Boeing and The Internet - Part I

In other parts of this site, there will be some observations by me about how the internet can be improved.  The additions that I propose, mostly to the lower layers of the protocol stack, may sound a bit audacious.  After all, who or how can needed change possibly be brought to such a global monster?  That's a fair question.  Let me offer two answers.  One, is that was the nature of the work I got into when I transferred into BCA in 1997.  And the other answer is that from its inception in the late 1950s (perhaps a surprising date to some who normally think it all started in 1964) through the end of 1992, Boeing was deeply involved in both the gestation and the evolution of the net. 

 

The first part of the story about Boeing's involvement in the creation of the Internet starts way back in the late 1950s. Boeing designed and built the first packet switched routers to be used in a practical application (Minuteman HICS).  This system would provide the starting point for the mid 1960s project to build ARPAnet.  The systems engineers in Boeing would make many small contributions to the evolution of the net between 1975 and 1988, but then again in 1989 Boeing would play a key role in a transformative change that made the modern Internet possible.  As chance had it, I just happened to work in the parts of the company that was involved in those things. 

 

In some quarters of The Boeing Company of the totem, we believed quite strongly that the net is still a work in progress.  Those of us who worked in those areas still believe that to be the case.  So while I admit that my proposals to make the net safer, more egalitarian, and capable of enabling applications that are currently impossible, will all sound impossibly audacious to many, I feel quite comfortable in making such proposals, because I've seen similar big changes up close.  So let's start by talking about some of the resistance that must be overcome, and then walk through a little bit of Internet history (I'll explain the mixed capitalization below).

 

In many IT organizations, the folks who do network engineering and construction do one thing very well, and another thing very poorly.  The thing the do well is build and maintain the network infrastructure using quality equipment.  They generally do a great job of provisioning capacity on a just in time basis, and networks tend to be very reliable.  If one were to give the typical network team or organization a grade for this part of their job, it would usually be an A.  In fact, often I think that network engineering is a part of the typical IT organization that routinely out-performs every other part of the whole.

But, the networking teams also tend to be self-isolated and know next to nothing about the actual businesses they serve, and often seem to take no interest in learning about them.  And heaven forbid that someone suggest integrating their network management tools into the operation applications of the business.  The big exception to this is the carriers. 

In the carriers, network management tools are tightly integrated into the business operations of the company.  Metered usage data and location data are but two examples of such integrations.  But try to suggest that an enterprise network team actually participate in such an integration with their tools, such as Cisco IOS, and one would be lucky to even get them to come to a meeting to talk about it.  They tend to get away with this nonsense because IT departmental leadership tends to know nothing of network engineering.  There is no career path that starts in network and leads to CIO, and it shows.

Network systems are data-rich environments that have much to offer the enterprises they serve.  After transferring to BCA in late 1997, the way I saw my job was to try to tap that value for the benefit of the company, especially in the factories.  Maybe I would get back to working on helping the company better communicate with itself one day, but for now, the task that took be to AR&T was on the back burner.  I did make an initial effort to take my projects with me, including the experimental servers that I had in my office in AR&T, but they soon pulled them back and shut down all communications improvement projects.  AR&T just wasn't going to go there any more. 

 

From my new home in BCA, I did have a voice in trying to set their research agenda, but since they no longer had anyone on staff who had any expertise in the communications area, there was no point in pushing on that, at least not for a while.  But in many ways, my higher level mission remained unchanged.  Whether it was working with the research teams in AR&T or doing battle with the BCS network operations teams, the goal was to make the factories, and to a lesser extent, the engineering areas of the commercial airplane company more productive in a continually improving way.  It is the job of infrastructure to facilitate, so let's get cracking on doing a better job of facilitating.

Boeing had a long history of network pioneering, starting with the part of the company where I started my new career.  That was in Ballistic Systems Division, the part of Boeing that designed and built the Minuteman missile system.  Boeing designed and built most of the system except for the actual missiles and their payloads.

Where Wizards Stay up Late by Katie Hafner and her late husband Matthew Lyon is generally considered to be the definitive history of the origins of the Internet.  However, there is one thing the got wrong, and I've chatted with Katie about this a couple of times via email.  The problem Katie and Matt ran into was that secrecy thing again.  Several critical aspects of the deep origins of the net were among the most classified, Top Secret - Need to Know, bits of technology the United States had at the time.  Some of it simply wasn't talked about.  And other parts were hidden by a kind of cover story, although not as elaborate as the Boeing Sonic Cruiser dog and pony show.

Another curious thing about the history of the net as it is now widely believed, is that most of the people who claim to have been there and in the know, actually did not get involved in it until about 1970, a good four or six years after the project was formally kicked off depending on which event one calls day 1, and thirteen years after work on the prototype network upon whose architecture the ARPAnet project was based.

The Wikipedia article I've linked here is a good basic discussion of one half the secret keeping approach used by the U.S. military.  The other half is the granting of clearance levels to individuals.  One might have a Secret or Top Secret clearance, but that does not confer the right to know, nor access to all data classified as such.  One also has to have a need to know or to be in a place that has controlled access requiring a given level of clearance.  One can't just go poking around out of curiosity in classified areas, the way I typically explored Boeing's facilities.

On October 4, 1957 the Soviet Union launched Sputnik I into low earth orbit.  As a satellite it was not that big of a deal.  Few early satellites were.  What was a big deal was that they could put something in orbit.  That meant they had sufficiently good computer guidance systems for their rockets that they could drop a bomb close enough to our Titan ICMB sites to destroy them in a first strike.  The U.S. military establishment went into panic mode.

