[ih] A paper

[Jack Haverty]

Hello Andrew,

Thanks for a very cogent explanation.

There's a few observations I think many people haven't considered.

IMHO, the time period of the early Internet, roughly 70s/80s, was a 
unique time in history.  The traditional methods of scientific 
discussion and debate had relied on papers, letters, 
conferences/proceedings, and other such "documentation" for centuries.   
But that changed with the advent of computer networking.   I was a 
student, and later a staff member, in Professor Licklider's group at 
MIT, and so thoroughly indoctrinated in his vision of using computers 
and networks to augment human interaction.   So that's what we did.

As the ARPANET grew, we naturally shifted our behavior to use "the net" 
as much as possible.   Instead of writing papers, we wrote emails.   
Instead of publishing documents, we put them on FTP servers.   The 
traditional methods increasingly atrophied, within the "ARPANET 
Community", as electronic interactions improved.    As a student, few of 
us could afford many subscriptions to journals, and the library stacks 
where you could read them much less convenient than the timesharing 
terminal down the hall, a portal into the net.  As the Internet emerged, 
and in particular CSNET and NSFNET expanded the connectivity into 
institutions not involved in building the network, I suspect the same 
transition occurred.

Unfortunately, computer storage was expensive, so you couldn't put too 
much online or keep it there for posterity.  Sadly, a lot of that 
"documentation" from the 70s/80s has now disappeared.     By the 90s, 
computer storage costs had dropped dramatically, and most importantly 
the World Wide Web entered the scene.   That made it possible for people 
to put much of their "documentation" readily available online and 
perhaps fragile mechanisms such as archive.org are preserving it for 
posterity.   Sadly, few of the organizations managing the computers of 
the day apparently saw any need to preserve things like archives of 
mailing lists or email exchanges or contents of the FTP servers.   Even 
the Datacomputer, where the mail system I built in the early 70s 
deposited much "documentation" for safekeeping, apparently didn't 
preserve it.   At least not that I know of.

So, IMHO there is a period of perhaps 15-20 years in the 70s/80s where 
much of the "documentation" to support historians is simply lost, except 
for "interviews and recollections".   Formal documentation was sparse, 
and the newfangled informal documentation transiting the network has 
been lost.   That may be a cause of the scenarios you describe of 
students experiencing the "I was there" reaction.

Even before the 'net, I think the historical record is incomplete with 
just the formalities of public "documentation" like papers as sources.  
In particular, many organizations don't publish much of their activity, 
whether for protection of their Intellectual Property or to avoid 
embarassment, or simply because there's no business advantage in 
spending your resources on it.   So IMHO we mostly don't know what they 
did or why, yet their actions probably had a strong influence on history.

Personally I spent quite a few hours on the phone a few years ago with 
Brad Fidler relating my experiences with developments like EGP.   My 
comments on the paper were primarily trying to get the facts right about 
that chunk of history.   Like much of the activities in the 70s/80s, you 
won't find much that was in formal "documentation" about that; all we 
have is personal recollections.

IMHO, the 70s/80s was a unique period of time because of that transition 
that the network triggered.

A second observation concerns "what we were doing".   IMHO this is often 
misunderstood.   People seem to think that we were building the Internet 
we have today, and want to know why we screwed it up in some way.

I can only speak from personal experience.   But I certainly thought we 
were building a system targeted toward government and military needs, 
deploying it to see what worked, and expecting the results of such 
research to be possibly folded into the subsequent "real" mechanisms 
that were being developed by traditional means and organizations.

The Internet was a research project, and its deployment was an 
experiment.  AFAIK, few of the people involved in implementing the 
network mechanisms had significant prior experience in building network 
equipment.  The contractors were organizations such as MIT, BBN, SRI, 
and such.  None of the names you'd associate with data communications 
(e.g., ATT) were involved.   We were expected to "think outside the 
box", and to try new ideas.   If there were two candidate solutions to 
some problem, one that we knew could work from others' experience, and 
one new, untried, idea, often the choice was to try something new.  We 
implemented such mechanisms not because we knew it would work, but 
because we didn't know that it wouldn't work.

I don't remember anyone ever suggesting that we were building the 
communications infrastructure that would survive for 50+ years and 
blanket the entire planet.   That was OSI's job, but we could keep our 
research network going for a few years until they were ready. If we 
"defeated" the other guys, we didn't even know we were at war.   At 
least I didn't...

A third observation is on the "politics" of the environment, or perhaps 
what were the mechanisms and motivations that moved it along.

My perspective is that the early years (70s/early 80s) the main driving 
force was a "benevolent dictatorship", where ARPA was the dictator and 
exerted its influence by simple control of funding. Within that 
dictatorship, there was a lot of freedom to explore your interests, as 
long as they were plausible steps toward the research goals.   For 
example, since the Internet was (still is?) an experiment, there was a 
desire to run some tests.   Those tests required measuring time, which 
was not an easy thing to do in a system spanning thousands of miles, and 
no GPS satellites yet. Most of us were happy with the resolution of a 
second or so that could be obtained by the classic "synchronize your 
watches" approach and a telephone.   That wasn't good enough for Dave 
Mills, who exerted tremendous effort to create NTP so that the computers 
on the network could synchronize to milliseconds.   IMHO, Dave's the 
reason your smartphone and computers today know what time it is.

