10 February 2015
ODNI Litt Transcipt Faulty
transcript was distributed by Brookings as an
transcript." Another version was posted to ODNI, which omits pages 24-35
of Q and A in the uncorrected version):
To: jya[at]cryptome.org, gnu[at]toad.com
Subject: ODNI-litt-15-0204.pdf is unreliable: typos
Date: Mon, 09 Feb 2015 20:11:40 -0800
From: John Gilmore <gnu[at]toad.com>
There are a couple of suspicious looking typos in that transcript.
In several cases, the transcript of Mr. Litt's statements says the opposite
of what you might expect. A good quality video of the talk is on YouTube,
and I cross-checked these typos against it.
Even so the President recognized the public concerns about this program and
ordered that several steps be taken immediately to limit it. In particular,
except in emergency situations NSA must now obtain the FISA Court's advance
agreement that there is a reasonable, articulable suspicion that a number
being used to query the database is associated with a specific foreign terrorist
organization. And the results than an analysts get back from a query are
not limited to numbers in direct contact with the query numbers and numbers
in contact with those numbers, what we call two hops instead of three as
it used to be.
The expected statement is that the results are "now" limited, rather than
are "not" limited. Indeed, that is what the video shows him saying.
There have always been very strong practical considerations against the use
of intelligence-derived information in criminal cases because if you use
the intelligence information a criminal case you have to expose the capability,
you have to litigate it, and it's something we always prefer to do.
The expected statement is that it's something we always prefer "not" to
do. Interestingly, in the video he seems to say exactly what the transcript
shows. So rather than a typo, it's just an interesting statement.
I'm not sure what Mr. Litt meant by saying there are strong reasons not to
use intel in criminal cases, so we always prefer to do it. Perhaps
he's referring to what Reuters revealed about the DEA practice of teaching
police to deliberately lie and mislead judges and defendants about whether
their information came from an NSA wiretap, referred to as "parallel
construction". This is not a new practice -- the LAPD was doing it
in the 1990s from ordinary wiretaps, where they called it "the handoff".
It was unconstitutional then (called "judicial deception" by the 9th Circuit)
and it still is. See for example:
Another typo that's more amusing than relevant, on page 25:
So I do think that there should be some comfort taken by this stage of the
game that we're not in fact randomly roaming around listening to communications
of burgers, finding out what they're going to have for lunch, but
that we're doing appropriate foreign intelligence collection.
For "burgers" I think he meant "burghers", that is, middle-class citizens
in Germany or the Netherlands.
Given this set of example typos, I don't think this transcript can be relied
on to be accurate. I discovered others like "you're barber" for "your
barber". It clearly hasn't been checked by anyone who actually understands
the policy issues -- or understands whether burgers eat lunch or are eaten