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

[David Walden]

Yes, my memory as I wrote an earlier message was wrong.  The second CTSS system was at Project MAC, not IBM Cambridge Scientific.

(Regarding reference 1 below, Joe Markowitz was at BBN in 1968 while the proposal to do the IMP development was being written; he was an active reviewer of what was being written.)  

On April 15, 2018, at 6:06 AM, John Klensin <jklensin at gmail.com> wrote:

Hi, while I'm on this list, I don't routinely follow it, so it took a
while for me to be pointed to this thread and longer to find the
energy to respond.

For context, I was involved with the Cambridge Project from the time
an early draft of the proposal started to be circulated to relevant
researchers within MIT, through the summer study, and then ended up
with lead responsibility for among other things, the software that was
intended to hold everything together, was a member of the small
steering committee (I don't remember what it was called, but that
wasn't it) that had practical oversight of the Project.  I worked
closely with Lick and more closely with those who were running the
project on a day-to-day basis.   Lick was actively involved (more than
I think Waldrop realized) but was leading the Dynamic Modeling work at
the same time and almost certainly more involved there on a daily
basis.   When I decided to do work leading to a Ph.D a few years after
the Cambridge Project wound down, Lick ended up on my somewhat-strange
committee.  If I recall, he was one of those who helped convince me I
should do the degree.  I'm happy to answer specific questions to the
extent that I have time and remember --the Project did zero classified
research--  but it has been a long time and MIT has, at least IMO, a
bad institutional memory problem for activities that are not linked to
active departments and/or sources of funds..   I have no idea whether
the original idea for what became the Cambridge Project originated
with Lick or de Sola Pool -- I worked closely with both, the latter
even earlier than I first met Lick, but, by the time I heard about the
idea, it was described very much in "joint effort" terms.  I also knew
(and know) enough about the interests of each to guess where some
ideas came from but find it difficult or impossible to try to
attribute most of the ideas to either independently.

I'll try to describe what it was all about, but it is probably
important that those trying to understand the effort (and almost
anything else related at MIT or Harvard at the time, especially if
there was DoD money involved, was that the late 1960s and first half
of the 1970s were times of great tumult in the academic and research
communities, with large differences in style among institutions about
how those things played out.  I don't believe we had anyone killed in
Cambridge, but there were a lot of loud demonstrations, marches, etc.,
There were some unpleasant confrontations between demonstrators and
the Cambridge Police and I can remember the smell of tear gas
Because it involved social and behavioral science research and
researchers, including some whom some of the most active of the
antiwar community were suspicious of for other reasons and because it
involved DoD (whether specifically ARPA or not, and it was ARPA)
funding) which meant to them that something nefarious was going on,
Some of those stories were on a par with some things we hear today
about the "real" reason the ARPANET work was funded; some were, at
least in my opinion, far worse.  The times were troubled enough that I
had some people who were working for me by day (because they were
comfortable with what they were doing and what they could see)  and
picketing us by night (some because of the principle of DoD funding
and others because of what "must" be going on elsewhere in teh project
although they could never find any sign of it).  The noise was loud
enough that, if one looks through contemporary articles, one can
probably find a lot of things that were the result of those kinds of
thinking (i.e., without strong connections to reality) and find then
with great ease.   We are a lot more interested in getting work done
than in trying to hold debates with those who were not willing to
listen and who, in many case, felt that anyone who disagreed with
them, their positions, or their truth should not be allowed to speak
at all.