As the Internet expanded, more players appeared, and commercial 
interests extered their influence, things changed.  Personally, I have 
no idea today how decisions are made that cause changes in the operating 
Internet that we all use.   From the "documentation", you might believe 
that it involves the IETF process which creates Standards.   But I've 
occasionally asked the question -- How many of the protocols and 
algorithms documented in RFCs are actually present and operating in the 
Internet machinery that I'm using right now? I've never gotten much of 
an answer, or seen much "documentation" that reveals how the design 
decisions actually are made to create the network machinery I'm using 
right now.   Or who makes those decisions.

So, not knowing what that process is, or who is actually making 
decisions, I find it difficult to judge how "political" it might be.

Personally, I agree that the Internet today needs regulation - as long 
as it operates in some way as effective as, and reminiscent of the 
"benevolent dictatorship" of the ARPA/NSF days.  Maybe that's just not 
possible though.   One of the design principles of the network (which 
may not appear in "documentation") was that the network must not have 
any single point of control, no one in charge.  One tangible result of 
that principle was EGP.

That was important for survivability in military scenarios, so perhaps 
now we're simply stuck with it.

Or maybe OSI will appear next year.

Thanks for reading,
/Jack Haverty




On 7/18/21 2:23 PM, Andrew Sullivan via Internet-history wrote:
> [I think I sent this earlier from the wrong address.  Apologies for 
> any duplicates.]
>
> Dear colleagues,
>
> In the sprit of full disclosure, I will note both that I work for the
> organization that hosts this list but I'm speaking for myself, and
> also that I have a personal relationship with one of the authors of
> the paper that has caused so much discussion.  But I have an
> observation.
> On Sun, Jul 18, 2021 at 08:37:32AM -0700, Bob Purvy via 
> Internet-history wrote:
>> I would hope after all that, especially Jack and Vint and Andrew's 
>> awesome
>> summaries, you will just withdraw the paper.
>
> I think that would be a shame, because I think the paper is making a
> point that some on this list seem to be missing, but that is extremely
> important for anything pretending to be an Interhet _history_ list,
> rather than just Internet recollections.
>
> In a previous career, I thought I was going to be an academic, and
> I knew a lot of historians because I worked on what was then called
> history and philosophy of science and technology (I think the jargon
> is now "science and technology studies").  Young historians working on
> the 1960s and 1970s kept having trouble publishing papers because
> they'd submit to a journal and get back a review that said in effect,
> "That's not what happened, because I was _there_."
>
> Now, the problem with our past selves is that we can't interview them.
> We can only interview our present selves, who have all the
> retrospective knowledge and story-telling about what we did _then_ and
> what it meant.  That isn't to say such interviews and recollections
> are not valuable, but they're also not documentary evidence.  And that
> is I think an important point that is related to something the paper
> is arguing.
>
> Regardless of what people doing engineering think, there are a lot of
> people today who believe the Internet needs plenty of regulation, and
> who have become convinced that the Internet is (or maybe I should say
> "is only") a political instrument.  This is one interpretation of
> Laura DeNardis's slogan, "Protocols are politics by other means."
> (For whatever it's worth, I think that interpretation doesn't hold up
> to a close reading of DeNardis's original text, but I have not noticed
> that popular discourse is much affected by close readings.)  The
> Badiei-Fidler paper is making the point that such a claim is poorly
> justified given the history of several protocols, and that indeed the
> historical record doesn't make it plain that the people designing
> things _themselves_ had perfectly clear interpretations of what they
> were doing at the time.  I don't know about you, but my experience has
> been that I often only really know where I am going after I get
> there.  This is in part because it is the effort of doing the work
> that reveals what compromises need to be made in a technology.
>
> Indeed, as several have pointed out in this thread, that was one basic
> problem behind the OSI approach: it appeared to want a
> fully-worked-out design that "everyone" could agree on before anything
> could be built and shipped, and the result was that the Internet
> people shipped first, and so that's what took over.  And it must be
> admitted, I think, that there is something at least partly political
> in such a decision: one approach valued Officially Approved Agreement
> and controlled distribution of documents over getting working things
> in the hands of those trying to make stuff work.  The Internet
> approach instead was to try things in the open and share documents
> as widely as possible.  I'm not sure what to call that kind of
> organizational decision other than "political".
>
> Yet, as Badiei and Fidler argue, there's no _inherent_ politics that
> is legible in any particular protocol.  To move from, "Some Internet
> protocols have politics built in," or even, "There is a fundamentally
> political decision in any organizational form," to, "All protocols are
> inherently political and must be designed to promote certain kinds of
> values," is hard to square with the history of several protocols as
> documented in the historical record.  That might not be surprising to
> people on this list, who problably didn't think they were practicing
> politics by other means when designing the networks.  But I think
> Badiei and Fidler are trying to make that case to scholars in other
> fields, who are busy interpreting the history that many here lived in
> a way that might not be good for the future of innovation on the
> Internet.
>
> Best regards,
>
> A
>