Friday, August 27, 2010

I Wasn't Wrong!

It's reality that was inaccurate:

"Sung Won Sohn, economics professor at Cal State University Channel Islands... estimates that the risk of a double dip has risen to 40% from only a 25% chance at the start of the year."


  1. I don't get it. I agree that economists make these numbers up and don't realize that's what they're doing, but you could come up with an analogous situation of a guy estimating the odds of getting four heads in a row with a fair coin.

    After he gets three heads in a row, he could say, "The probability went from 1/8 at the start, to now being 1/2."

    What's wrong with that?

  2. Oops I should have said it went from 1/16 to 1/2.

  3. Well, Bob, this is mostly a light-hearted post, but my point is that the passage seems to be treating "the risk" as if it were a real thing "out there" that had really changed, and not just a number being spit out of a model.

    In a case of probability where the frequency interpretation can be applied (like coin flipping), there is some sense in claiming there is an "objective risk" -- even there I'm doubtful, but let's not get into THAT -- but in much more individual cases, like large historical events, one really can't say that sensibly at all. With the coin flips, you could keep re-rerunning the cases, and in the limit you'd get your odds -- so maybe they're "out there." But how is this economist going to do that -- he's going to re-run 2010 thousands of times, and show the situation at the beginning of the year would double dip 25% of the time, but once it had evolved to "this" situation, it would double-dip 40%? And what does it even mean we are in "this" situation again? It only makes sense describing his model, not the real world.

    That's what led Mises to reject case probability altogether. But I think it's better to look at case probability subjectively. If the guy said, "At the beginning of the year, I would have given you 3-1 odds no double dip, but now I'd only give you 6-4," I'd be fine. But he claims (or perhaps the reported claim that Sohn claims) to be estimating something called "the risk," and "estimates" implies that is an objective "thing." (We don't say, "I estimate that I am very angry" but we do say, "I estimate its 100 yards to the green.")

    Now, I really have no idea if Sohn thinks of "risk" in that fashion, but the reporter reports it as if he does, and, like I said, I was just making a wee joke, and not trying to write a scholarly critique of Sohn's work.

  4. And, of course, we could do axiomatic probability a la Kolmogorov, but then the probabilities are deductive consequences of the axioms, and not properties of the world.

  5. Even the coin flip only involves ~50/50 odds from the standpoint of a human observer. In principle, a super-computer with all the right data could use the laws of physics to predict its trajectory.

  6. Mistakes:

    "Now, I really have no idea if Sohn thinks of "risk" in that fashion, but the reporter reports it as if he does"

    Should be "or" not "but."

    Previous par. should be "reporter" not "reported."

    PSH: Yes, we may really have "true" probability only in quantum mechanics.

  7. I agree with you (I think), in that it would be more accurate for Sohn to say that "The risk has risen" instead of "I estimate that the risk has risen". This is because a statement about a risk is necessarily an expression of your confidence in an estimate. So the latter is going "one meta-level too high", and you are correct that reveals a failure to recognize a map-territory distinction.

    But I can't go so far as to take this model-centeredness as a reason to reject assigning case probabilities. People bet on things like "Obama winning his next election". Must I regard the bookie numbers as saying only things about models, and not about real out-there events?

  8. Well, Silas, I think that "the chances of Obama winning the election" says SOMETHING about the world out there, in that it describes how the world looks to the oddsmaker of the average bettor. And that "how it looks" may be based on very objective things.

    On the other hand, I don't think anything in the world actually contains some "chance" of tumbling this way or that way. The "odds" are a human construct, a statement of what sort of return we'd need to feel good about placing the bet.


Killing the Spirit

"The more fervently all human energies are thrown into the great enterprise of salvation through world -- immanent action, the farther...