That isn’t how you defines a mistake. To flip the logic, if you know something is a bad idea and do it anyway that’s not a mistake either, it’s intentional.

> if it was only known after the fact that it was a mistake, then i dont think you can call it a mistake

Mistakes are by definition only known to be mistakes after the fact. If you know before you do something that it’s going to fail then it is not a mistake.

>and cannot have blame placed on those who made the decisions under incomplete information.

That's ridiculous - everyone makes decisions under incomplete information - does that mean we can't be blamed for anything ?

Blame - yes.

But it was not a mistake - it was the outcome of a deliberate process.

A deliberate process that leads to an outcome you didn’t intend or want is one of the definitions of a mistake.

> an action or judgment that is misguided or wrong

Poker players are trained to think a decision was a mistake only if the expected value of the decision was negative, not if the outcome happened to be negative. This is good because it removes the chance component from the outcome when evaluating the quality of the decision making process.

Whether that’s a good definition or not of ‘mistake’, it’s extremely context specific. Considering the outcome of Abenomics and the generally used notion of ‘mistake’, still strikes me as valid to call it a mistaken program in retrospect.

It's context independent and can be applied anywhere. It's the maxim of not factoring in good or bad luck when evaluating the quality of a decision that's made under incomplete information (such as not being able to see the future). Simple and sensible.

> Considering the outcome of Abenomics and the generally used notion of ‘mistake’, still strikes me as valid to call it a mistaken program in retrospect.

Fair enough. I was commenting in the abstract, I have no view on Abenomics.

>Simple and sensible.

No, it only works in artificial environments where you get to sample a bunch of outcomes under the same decisions, with a known fixed probability distribution, to work out to an overall net positive.

Economic policy is anything but this, a wrong decision can have catastrophic events where you don't get to make more than one decision, probability distribution is extremely complex and ever changing, and you need to account for that. You definitely don't want to apply artificial game logic.

It can only be applied rigorously in artificial environments where distributions are known or where distributions can be easily sampled from.

But the mindset/maxim can and should be applied everywhere. If an economic policy had bad outcomes because of bad luck, and the unlikely adverse event was appropriately considered as a possibility when the decision was initially made, then the sheer bad luck shouldn't count against the decision when we are retrospectively evaluating it.

>If an economic policy had bad outcomes because of bad luck, and the unlikely adverse event was appropriately considered as a possibility when the decision was initially made, then the sheer bad luck shouldn't count against the decision when we are retrospectively evaluating it.

No, you need to weigh the severity of bad outcome. If you play Russian roulette for 100M - you don't say "excellent opportunity - 5/6 chances I walk away with 100M". No ammount of money is worth that kind of risk to me.

Of course, that's basic.

Anyway, economic decision makers do do this kind of scenario analysis, where unlikely possibilities are considered, using for example SDGE.

Good point. Still, I'd argue that changing a nationwide taxation system based on incomplete information is reckless (depending on incompleteness).

They could have tried it in a small area beforehand, instead.