Economists could calculate error bars that would say how closely the calculated aggregate figures approximate their exact values according to definitions. This is normally not done, and as Morgenstern noted in the book discussed elsewhere in the thread, the results would be quite embarrassing, since they'd show that economists regularly talk about changes in the second, third, or even fourth significant digit of numbers whose error bars are well into double-digit percentages.
However, when it comes to the more essential point I've been making, error bars wouldn't make any sense, since the problem is that there is no true value out there in the first place, just different arbitrary conventions that yield different results, neither of which is more "true" than the others.
Economists could calculate error bars that would say how closely the aggregate figures approximate the exact value as defined. This is normally not done, and as Morgenstern noted in the book discussed elsewhere in the thread, the results would be quite embarrassing, since they'd show that economists regularly talk about changes in the second, third, or even fourth significant digit of numbers whose error bars are well into double-digit percentages.
There's an old joke: "How can you tell macroeconomists have a sense of humour? They use decimal point...
This thread is for the discussion of Less Wrong topics that have not appeared in recent posts. If a discussion gets unwieldy, celebrate by turning it into a top-level post.