Organizationally, the project was originally intended  to be a joint
MIT-Harvard effort. It was also intended, from the beginning, to be
organized the way Project MAC was originally organized (in retrospect,
probably unsurprising given Lick's involvement in shaping both), i.e.,
some centrally-funded and managed core activities, support from the
Project for complementary activities of various faculty and
departments, and some more independent activities with their own
independent (e.g., non-DoD) support that were nonetheless
collaborating (the latter group of activities was important with
Project MAC but was never significant with the Cambridge Project and,
as far as I can remember, never came together),   There were many
protests and some debate about that at Harvard.  The _Crimson_ article
cited was part of that fabric; perhaps something about its balance and
dedication to reasoned debate can be inferred from such balanced and
objective comments as " M.I.T. is the Defense Department's house
whore,...". Others may remember actual details of the Harvard
discussions better than I do, but Harvard eventually decided that
there would be no formal Harvard-as-University participation, but that
interested departments and researchers at Harvard were free to
participate and accept funding.   Many did -- there were at least
three Harvard senior faculty, from at least Schools on the internal
advisory committee and far more on a large faculty (and probably some
students -- don't remember offhand) advisory group.   So, we ended up
with a central staff at MIT with work focusing on a general
architecture and software substrate for a wide range of applications,
integration of a variety of tools, data representation issues, design
and construction of a researcher-friendly and statistically-oriented
database management system, and a good deal of work what was necessary
to apply different kinds of tools and models to the same underlying
data.  Wrt the latter, a common attitude, and arguably the state of
the art, at the time was that people would build highly integrated
"statistical packages" with a particular view of data and that
researchers should design their work and hypotheses around what could
be done with one of those packages.  One of the key ideas behind the
Cambridge Project was that it was important to have an environment in
which data, models, and hypotheses should drive analysis not the
available tools (not at all z new idea, but one that was hard to
realize at the time).[1].[2]

It may also be relevant that the Cambridge Project was funded out of
ARPA Behavioral Sciences (sometimes Human Resources, IIR), not IPTO.
There were certainly some conversations at/with RADC about command,
control, and intelligence functions but they were more about the
applicability of our work to those functions than any focus of the
work on those topics.  Mostly or entirely after the Cambridge Project
as such ended, a company that was more or less spun off from MIT
provided support for the systems that the Cambridge Project was
developed to several universities and commercial enterprises in the US
and Europe (and maybe elsewhere, but I don't remember) and to parts of
DoD, notably what was then OSD Program Analysis and Evaluation (main
application there was the DoD budget, not, e.g., warfare)..  As with
many other things funded by ARPA, there was far more effort to explain
possible specific military applicability of the research work rather
than its justification as research after the Mansfield Amendment (and
the transition to "DARPA") than earlier.  Like many other ARPA
activities at the time, the explanations changed more than the actual
work,  It occurs to me that some of those explanations might be the
foundation for the NBC reporting referred to below.

A few other things to add a bit of data and help parse facts from
misunderstanding or fantasy (I'm running out of energy and this note
is already too long or there would be a much longer list):

(1) I have no idea where Levine got his list of "data banks" that the
Project was going to acquire, maintain, and distribute.  I don't
remember such a list from any of the early proposal drafts, nor do I
remember any discussion of them during the summer study.  In any
event, while individual researchers almost certainly had their own
data of interest and saw some of the work of the Project as providing
better tools for analysis and modeling of them, there was never any
central archive or effort to build one -- I'm quite confident about
that because it almost certainly would have been in my area of
responsibility.

(2) The document at http://www.dtic.mil/get-tr-doc/pdf?AD=AD0783626
was one of a collection of annual and them semi-annual reports.  They
are all public; they were all available through NTIS and probably
still are, although some of the scans were, IIR, even worse than this
particular one.   In any event, I have the MIT-produced paper versions
of all of them.  If the NTIS copies are no longer avaialble and
someone has appropriate scanning resources, I'd be happy to make them
available.

(3) There was never any "Project CAM" or something referred to that
way, at least in conjunction with the Cambridge Project.  The only
times I remember hearing that term during the Cambridge Project's
existence were in conjunction with a conspiracy theory (whose details
I don;t remember) involving "MAC" spelled backwards.

