Sunday, July 19, 2015

Stupid Economist Tricks


Deeply annoying featurette from the Times Upshot:

A Quick Puzzle to Test
Your Problem Solving

Now it’s your turn. Enter a number sequence in the boxes below, and we’ll tell you whether it satisfies the rule or not. You can test as many sequences as you want.

Your guesses:

So you type in, say:
3, 6, 12
And the thing replies
So then you're asked to type in your version of what the rule is, given only a single chance to get it right, and you type in, say,
each term is 2 times the previous term
and submit, and then old Leonhardt totally ignores what you typed, even though he's shared his byline with you.
The answer was extremely basic. The rule was simply: Each number must be larger than the one before it. 5, 10, 20 satisfies the rule, as does 1, 2, 3 and -17, 14.6, 845. Children in kindergarten can understand this rule.
But most people start off with the incorrect assumption that if we’re asking them to solve a problem, it must be a somewhat tricky problem. They come up with a theory for what the answer is, like: Each number is double the previous number. And then they make a classic psychological mistake.
I call shenanigans. My answer satisfies the problem just as well as his does.

As does the one allowing as correct
1, 5, 8
6, 9, 7
10, 0, 3
or, but only if you're speaking Italian,
3, 8, 7
(If you can't guess the rule for this one, click here.)

Leonhardt writes:
They don’t want to hear the answer “no.” In fact, it may not occur to them to ask a question that may yield a no.
Remarkably, 78 percent of people who have played this game so far have guessed the answer without first hearing a single no. A mere 9 percent heard at least three nos — even though there is no penalty or cost for being told no, save the small disappointment that every human being feels when hearing “no.”
And describes the fear of this disappointment as "confirmation bias". That is wrong!

We didn't realize that he was asking us to make up more data and find whether our invented data fit our own hypothesis or not. How exactly would you make up a disconfirming example? If I had been looking for a "no" I would have gotten a second "yes" and then gotten frustrated and clicked for the answer. What were those weirdos in the 9% doing?

You leap for the
xn = 2xn-1
solution without checking a broader hypothesis for a really good reason: because you know this is a problem, not some random data. And because it's a math problem, not an empirical one. In math there's no reason to look at more data, once you've seen the pattern. You know the author projected a pattern on it, and there's only supposed to be one working answer.

What's really wrong with it is that it's just a shitty problem, with too many solutions, if you count the stupid one Leonhardt wants. And the solution Leonhardt "wants" is a possible solution to most of the number series problems any of us have ever done, because they're almost always about rising number sequences (something like 12 out of 14 from this handy middle-school explainer).

We know from years of practice that "each number must be larger than the previous" is never the answer, because it's virtually always present, as a piece of background information.

We don't give Leonhardt's preferred answer just the way we don't answer
2 + 2 = some integer between 3 and infinity
The answer we give instead is the most precise true answer possible and therefore the best.

Also remarkably offensive is Leonhardt's bothsiderist exemplification of how his principle works out in government policy:
A version of this same problem compromised the Obama administration’s and Federal Reserve’s (mostly successful) response to the financial crisis. They were too eager to find “green shoots” of economic recovery that would suggest that the answer to the big question in their minds was, just as they hoped and believed: “Yes, the crisis response is aggressive enough, and it’s working.” More damaging was the approach that President George W. Bush’s administration, and others, took toward trying to determine whether Iraq had weapons of mass destruction a decade ago — and how the Iraqi people would react to an invasion. 
Neither case was anything like his stupid problem. These are empirical problems, and the job isn't to make up data to find out what the secret hypothesis is, but to make up hypotheses to account for the real data.

Both at some level typical cases of looking for analyses that would justify government policy, but Obama's was what Hamlet called "indifferent honest", i.e. not so bad by human standards, and Bush's was like Claudius,
The harlot's cheek beautied with plast'ring art
Is not more ugly to the thing that helps it
Than is my deed to my most painted word.
Obama and the Fed and I believe everybody except Geithner understood that the crisis response was a great deal less than really needed, but was the best they could push through the debt-crazed institutions of Congress and the press:
Senate Democrats, after ignoring calls for additional infrastructure spending by Senate Republicans, forced a near unprecedented level of changes (near $150 billion) in the House bill, which had more closely followed the Obama plan.
They were eager to show that it wasn't an utter waste, and they were right, as Leonhardt acknowledges with that "mostly successful". Also, the more data they got, the more successful it was, in cost-benefit terms, as the bailed-out companies all paid back their loans and the total expenditure shrank.

The Bush administration's interpretation of the Iraq WMD "evidence", on the other hand, was criminally dishonest, actually rejected data they didn't like (like Joe Wilson's from Niger or the evidence of Curveball's mendacity), and caused tens of thousands of deaths and was entirely unsuccessful, so it's really not the same thing.

No comments:

Post a Comment