Zvi depends on Twitter for a lot of analysis, which is a really serious Goodhart’s law situation, and it resulted in him being really wrong about Goodhart-intensive areas like predicting that the Shanghai lockdown would fail. Again, his analysis is still generally top-notch but it’s far from perfect. It’s more about outperforming large popular news outlets by as much as possible, which he does plenty of. But international affairs is a different beast entirely when it comes to systematic error.
Economic priorities can be considered more important than public health, especially with Omicron and variants, and even more so during a major war, and yet even more so when both sides deliberately use the threat of unpredictably spiralling economic crisis to erode the opposite side’s consensus to maintain force. Persuading people to go to work, instead of not going to work (e.g. due to fear of brain damage), seems to be a good idea that powerful people can rally behind right now.
I never stated that the military is a monolithic entity, it is consensus-driven depending on whatever consensus is possible in the context of compartmentalization e.g. what actions seem obvious to a lot of different people. Governments too corrupt to occasionally rally behind a goal have not been a dominant force in global politics for centuries. Any sort of effective central command structure is anti-inductive to study because it’s built to fool foreign intelligence agencies and domestic business interests. And, obviously, partisan gridlock is usually not the dominant factor influencing core national security priorities.
It is also really difficult to disentangle ordinary bureaucratic actions with military operations carried out through bureaucracies. This is actually extremely common. Among other things, it helps protect military agencies from blame since it’s really important that they remain as popular as possible, for several reasons.
The “pure incompetence” model appears extremely naive compared to the “mixed competence-incompetence” model, and the CDC’s lack of Disease Control and Prevention is only a single data point in a large-n system; even if that was entirely a convulsion, which I never ruled out in the CDC’s case.
Zvi depends on Twitter for a lot of analysis, which is a really serious Goodhart’s law situation, and it resulted in him being really wrong about Goodhart-intensive areas like predicting that the Shanghai lockdown would fail. Again, his analysis is still generally top-notch but it’s far from perfect. It’s more about outperforming large popular news outlets by as much as possible, which he does plenty of. But international affairs is a different beast entirely when it comes to systematic error.
Economic priorities can be considered more important than public health, especially with Omicron and variants, and even more so during a major war, and yet even more so when both sides deliberately use the threat of unpredictably spiralling economic crisis to erode the opposite side’s consensus to maintain force. Persuading people to go to work, instead of not going to work (e.g. due to fear of brain damage), seems to be a good idea that powerful people can rally behind right now.
I never stated that the military is a monolithic entity, it is consensus-driven depending on whatever consensus is possible in the context of compartmentalization e.g. what actions seem obvious to a lot of different people. Governments too corrupt to occasionally rally behind a goal have not been a dominant force in global politics for centuries. Any sort of effective central command structure is anti-inductive to study because it’s built to fool foreign intelligence agencies and domestic business interests. And, obviously, partisan gridlock is usually not the dominant factor influencing core national security priorities.
It is also really difficult to disentangle ordinary bureaucratic actions with military operations carried out through bureaucracies. This is actually extremely common. Among other things, it helps protect military agencies from blame since it’s really important that they remain as popular as possible, for several reasons.
The “pure incompetence” model appears extremely naive compared to the “mixed competence-incompetence” model, and the CDC’s lack of Disease Control and Prevention is only a single data point in a large-n system; even if that was entirely a convulsion, which I never ruled out in the CDC’s case.