandoned, effective/ineffective), when in fact all organizations have elements in all 4 quadrants, and both dimensions have degrees that matter, not just on/off.
I think this would benefit by recognizing that this is mostly an specific case of the principal-agent problem, and it doesn’t matter too much whether the agent is an individual, an automaton, or a collective bureaucracy.
I like the point about the principal-agent problem. I disagree only slightly, and think it would be worth distinguishing between individuals and automatons on one side, and a collective bureaucracy on the other: this is because the bureaucracy is in fact a chain of principal-agent links. It seems to me a big reason that any kind of organizational management is hard is because whatever anyone tries to do, it will be mediated by many principal-agent steps, with the predictably accumulated error.
I have just realized I never considered the possibility of an automaton (software, say) as an independent link in the chain. I don’t know how correct it is, but it sure makes a lot of the problems I encounter on a daily basis more understandable.