National Security Letters vs. Counterrorism investigations

[David Farber <dave () farber net>]

Begin forwarded message:

From: Krishna Kumar <kmkumar () earthling net>
Date: March 15, 2007 9:24:49 PM EDT
To: dave () farber net
Subject: National Security Letters vs. Counterrorism investigations



for IP


While a lot of attention is focused on the firing of the US attorneys,
there is another FBI/DOJ scandal -- the abuse of National Security  
Letters.
But I saw a chance to link two bodies of data, and I think it does shed
some more light on the problem we have.  What follows is a brief  
analysis
comparing the number of NSL issues to the number of threats the FBI  
tracks,
number of  FISA warrants, number of terrorism/anti-terrorism  
prosecutions,
and number of terrorism prosecutions.  In 2004,  56,000 NSL requests
resulted in 93 prosecutions.

The following data is coming from the DOJ IG NSL report.

Here are the base figures:

2003:  39,346 NSL Requests
2004   56, 507 NSL Requests
2005:  42,221 NSL Requests

Now a NSL request -- as opposed to a actual NSL -- could  be for the  
same
person.  For instance, if they wanted to  investigate me, they would ask
for a home phone, mobile #, work #, work email, home email, credit  
history,
and bank records..  So one LETTER might contain 10 or 12 requests on one
individual.

Unfortunately, the government doesn't really track the number of NSL
letters.  But they do track the number of PERSONS involved for the same
years.

2003:   16, 751 (10,232 non US person)
2004:   17,437 (7494 non US persons)
2005   18,011 (8526 non US persons)

Now, we know that 73% of REQUESTS deal with counter-terrorism
investigations (as opposed to counter intelligence investigations, or  
cyber
attacks).  What we don't know is the number of PERSONS involved in  
counter
terrorism investigations, or the number of counter terrorism  
investigations
that involve a NSL request.

(That is in chart 4.4 of the IG report -- blacked out and marked secret.
If anyone wants to leak it to me...)

But just for fun, let's assume that 73% of the persons are involved with
counterterrorism as well.   That would give us the following figures:

2003: 12, 228
2004:  12, 729
2005:  13,148

The numbers are an estimate.  It's possibly that a bulk of the non US
persons are MORE involved in counter terrorism, and that would skew the
numbers somewhat.

Now our terror expert is going to point out the FBI is about  
intelligence
gathering.  Now the numbers get very murky here - the relevant fact  
in the
newest IG report is marked "Secret".

Let me drop a mental footnote:  a NSL is something that a FBI officer  
can
initiate on their own. As far as I can tell, during a regular  
investigation
you could go to a judge or grand jury and ask for the same thing if you
have probable cause.  FBI divides this into three pieces: threat
assessment, preliminary investigation, and full investigations.  You  
can't
get a NSL during a threat assessment. The IG starting tracking these in
2003 because that is when the guidelines allowed for a NSL to be  
requested
during a preliminary investigation.

But just for fun, let's look at the threats the FBI is dealing with.   
This
is from another IG report issues in February about DOJ and terrorism
statistics.  Now, the IG took the FBI to task on this report for  
failing to
properly report on how it was tracking terrorism, so these numbers  
may be
off.

FBI:  Number of terrorism-related  threats tracked in FY 2004

4499 reported,  4059 after IG

In this case the  FBI told the IG that these reported numbers only
accounted for the threats tracked by the Threat-Monitoring Unit and were
only 40% of the estimated threats (local offices were also tracking  
threats
as well).

Let's accept that at face value.    If that's the case, then the FBI was
tracking an estimated
11,000 threats during FY2004.  It's only a data point for  year, and is
extremely rough.  But compare that with the estimated 12,000 individuals
that the FBI asked for NSLs on during 2004.   It is a nice easy  
comparison.
And that number is being generous to the FBI - their official numbers  
only
reflect 4000+ threats in 2004.

In the NSL IG report, FBI officials say that NSL are their "bread and
butter".   In particular, what they are useful for is to build a case  
for a
FISA surveillance warrant.   That makes some sense.   Now, we don't know
much about FISA cases - they are all secret - but thanks to another
internet source we can find the  number of FISA orders issued over  
the same
period of time.  This is coming from EPIC data:

2003:  1727 FISA orders presented to secret court
2004:  1758  FISA orders presented to secret court
2005:  2074 FISA orders presented to secret court

We don't know if a FISA order deals with only one person, or with a  
class
of individuals.  But we can quickly look compare this against the  
number of
people being investigated under NSL letters.  Again, I'm being generous
here:  I'll use the ESTIMATED number of people under counterterrorism
investigation against the FISA numbers;   it is possible that a FISA
request may have been based in counterintelligence rather than  
terrorism:

2003: 12, 228 people investigated, 1727 FISA orders
2004:  12, 729 people investigated, 1758 FISA orders
2005:  13,148 people investigated, 2074 FISA orders

That is about a 15% hit rate, and keep in mind that is a generous
interpretation.

