Print

Print


*****  To join INSNA, visit http://www.insna.org  *****

 

These editorials fail to point out the problem of false positive from data
mining.  Even if network analysts develop a extremely discriminating
algorithm for idenifying terrorist cells, that algorithm will certainly
identify innocent people as terrorists.  Even with a false positive rate as
low as 1 in a million, the system would target thousands of innocent people
for investigation every day.  

See:  Why Data Mining Won't Stop Terror
http://www.wired.com/news/columns/0,70357-0.html?tw=wn_story_

Maybe the question that survey researchers should be asking Americans is
whether they support a data mining system that would bring 1 million
innocent persons under suspicion and possible FBI investigation every year.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Christopher Weare 
Research Associate Professor 
School of Policy, Planning, and Development 

VKC 371
University of Southern California 

Los Angeles, CA 90089-0626

 213-740-4690 voice 
[log in to unmask] 


-----Original Message-----
From: Social Networks Discussion Forum [mailto:[log in to unmask]] On
Behalf Of Steven Corman
Sent: Tuesday, May 16, 2006 10:33 AM
To: [log in to unmask]
Subject: Re: NSA Data Mining

*****  To join INSNA, visit http://www.insna.org  *****

BTW Valdis some caller the Diane Rheem Show on NPR yesterday was quoting you
saying similar things.  

Here's the deal: I just can't believe they have played hanky-panky with the
law (being charitable, there) and amassed this huge wad of data just to go
on a complete fishing expedition.  It's more plausible that they have
identified a pattern from looking at known cells that they think they have a
good chance to detect.  

The critiques linked in the your message all assume that they are looking
for a single person.  What if they are looking for small groups?
We know from captured documents that cell members are (or were) trained to
remain detached from the host society.  It's easy to imagine that
manifesting itself in a strange pattern of phone calling: only or mostly
within a small group, no calls to family, few calls in the same area code
outside the group, and so on.  

That would surely describe some non-terrorists too, but throw in some other
variables like age of the phone numbers, geographic areas where there have
been known calls from the Bad Guys, some time-series techniques, and it does
not seem far-fetched that you could get some useful leads.  

The legality of how they have done this is what worries me, but apart from
that I have yet to be convinced that it couldn't work.

Steve


-----Original Message-----
From: Social Networks Discussion Forum [mailto:[log in to unmask]] On
Behalf Of Valdis Krebs
Sent: Tuesday, May 16, 2006 9:49 AM
To: [log in to unmask]
Subject: [SOCNET] NSA Data Mining

*****  To join INSNA, visit http://www.insna.org  *****

Two good editorials on why data mining just phone links may not be useful...

http://jeffjonas.typepad.com/jeff_jonas/2006/05/the_six_degrees.html

http://www.nytimes.com/2006/05/16/opinion/16farley.html

Valdis

_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send an email
message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in
the body of the message.

_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send an email
message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in
the body of the message.

_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.