[ih] Yasha Levine's Surveillance Valley: The Secret Military History of the Internet -- Some Questions

[Bill Ricker]

I wasn't there but ... maybe I can connect some dots.

This smells like coincidence ... that Licklider's CAM and MAC projects
(and many other defense projects) were\ at MIT makes a connection
"obvious" even if there wasn't one.

A budget-balancing transfer of funds from AGILE to Licklider seems
perfectly reasonable from a bureaucratic point of view. It doesn't
mean the money was for ARPAnet even though that is the last project
that put Licklider over budget; they could be listed together on one
memo because they're the two amendments to a previously approved
budget. It does means AGILE had uncommitted funds when Lick was
overcommitted. The Psych portion of Lick's portfolio was certainly of
common interest, but a transfer might have occurred from any
undercommitted team, as failure to spend funds may lead to a reduction
in budget!

I would be very impressed if either of the captains of research
expected the ARPAnet to be actually useful to AGILE researchers in the
near term. Although the possible benefits of a future MILNET for
collaboration between applied social science researchers in in-country
anti-insurgency INTEL centers and their peers back home (in academe or
CIA HQ) could perhaps be foreseen, the undersea cables and high
bandwidth satellites needed to connect a SAIGON operating center to
back home were decades in the future. (There was eventually a low
bandwidth link to UK and from there to NATO and a treaty verification
seismology lab in Scandinavia but even that was far future at the time
in question.)

Active INTEL databases have been classified since forever. As Vint
notes, there was a classified adapter for MILNET nodes of the
(d)ARPAnet, for passing data from from one MILNET node to another.
That would technically be "over the ARPANET"  since until TCP/IP
cutover, it was only one network, but with an encrypted tunnel of some
sort. I'd be shocked if active INTEL data was sent that way, I doubt
the adapters were certified for higher classifications;  but ...
anything's possible, especially as exceptions.  Sending to MIT? That's
distinctly odder.

I am unaware of the Natick Army Labs being involved in anything like
this ... they developed the tropical chocolate bar and new uniforms.
Could they have had an AGILE branch? I guess plausible. Might a
researcher working with NSA or CIA have collocated with Natick to have
secure facilities instead of at Draper, Lincoln Labs, MITRE, BBN, etc,
for whatever reason ? IDK, possible, but seems very odd. But if they'd
had a compartmentalized sideline, no one would know. That's the beauty
of black programs and conspiracy theories, lack of evidence is
inconclusive.  Were they home to contract managers for some ASA
research project with MIT? Perhaps. Before NSA could use its name
publicly, they'd have let contracts as ASA (or successor names) and
the Navy equivalent.

The mystery files at MIT make me think of CCA's Model 204 work for
"The Community", which may well be an MIT Intelligence-research
spinoff. (While possibly connected to CAM or more likely AGILE, it
might have been more applied and directly funded CIA/NSA R&D contract
funneled through ASA?) The inventor of Model 204's key internals, Pat
O'Neil, was a professor at MIT immediately before CCA, and had been
working on the special index structure for nearly a decade.  Just
guessing but looks like development may have been at MIT as contract
research and fielding, support, and future maintenance/support was
spun off to CCA, formed conveniently down the block?
   (For decades Model 204 was the only DBMS capable of big-data and
text-retrieval. The opening sequence in "3 Days of the Condor" movie
(likely 7 days book too?) showed you an AGILE/CAM type team using CCA
software to digitize printed source documents into a document
retrieval system ahead of the unclassified state of the art, if I'm
connecting the dots right. I worked with tape extracts from a Model
204 Text DBMS in an unclassified setting in the late 1990s -- the
National Library of Medicine MEDLINE bibliography&abstract system was
then, likely still is, based on Model 204. Lucious metadata, it had
ontological search before the phrase was coined. You can access it as
PUBMED, thanks to Al Gore -- which undercut our startup's business
model, oops. )
   The tapes being seen at MIT does not mean they were sent over the
ARPAnet. In those days, was it not the case that a courier with
several tapes in a locked bag taking the train from DC to Boston had
better bandwidth, latency, and error recovery? I was still getting
tapes sent from NLM's Model 204 via USPS/UPS in 1990s. (And a weirder
EBCDIC variant I've never seen.)
    I'm guessing the mystery tapes at MIT were test data sent to
O'Neil to test his pre-production DBMS ?  Back in those dark ages,
they might not have thought to make the test data anonymized/mangled.
( People still forget that today in a post HIPAA/PCI world!)  Or,
realizing that a real intel DB being released to an academic
environment would have been a security problem for NSA/CIA, maybe they
made a test file with data they swiped from Commerce's Census dept?
Just brainstorming here.
   Pat O'Neil is Professor Emeritus at UMass/Boston, where he
co-founded the CS department on his return from industry.
[ https://www.cs.umb.edu/~poneil/  ]  He might be able to shed light
on the NBC reports of MIT having had tapes that belonged at Langley or
Ft. Meade, and which of Licklider or AGILE or CIA/NSA/ASA was his
original funding source.

You could also check with Don E Eastlake iii (on some IETF/W3C groups)
on CCA DBMS history.
https://www.informit.com/authors/bio/5f1734d3-42df-49f0-b2e2-61007b188cd1

// Bill Ricker
// Friend of Padlipsky