Thursday, March 08, 2007

The Lamp Post Rule   posted by Matt McIntosh @ 3/08/2007 10:46:00 AM

There's a problematic tendency, particularly prevalent when dealing with emotionally-charged subjects, for people with little or no understanding of a complex subject to handle numbers that come out of that subject in much the same way that Moses handled the Ten Commandments: They are handed down from higher intellectual powers, whose ways are mysterious to us but whose authority is without question. They are the outputs of a black box whose inner workings are completely opaque, but which is quite useful to those looking for a blunt object with which to bash their opponents over the head.

Examples are not hard to find: Creationists making reference to Haldane's limit, global warming deniers talking about arctic ice cores, price control fetishists banging on about Card & Krueger, race-skeptics quoting Lewontin's infamous 85/15 figure, etc etc etc. We've all seen it, many of us have probably even done it at some point, and it's a stupid human trick that's centuries old.

Good old Arty Schopenhauer knew what this one was about:

This is chiefly practicable in a dispute between scholars in the presence of the unlearned. If you have no argument ad rem, and none either ad hominem, you can make one ad auditores; that is to say, you can start some invalid objection, which, however, only an expert sees to be invalid. Now your opponent is an expert, but those who form your audience are not, and accordingly in their eyes he is defeated; particularly if the objection which you make places him in any ridiculous light. People are ready to laugh, and you have the laughers on your side. To show that your objection is an idle one, would require a long explanation on the part of your opponent, and a reference to the principles of the branch of knowledge in question, or to the elements of the matter which you are discussing; and people are not disposed to listen to it. For example, your opponent states that in the original formation of a mountain-range the granite and other elements in its composition were, by reason of their high temperature, in a fluid or molten state; that the temperature must have amounted to some 480 degrees Fahrenheit; and that when the mass took shape it was covered by the sea. You reply, by an argument ad auditores, that at that temperature - nay, indeed, long before it had been reached, namely, at 212 degrees Fahrenheit - the sea would have been boiled away, and spread through the air in the form of steam. At this the audience laughs. To refute the objection, your opponent would have to show that the boiling-point depends not only on the degree of warmth, but also on the atmospheric pressure; and that as soon as about half the sea-water had gone off in the shape of steam, this pressure would be so greatly increased that the rest of it would fail to boil even at a temperature of 480 degrees. He is debarred from giving this explanation, as it would require a treatise to demonstrate the matter to those who had no acquaintance with physics.

There's always going to be a lot of assumptions underlying any specific truth-claim pertaining to a complex subject. Altering one or more of these assumptions will have an impact on the plausibility of the specific claim in question, and often it takes a trained eye to see what assumptions are being made, which are likely to be sound and which are highly questionable. But the unfortunately large intersection between the set of subjects people have strong feelings about and the set of subjects that take a non-trivial amount of education to adequately comprehend ensures that people are always going to end up arguing beyond their range of competence. So how do we deal with this combination of ignorance and importance?

I propose an informal rule for any sort of argument where the participants involved are not competent to evaluate specific truth-claims: All arguments conducted in a state of relative ignorance must be algebraic. I don't mean speaking in math -- I mean that such arguments should focus on the relations between variables rather than on what specific values to assign them. And if someone does plug specific values into the argument, they have to either A) be unobjectionable, i.e. everyone in the argument can agree that the values are reasonable, or B) be supported with at least a cursory explanation of how the values were produced (or a link or reference to such an explanation).

Anytime you see someone supporting their argument with specific numbers in a discussion with no hint at how they were arrived at, your bullshit detector should go off and you should demand an explanation of the method by which those numbers were produced. If upon being pressed your interlocutor cannot or will not adequately explain this (or provide a link or reference to someone else's adequate explanation), an argumentative foul has been comitted; those specific numbers and whatever parts of his argument require them may be disregarded and the offender's credibility reduced as punishment. Call it the lamp post rule.

And if this ever becomes as commonly referenced as Godwin, rememeber, you read it here first.