Air Force Colonel Edward N. Hall campaigned for the idea of using a lot of smaller missiles distributed in what he called a farm.  As I understand it from being told by the engineers with whom I worked, who in turn were mostly the second generation of engineers working on the system, it was J. C. R. Licklider who first suggested what he called a "hot potato" communications network to link all of the silos together with their launch control centers.

It was Paul Baran who worked out the architecture of how such a system should work, using what came to be called "packet switching."  It was Boeing that got the contract to design, build, and install the equipment to make it come to life.  The network was called the Hardened Intersite Cable System or HICS.  Finding pictures of the cables is easy.  Finding pictures or discussions of the boxes and software to which they were connect - well not so much.

But before Boeing could validate the first silos as operational, U-2 spy planes flying over Cuba took photos that led to the start of the Cuban Missile crisis.  The fate of the entire world was literally hanging in the balance, as Boeing crews rushed to get Minuteman operational.  I cannot stress enough how sensitive this stuff was.  Even 30 years later when Katie and Matthew were researching their book and interviewing people, they still did not talk about HICS.

The original HICS equipment and software was replaced a long time ago.  By today's standards, it was quite primitive.  So what I am about to describe is no longer sensitive. 

The system as Baran conceived it, and as Boeing engineers built it, had all of the key features of internet routing, but for a closed and well documented system.  The robust flexibility of BGP and TCP were still very much in the future, and were not needed.  The boundaries of each field of approximately 200 silos or launch facilities (LFx) and 20 launch control centers (LCCs) were fixed.   

 

BTW, let me explain my mixed capitalization of the word internet.  Up until about 1990, we always capitalized the word like it was some sort of proper name.  But since then, not so much.  Besides, its more of an idea than a thing.  There is no "thing" called the Internet.  There are just a bunch of protocols, and business necessities for interconnecting networks that allow the whole to function as though it was a thing.  But the substance of it is a lot of privately owned and operated networks, many of which have their own peculiar features.  I tend to capitalize when talking about the net prior to 1980, and not for things after 1990.  In between, it gets mixed a bit.

As Baran's concept was implemented, there was no notion of a transport layer (layer 4).  Rather, a packet was just a 56 bit "word."  Each bit could be assigned some special meaning.  If a bit were flipped from 0 to 1, that might be a request for silo status data, and so on.  Not all 56 bits were ever assigned functions.  

On page 42 of Katie's and Matthew's book, they describe the meeting in Charles Herzfeld's office in the Pentagon where Bob Taylor asked him for the funding to build what would become the ARPAnet, the starting point of the Internet.  Now let me quote directly from the text.

"is it going to be hard to do?" Herzfeld asked.

"Oh no,  We already know how to do it," Taylor responded with characteristic boldness.

"Great idea," Herzfeld said.  "Get it going.  You've got a million dollars more in our budget right now. Go."

Taylor left Herzfeld's office in the E-ring and headed back to the corridor that connected to the D-ring, and his own office.  He glanced at his watch.  "Jesus Christ," he said to himself softly. "That only took twenty minutes."

So what was it exactly that Taylor thought "we already know how to do?"  The basic lower layer architecture of course.  They would start with the Minuteman HICS architecture, and somehow figure out how to make those 56 bit words into flexible length packets that would have the ability to transport any string of 1s and 0s that was desired.  But, this was not a job for Boeing.   Also, they would have to figure out how to interconnect two private networks, using some sort of a boundary gateway.  And, there would have to be a flexible addressing system that had unlimited growth potential.

The Minuteman HICS system did not have an easy flexible way to link one missile field's network to another network of any sort.  There were two other links, one to the Looking Glass airborne launch control centers, and to a satellite communications system.  I think the details of both of those are still classified, so nothing about them is said here except to say that they did not provide a foundation for the Border Gateway Protocol (BGP) that provides the network to network interfaces that makes the Internet what it is.  But, Minuteman HICS was the starting point.

 

Boeing had its hands full with other things.  So the contract to figure out those things and get ARPAnet going actually went to the small engineering company where Licklider had worked before working for the Department of Defense in the Defense Advanced Research Projects Agency where Herzfeld was the director of the Ballistic Missile Defense Program.  That was Bolt, Beranek and Newman, or BBN.  Paul Baran was at the Rand Corporation, the consulting firm tied to both.  Also, the way Bob Taylor was thinking about the ARPAnet project suggested that he needed to start with a core team of university computer communications researchers, and not the aerospace industry.  They would be added later, once they had a minimally functional starting point.  So for the first decade of the ARPAnet project, Boeing and the other aerospace companies were not directly involved, except for a bit of overlap with some researchers who held down both corporate positions in aerospace and academic positions with some of the schools that were involved.  BTW, this overlapping is not insignificant, and is yet another example of why Aerospace doesn't fit the team versus family concept promoted by the folks at Netflix and for very different cynical reasons by the minions of Jack Welch.  That is discussed elsewhere on this site (Netflix).

From this point on in Katie's and Matt's telling of the history of the net, I don't see any issues, provided that one smiles at the attempts by some to deny that the architecture of the net was somehow designed to survive a nuclear strike.  In fact, that was exactly the whole point of the entire Minuteman system, including the HICS and the electronics to which it was connected.  That robustness has made it very difficult totally block internet access in a country experiencing some sort of crisis, either manmade or natural in origin.  If one interconnected private network is down, there is bound to be another one that isn't.  And satellite network interfaces are basically impossible to detect, unless the connected node also has a GPS receiver, and the GPS data is being fed back through the internet link, and that data is being received someplace where an application is identifying it as such, making it available for use, and someone with access cares to look at it.

So that's the story of HICS as it relates to the gestation of the Internet and Boeing's role in it.  And as far as I know, this is the first time that story has been told.