(4) During most of its existence, the Cambridge Project was on the 5th
floor or what was then 575 Technology Square, across the plaze from
545 (before that space came together, there was a group in MIT
Building 26 near the original MIT computer center facility, I
continued to sit in 545 Tech Square, etc.   That is relevant to the
Datacomputer discussion because we had the south side of that floor
and they had the north side.   But, if I remember (and my memory is
very vague about this), while Tom Merrill was PI on that project, I
think CCA continued to do business out of their other offices (up near
Fresh Pond and a few blocks from BBN).  Could easily be wrong about
that, but IBM never had anything to do with the Datacompiuter -- it
ran on PDP-10s, Ampex videotape drives, and some specialized hardware.
What I do know is that, while the people involved knew each other
(common elevator lobby and shared history among the more senior
folks), no data ever moved between the two projects although Lick and
others had a lot of fantasies about that if and when the Datacomputer
work ever reached useful production status.  Also, IBM's Cambridge
Scientific Lab was definitely in 545.     The only two CTSS systems I
was ever aware of belonged to Project MAC and the MIT Computation
Center.  They were networked via  the high-bandwidth method of people
carrying magnetic tapes a block of two :-(  I don't think IBM every
actually owned one, although I might not have known.    CP/CMS didn't
speak SNA.  It did acquire RSCS although I don't remember whether
before or after the transition to the VM/CMS product.   RSCS of course
became the primary transport protocol for BITNET.  Almost certainly no
ARPANET connections to the Cambridge Scientific Center, at least early
on -- the Host-IMP protocols didn't exist for the machine and there
weren't any spare ports on the obvious IMPs.   And the CIA office in
545 was a fairly open secret if it was a secret at all, at least by
the time I had an office there around 1965-1966.

    john


[1[  Klensin, John C., J. Markowitz, D. B. Yntema, and R. A. Wiesen,
“The Approach to Compatibility of the Cambridge Project Consistent
System”, ACM SIGSOC Bulletin, Fall 1973.
[2}  Klensin, John C. and Douwe B. Yntema, “Beyond the Package: A new
approach to social science computing”, Social Science Information, 20,
4/5, (1981), pp. 787-815.



