RE: Thinking Correct Thoughts

From: Lee Corbin (lcorbin@tsoft.com)
Date: Mon Aug 27 2001 - 00:59:33 MDT


Harvey Newstrom writes

> Lee Corbin wrote,
>> Now, I think that there *is* an implication in that sentence that
>> Mexicans are poorer than business executives, and are poorer
>> than some other ethnic groups. But Olga---this is a fact!
>
> This is not a fact. This is sloppy logic. If you were working on an AI
> database and wanted to code the logical "fact" that "Mexicans are poorer
> than business executives", I would object most vigorously. This "fact"
> would break a lot of the logical relationships of other "facts" and would
> probably poison the whole project.

You are perfectly correct: those words that I quoted when taken
out of context are erroneous, and no AI should be administered such.
There are actually very few statements that are impossible to
criticize (Eternal Truth #2: Every statement needs to be further
modified.) Natasha's original remark specifically talked about
Los Angeles and indirectly alluded to the fact of recent immigration,
housing problems, and so on. But what was important here was not
whether the particular item in question was indeed true: what was
important is that the nature of the "allegation", i.e., Mexicans
in L.A. could be expected to have fewer resources on average than
business executives (who have expense budgets, for one thing),
especially if it came to purchasing a gadget that has immediate
and valuable business applications, and relatively fewer social
ones.

It was, and is, ridiculous to attack the notion that in this case
anything was wrong with the original statement implying that [in
Los Angeles] Mexicans [statistically] are poorer than business
executives [in the arena of being able to afford cellular phones].
I didn't think that you would need to have such qualifiers spelled
out. They were obviously there in the context of the remark.
Unless you just wanted to attack something.

> It implies that "all" Mexicans are poorer than "all" business executives,
> which is untrue. It implies that the groups of Mexicans and business
> executives are mutually exclusive which is not true. It implies that
> Mexicans cannot be rich like business executives, which is not true.

All of these things are so. But right now, humans are different than
AIs in the sense of not making such ridiculous blanket generalities.
As I said, it is impossible in most cases to so completely qualify
all the words of a sentence to the point that someone cannot infer
a faulty generalization such as these.

> This is the exact reason that statistical numbers for a whole group cannot
> be applied to any individual. It is one of the more common logical
> fallacies.

Yes, quite right. But I almost never get a chance (fortunately?) to
talk with people who improperly impute statistical characteristics
to individuals.

Lee



This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:40:18 MDT