"Eliezer S. Yudkowsky" wrote:
> I have this wacky idea that says it's ethically prohibited to
> apply Bayesian priors to people. In other words, you should judge a
> person's properties ... only by information that is directly revealing
> of those properties, and not through other properties that happen to
> be weakly or strongly associated...
Very decent of you!
I have two memories, half amusing and half dismaying, of
situations in which I was the victim of statistical profiling
in an inconsequential but extremely irritating fashion.
Both incidents happened sometime in my late teens, and both
involved accusations of cigarette smoking because, well,
that's what teenagers do, innit? I have never in fact smoked
cigarettes; in fact I've never smoked anything in my life
**except** pot ;-> (a long time ago).
The first of these incidents happened when my mother discovered
cigarette ashes in the bathtub and decided I had taken
up smoking in the bathroom. My mother herself smoked like
a chimney, so if there were any ashes in the bathtub, she
was herself undoubtedly the source of them. If I was
doing anything at that age that caused me to spend a long
time in the bathroom, it had nothing to do with cigarettes
<blush>. The upshot of this was a very surreal, but
nevertheless acrimonious row with my parents. (The fact
that I didn't get into any conventional kinds of trouble
confused the hell out of my parents and I think was
actually a source of disappointment to them.)
The second incident happened during a visit to the dentist.
I had been going to the same dentist (specializing in
children's dentistry) since I was a little kid (in early
days it was just the one guy, but his two younger
brothers joined the practice when they graduated from
school). My parents forked over for orthodontia from
this guy, too, so I'd been through a lot in that office
(just the characteristic smell of the Newark [Delaware] Medical
Building always sent a hot wire straight into my amygdala;
I've no doubt I would still react strongly to that
smell).
Well, one day when I was 18 or so I was there for a regular
checkup (with one of the two younger brothers), and he noticed
that there was a lot of staining this time. Probably this was
because I had taken to drinking tea, but the dentist immediately
jumped to the conclusion that I had taken up cigarette smoking.
When I interrupted the lecture he had launched into on the evils
of cigarettes to reassure him that I did not in fact
smoke, he didn't believe me, and he expanded his lecture to
include his disappointment in the fact that I would attempt
to lie my way out of his discovery. I was not only angry
enough to bite his hand off, I was horrified by his lack of
professionalism -- his business was to take care of the
toofies; the health of my lungs and my soul were strictly
outside his area of expertise (though maybe, being children's
dentists, they were on some kind of in loco parentis trip;
I wasn't having any of it). I was promptly and unceremoniously
dropped as a patient after that visit, ostensibly because I was
now too old to be among their clientele, but I suspect it had more
to do with that confrontation with the younger Dr. C.
Of course I know that other groups of people in this country
suffer far more humiliating (and even life-threatening) consequences
from their Bayesian correlates -- I now live in the state of
New Jersey, which has been caught in a scandal in recent years
having to do with certain groups of people being pulled over and
harrassed by the state police on the New Jersey Turnpike and
elsewhere based on statistical profiling. Nevertheless, I
always grind my teeth in response to sanctimonious propaganda
about the evils of "stereotyping" -- **everybody** stereotypes;
it's just another word for guessing probabilities based on
incomplete information, and sizing up situations quickly
using rough-and-ready rules of thumb. If people sometimes
get pinched by the Procrustean beds nailed together for them
by other people's Bayesian calculations, those are the breaks.
But DON"T LET IT HAPPEN TO ME!! :-< :-<
;->
Jim F.
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 10:00:02 MDT