RE: group based judgment

From: Rafal Smigrodzki (rms2g@virginia.edu)
Date: Fri Jul 26 2002 - 12:43:35 MDT


Harvey Newstrom wrote:

Women can't be terrorists? Old people can't be terrorists? Americans
can't be terrorists? Exactly what magical formula are you using to make
people above suspicion? And what does any of this have to do with
searching people and luggage on airplanes?

### You are drawing very far-reaching conclusions from my words. I never
claimed what is imputed by some of your questions. Please ask for
clarification if you have doubts about what I mean. But ad meritum - women,
especially older ones, are less likely to be effective terrorists. I am not
using a magical formula - merely the observation (doubtless well-known to
security professionals) that out of the thousands of hijackers ever known to
law enforcement, none were elderly women (well, I am not a professional, so
I might not know about a couple exceptions). This means that terrorist
groups will find difficult to recruit elderly women. We also know it is
difficult for a young man to pose as an elderly female. Therefore, as a
matter of simple inference, I conclude that advanced investigation
(including procedures cumbersome to the subject) of grandmas is less likely
to yield a positive result than a similar investigation of a young, restless
male. Do you find anything wrong with this reasoning?

----------

I'm not going to go through this whole argument again. As the article
Hal posted shows, profiling *can't* be used.

### This article is flawed. Here we are (quoted from the article):

"All of this open scrutiny makes it possible to learn an anti-profile to
defeat CAPS, even if the profile itself is always kept secret."

### The knowledge of an anti-profile is not sufficient - you actually have
to be able to find co-conspirators fitting this anti-profile, and by the
definition of a good profile, it is difficult. See below.

"In essence, the terrorist cell is able to relegate its harmful activities
outside of the 6% CAPS flag zone"

### Only if they can find a sufficient number of CAPS-safe recruits. See
below.

"It is entirely probable that even a rudimentary CAPS profile can flag many
individual terrorists, shrinking the viable pool of recruits that a
terrorist cell can send on a mission."

### Here the authors admit the essential utility of CAPS but then blissfully
go on claiming the opposite.

"It only takes one person to do harm"

###Depends. A terrorist cell with 20 members launching a coordinated attack
could bring 18 planes down (assuming an 8% chance of 2nd level exam with
100% effectiveness, and a 0% effectiveness of the basic screening) in a
random system. In a CAPS system with the same assumptions about frequency
and efficacy of examinations, if only ten of the cell members are CAPS-safe,
this reduces their effective number to ten and the number of downed planes
to 9. The same spending on security but supplemented with CAPS halves the
number of dead people.

"Do terrorist cells have a diverse enough membership to successfully use
this algorithm?"

### No. The cases mentioned in the article are outliers. Close-knit, tribal
and denominational groups do not have outreach programs. Even knowing the
anti-CAPS profile, they cannot benefit from it. They cannot (contrary to the
main, explicit but totally erroneous assumption of the article) freely
recruit grandmas from Kansas. Any attempt to recruit is always dangerous for
them, especially if targeted at constituencies very unlikely to be terrorist
material (the very ones that are CAPS-safe).

------

 "The computer tracked a simulated terrorist cell as it used the Carnival
Booth algorithm to recruit new members and probe the system with each
neophyte"

### Yeah? So where do they get these neophytes from?
-------

"Other distributions modeled highly tuned profiles with a mean CAPS score of
70/100 for terrorists and 20/100 for non-terrorists and standard deviations
of 20."

### The authors make guesses at the efficacy parameters of CAPS and presume
to derive rigorous policy recommendations from such conjectures,
supplemented by unrealistic assumptions about the sociology of terrorist
groups. But no - garbage in, garbage out.

------

"While El Al does keep a database of individuals' nationalities, genders,
criminal records and flight histories, this tracking should not be confused
with the a CAPS-like profile"

### Here again the authors point to a clearly efficacious case of profiling
and then discount it as irrelevant.

Some general remarks about the article. The authors have apparently little
knowledge about the mechanics of CAPS, and admit it, yet proceed to make
sweeping generalizations about profiling in general, not befitting a
rigorous study. The ability to write down some trivial equations and run a
simulation does not make a security expert. I guess that real security
experts (such as those who designed CAPS or work at El Al) could find many
more flaws in this article.

------
  It doesn't work. We can't
exclude any gender,

### This is not meant by profiling as defined by CAPS as currently
implemented.

-----

 we can't exclude any age, we can't assume that
birthplace records are accurate,

### This is not assumed by me or CAPS.

-----

 we can't assume the bag tags haven't
been switched, we can't assume that poor old grandma isn't unknowingly
carrying something put in her bag by others. The bottom line is that it
is just plain safer to search all the bags.

### Full agreement. But why do you have a problem with profiling?

Unless you define what you mean by profiling I will drop this subject.

Rafal



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:15:41 MST