On Sat, Apr 14, 2018 at 11:26 AM, Eric Gade <eric.gade at gmail.com> wrote:
>> Lick was my thesis adviser, and
>> subsequently my boss when I was a member of the MIT research staff in
>> his group.  So there is a lot of overlap between my personal experiences
>> in 1969-1977 at MIT and the events and interactions chronicled in "Dream
>> Machine".
>
>
> Hi Jack, thanks for writing back. It's great to have a person who worked
> with Licklider be a part of this email record.
>
> Also thank you to all the others for responding. I want to clarify a couple
> of things, mostly because I don't want to be unfair to the book's author
> despite my evaluation of his research. Levine seems to suggest that there is
> some connection between counterinsurgency psychological/sociological
> research in Vietnam and the origins of Licklider's research group(s) and
> work in building the ARPA C&C/IPTO community. That is to say, he believes
> there are common intellectual origins if not necessarily applications. What
> has been covered by Waldrop and others -- and what is even apparent in the
> oral histories recorded by Licklider and others -- is that to the extent
> this is true, there was apprehension on the part of the interactive
> computing researchers. Either way, this is a bold claim and my own feeling
> is that it requires much more evidence to support it.
>
> The NBC reporting is -- to his telling -- evidence of similar tactics being
> used on the ARPANET, although the Congressional Record testimony seems
> pretty clear that the report confused a bunch of things. Again, it doesn't
> seem to me that enough convincing evidence is presented, but these reports
> are interesting nonetheless and I'd never heard of them before in my own
> research.
>
> One final note about the Cambridge Project. Waldrop also discusses the
> Cambridge Project in "Dream Machine" -- he even recounts a story where
> Licklider, surrounded by protestors who were attempting to burn copies of
> his proposal, showed the youngsters that they needed to fan out the pages if
> they wanted to get it to burn properly (and even lit his own report on
> fire). At the time, this was a known project. The Harvard Crimson even
> reported on it:
> http://www.thecrimson.com/article/1969/9/26/brass-tacks-the-cambridge-project-pi/
>
> As I mentioned, I have not been able to get a copy of this proposal. The MIT
> archives will almost certainly take their time getting back to me. The
> citation Levine uses for the report is:
> J.C.R. Licklider, "Establishment and Operation of a Program in Computer
> Analysis and Modeling in the Behavioral Sciences" December 5, 1968. MIT
> Institute Archives and Special Collections, MIT Cambridge Project records.
>
> Levine does not seem to quote from this proposal and only cites it once when
> he lists the "data banks" that the Cambridge Project would create (and "make
> available through ARPANET"):
>
> Public opinion polls from all countries
> Cultural patterns of all the tribes and peoples of the world
> Archives on comparative communism [...] files on the contemporary world
> communist movements
> Political participation of various countries [...] This includes such
> variables as voting, membership in associations, activity of political
> parties, etc.
> Youth movements
> Mass unrest and political movements under conditions of rapid social change
> Data on national integration, particularly in "plural" societies; the
> integration of ethnic, racial and religious minorities; the merging or
> splitting of present political units
> International propaganda output
> Peasant attitudes and behavior
> International armament expenditures and trends
>
> (It is unclear is Levine is listing these himself or quoting from the
> proposal; without seeing a copy we cannot verify)
>
> My understanding is that the project ran for ~5 years. The only documentary
> evidence for it that I've been able to find online is the following report,
> presumably written near the end of the project:
>
> http://www.dtic.mil/get-tr-doc/pdf?AD=AD0783626
>
>
> Without some revelation from people on this list, I don't see enough
> evidence to overturn the narrative clearly expounded by Waldrop, Weinberger,
> and others that the ARPA computing community as established by Licklider was
> a kind of lucky moment where lots of funds could be spent on risky/open
> projects and that most of the rest of ARPA had little idea what these guys
> were even doing, let alone others within the Pentagon.
>
> On Sat, Apr 14, 2018 at 7:15 AM, Dave Walden <dave.walden.family at gmail.com>
> wrote:
>>
>> Jack,
>>
>> My memory is that CCA (Computer Corporation of America -- Tom Merrill's
>> company) did the DataComputer.  They might also have been at 545 Tech
>> Square at the time but I am unsure of that.  IBM (the "Cambridge
>> Scientific" lab?) was also there (as you note) and did other important
>> things (my memory is vague, so I am uncertain of the following things
>> ... the beginning of CP/CMS operating system, Script text processing
>> system, creation of GML, I think they may also have had the other CTSS
>> system, etc.) but I don't remember this group being connected to the
>> ARPANET (IBM was pushing SNA -- proprietary networking).
>>
>> Dave
>>
>> On 4/14/2018 3:36 AM, Jack Haverty wrote:
>> > Lick's group was part of Project MAC, aka LCS (Laboratory for Computer
>> > Science),  It occupied part of 545 Technology Square, along with the MIT
>> > AI Lab.  LCS had many subgroups.  In addition, the building complex
>> > housed an IBM research group (that did the DataComputer, which was
>> > attached to the ARPANET), and even a stealth office of the CIA (really -
>> > but that's another story), which I accidentally "outed" one day while
>> > trying to run computer cables up to the roof through the elevator shaft.
>> >   Oops.
>> >
>>
>> _______
>> internet-history mailing list
>> internet-history at postel.org
>> http://mailman.postel.org/mailman/listinfo/internet-history
>> Contact list-owner at postel.org for assistance.
>
>
>
>
> --
> Eric
>
> _______
> internet-history mailing list
> internet-history at postel.org
> http://mailman.postel.org/mailman/listinfo/internet-history
> Contact list-owner at postel.org for assistance.
>