Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: profiling (was: finding information on the Internet)



It has been pointed out [off list] that on
04/09/2003 02:53 PM, I should have written:
^^^^^^

All you will accomplish by this is
a) Creating a racist society.
b) Causing all evildoers who happen to fit the profile
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
to take off their
kaffiyehs, shave their beards, and buy two-way
tickets.

I hope it was obvious from context that I did
not mean to suggest that all evildoers actually
fit the profile.

======================================================
On 04/09/2003 04:19 PM, Rick Tarara wrote:

I'd reeeeeally like to hear your alternative here. It seems that not doing
something like Tim's example either requires we strip-search all passengers
OR we tolerate x number of planes blown up by 20 year-old Arabic males
traveling one-way carrying backpacks? ;-(

This is so far outside the charter of the list
that we probably shouldn't pursue it. Maybe
somebody can suggest a more appropriate forum.

And a proper answer would require more detail
than I can afford to provide right now.

... but here's a sketch:

The options are:
1) Search nobody.
2) Search everybody.
3) Search some people based on a "profile".
*) Other options not yet mentioned.

Actually option (3) is a family of options depending
on a "threshold" parameter; if you set the threshold
really permissive it becomes equivalent to option (1),
and if you set the threshold really restrictive it
becomes equivalent to option (2). For intermediate
values of the threshold we must examine the ROC
(receiver operating characteristic). On a ROC graph,
one axis represents false positives while the other
axis represents false negatives. Draw the curve,
parameterized by the threshold. A curve that
passes near (0,0) is good.

One of the biggest mistakes a strategist can make
is to underestimate the enemy. Most "profile"
strategies make this mistake. If the enemies are
not deaf, dumb, blind, and stupid, they will
figure out how the profile works and find a way
to "game the system".

As a result, the profile option has just about the
worst ROC imaginable. The rate of false negatives
is very high (comparable to option (1)) unless the
threshold is so restrictive that the rate of false
positives becomes comparable to option (2).

To summarize: assuming non-extremal threshold
settings and non-stupid enemies, profiling has
-- low effectiveness (comparable to option (1))
-- high direct cost, out of proportion to any
possible benefit
-- abhorrent indirect costs and side effects,
(institutionalized racism etc.)

So..... One may conclude that option (3) is not
a good idea.

====

There are other options that work better than
anything so far mentioned, but they don't work
very well, and to present them would require a
very long, very off-charter discussion.

Also: I predict that the next major atrocity won't
be a bomb on an airliner. I hesitate to make such
predictions. On 6 March 1995 I asked a security
committee why they were fixated on bombs to the
exclusion of poisons, and airliners to the
exclusion of office towers and subways. They
rolled their eyes patronizingly. They assumed
I was naive or daft or both.

Two weeks later they were really, really angry.
They wanted to burn me for a witch.