An experienced counterterrorism person will jump in now and say  
"Look, it's
not about convicting terrorists.   It's about disrupting their  
networks -
arresting them for little things."  That analysis would be correct.

It looks as if DOJ looks at three types of counterterrorism activity:
Terrorism, Terrorism-Related, and "Anti-Terrorism".  "Anti-Terrorism" is
what your analyst would say is the Al Capone method of justice - arrest
them for the little stuff.

TRAC is not useful here:  it gives you a total figure (6472) for a five
year spread.  So let's go back to the February IG report.

But we can use that report to get a few additional data points.  The  
new IG
reports pulls in information from the FBI, EOUSA, and the Criminal  
Division
of DOJ.  And again, one of the findings is that these numbers are
potentially inflated.   One key factor is that is a person is CONVICTED
because of a terrorism related investigation, that conviction is  
marked as
terrorism-related or anti-terrorism.  But it may be a coincidence:  the
example is you bust a warehouse that you think has a bomb and find the
guard is an illegal.  Is convicting the guard really a terrorism-related
conviction?  So in certain areas the IG went in and did their own
interpretation of the data.  I'll report both.

2003 Number of terrorism-related convictions  EOUSA  558 (281 after  
IG review)
2004 Number of terrorism-related convictions  FBI  206  (141 after IG  
review)
2004 Number of terrorism-related convictions  EOUSA  379 (240 after  
IG review)

I can't explain why the FBI and EOUSA have different statistics for what
appears to be the same conviction type in 2004.  But averaging it out,
we're getting about 225 terrorism-related convictions for 2003 and  
2004 if
you believe the IG numbers, and about 400 a year if you believe the DOJ
numbers.

I could walk you through the same analysis on the "Anti-Terrorism"
convictions.  But there is another stat which can shortcut that  
argument:

Number of defendants the U.S. Attorneys terminated terrorism and
anti-terrorism cases against  in FY 2004:

643 reported
401 after IG investigation

So, even assuming the DOJ numbers, we've got an estimated 12,729 people
investigated in 2004 with a NSL.   And we've got 643 potential  
defendants.
That is roughly a 5% hit rate.  These are rough numbers.  People
investigated during 2004 may not be prosecuted until the future.

There is one last number to look at.  We tend to justify intrusions into
our civil liberties because of the threat that terrorists pose.  So  
let's
look at the really bad category:  what  the government records as actual
"terrorism" cases.

There is another report floating around -- from Syracuse University --
where they have been tracking "international terrorism" cases through  
the
EOUSA.  I can't tell from their data, but I suspect their definition of
"international terrorism" is the same as the FBI definition of  
"terrorism".

I could be mean.  After reading the 2006 DOJ White Paper on
counterterrorism and reviewing their selected excerpts, its clear  
that are
some bad people out there.  But a lot of these folks are being  
arrested for
going jihadi in Iraq or the Middle-East rather than at targets in the
Homeland.  And the DOJ numbers do include our occasional domestic
terrorists as well.

A US attorney's office does not have to bring an indictment when told  
of a
wrongdoing.  So what I am going to list now is the number of  
prosecutions
by year and how many cases were also turned down, based on the TRAC  
data of
"international terrorism":

CY2003:   66 prosecutions, 200 declined
CY2004:  93 prosecutions,  172 declined
CY2005:  46 prosecutions,  209 declined

When we look at the 2006 USDOJ White paper on Counter terrorism the  
numbers
seen to be compatible with the TRAC info.

DOJ reports 441 defendants have been charged between FY2001 and FY2006.
This works with the TRAC figures, because there was a major bump in  
CY2002
--355 according to TRAC.  DOJ reports that of the 441, 261 were  
convicted,
150 are pending, and 29 were dismissed -- breaking down the dismissed
numbers we see that 4 were deported and 8 found not guilty.  So there  
is a
high conviction rate for those charged.

(TRAC does note that the average sentence in a terrorism case is now 20
months, which makes you suspect that the charges are NOT that serious to
begin with and  may be more "terrorism-related", but that is a side  
issue)

The numbers I want to focus on comparing the number of requests to the
number of prosecutions.  I'll use my imaginary numbers for the number of
PERSONS to keep the numbers reasonable, and I'm using the TRAC info  
for the
number of prosecutions.

2003:  12, 228 people investigated;  66 prosecutions
2004:   12, 729 persons investigated, 93 prosecutions
2005:  13, 148 persons investigated, 46 prosecutions

Back of the envelope, that is about .05 percent hit rate.

TRAC is based on FOIA requests to the EOUSA, so it is not a perfect  
number.
Looking at the February 2007 IG report, we can pull the following  
numbers.
Information is courtesy of the EOUSA:

FY2003:  Terrorism convictions:   103 (the IG thinks there are only 86)
FY2004:  Terrorism convictions:   118 (the IG thinks there are only 97)

Now one key difference is the TRAC info is based on Calendar Year, while
the DOJ info is based on Fiscal Year.  So you can't compare them  
directly.
But they are in the same ballpark (103/86 vs. 66,  118/97 vs. 93).  
And the
DOJ numbers include domestic terrorists, not just "international
terrorism."   But even if you take the DOJ numbers at face value, when
compared to the number of NSL requests you are getting a 1% hit rate.

Let me go back to the TRAC reports for a moment.  I mentioned before  
that a
US attorney's office can DECLINE to bring a prosecution.

And the FBI is not the only government agency that can refer a case  
to the DOJ.
The TRAC data shows five government agencies are the primary referring
agencies on "international terrorism" over a five year period:  FBI,
Immigration (ICE/CBP), SSA and TSA.  As you can expect, the FBI is the
largest by far - 913 of 1391 over five years.  And keep in mind these  
are
the actual "terrorism charges";  not the little nickel and dime charges.

But what is REALLY interesting is that the US Attorneys declined to
prosecute 82% of the referrals by the FBI for the most serious
"international terrorism" cases.  These aren't cases that were  
dreamed up
by bored agents in Boise.  Thanks the internal changes at the FBI, the
cases the FBI are bringing for prosecution probably have been through
several levels of review, and well as having the local Joint  
Terrorism Task
Force look at them.  And yet US Attorneys are declining to prosecute.

Our counterterrorism expert will now point out that a US Attorney's  
office
may not have the expertise to understand and prosecute a terrorist case.
Perhaps true, but a quarter of the international terrorism cases all  
landed
in ONE US Attorney's office - the Eastern District of Virginia - and  
with
that caseload you can imagine some expertise been retained.

You can understand why the Administration views NSL as such a valuable
tool.  If any threat comes into the FBI, a NSL will likely be  
issued.  And
I mean any threat - if you see a guy with a turban acting funny, or  
someone
reading Arabic at a computer in a public library --  once it gets  
reported
to the FBI it is threat.

Back to the FBI for a second.  I think it's clear that the use of NSL is
now standard in ANY case that is labeled a threat.  The definition of a
preliminary investigation is too loose.  And at the same time,  82%  
of the
cases the FBI are bringing to the a US Attorney are being declined for
prosecution.

According to the DOJ, funding for preventing terrorism at the FBI have
increased from $340 million in FY2001 to $1.129 billion in FY2005.   
That is
an increase of over $700 million dollars.

NSL may have once been a useful investigative tool.   Perhaps for
counter-intelligence operations  they still are. But by giving the FBI a
tool to easily investigate suspects and threats,  they are making it to
easy.  Based on a very rough look at the numbers,  the referrals the FBI
are making the US Attorneys for prosecution are not good, and that  
probably
means the quality of their investigations is not good either.  We  
need to
take some of the billion dollars we are spending every year on
counterterrorism at the FBI and use it to go after the much smaller
universe (646 people?)  of bad people out there.  NSL aren't like "bread
and butter."  They're more like crack, and the FBI is addicted.

The problem is NSL were never meant to be used during a "threat
assessment."  But we are using these to build up a useful intelligence
product.  The FBI, according to the IG, is retaining the phone  
records and
dumping them into a database that anyone could use. But that was not the
intent of the NSL - the Patriot Act changes were supposed to make it  
easier
for the FBI to use this tool.  By giving the FBI the ability to use a  
NSL
in a preliminary investigation, we've created a monster.

At it's heart, if we commit to a rule of law in fighting the bad guys,
this entire process is supposed to be a pipeline.  But a breakdown of  
the
numbers shows that this is an extremely constipated pipeline - one where
your taking a vast amount of intelligence at the top (the NSL) and  
turning
out very few convictions in the end (the terrorism prosecutions).

Our hypothetical counterterrorism expert would probably shrug at  
this, and
acknowledge that this is hard work, because you're looking for a  
needle in
a haystack.  So let me make one final point.  If we RELY on NSL to clear
suspects, they are an inherently flawed tools.   What they measure are
connections to earlier suspects;  if a terrorists wanted to fly under  
the
radar all they would have to do 1) use disposable cell phones;  2) make
sure not to call associates in the country and 3) don't use credit  
cards -
do everything with cash.  With those simple steps you would be  
invisible to
the FBI.  What we need is REAL investigations, and not this simple  
tool to
weed out the good from the bad.

The final policy prescription is that before the Patriot Act was passed,
FBI local offices did not have the authority on their own to issue a  
NSL -
it needed to be done by headquarters.   Given this waste by the FBI, we
need to change the authority of WHO can issue the letters - either
restricting it to a group in headquarters or some other central  
authority.





-------------------------------------------
Archives: http://v2.listbox.com/member/archive/247/@now
Powered by Listbox: http://www.listbox.com