Anecdotes From the VNET Era
Edson Hendricks,
inventor of the VNET/NJE Network. More about Edson Hendricks here:
About
the interview with Edson Hendricks, investor of the VNET/NJE Network:
author Leanne Jones
interviewed Edson Hendricks in person, on the phone and via email. In the
editing and fact-checking stage, editor Bruce Batchelor
sent dozens of emails to Edson as well. The following explanations and
anecdotes by Edson were gathered from the email stream.
Chronology:
[Edson
writes:] By 1974 VNET was already in 24/7 operation all over IBM,
including IBM staff who were working on-site at IBMÕs
larger academic, government and corporate customers. At that time, almost
*nobody* outside IBM had any idea that IBM had a big internal computer network.
We kept it pretty quiet. I recall that in our meeting (me, Vint Cerf and Jerry Saltzer in
the summer of 1975 in VintÕs office), Vint was very surprised to learn about it, and he seemingly
had no idea what it was or how it worked. He seemed extremely convincing
to me that it all came as a big surprised. Vint Cerf
did not mention to me when we met anything about TCP/IP that I recall, nor
do I recall it coming up at the IIASA conference in Austria in August 1975 at
which Vint and I were the only USA attendees. I was
on tour in Europe in connection with the first IBM official product
release of my networking software. (The software was called RSCS then –
it got named VNET later.) The conference was at the International
Institute for Applied Systems Analysis, IIASA, in Baden, Austria, where I gave
a paper and the first public explanation of the new IBM internal computer
network. I donÕt recall hearing about TCP/IP until around 1979.
At that IIASA conference the Europeans were all over me, because
they could see that I was talking about a connectionless network, and wasnÕt I
aware of the ÒCigaleÓ work Louis Pouzin was doing with that? No, I wasnÕt, until then. My
unfamiliarity with Cigale attests to my total
unfamiliarity with Cigale and Cyclades until
1975, after my first ÒconnectionlessÓ design had been in regular operation
within IBM for almost two years. Cyclades/Cigale had
absolutely no influence on my design at all, because I was unaware of
it.
IP:
The whole point
of IP is the combination of connectionlessness and
datagrams. Datagrams are free-flying packets with destination addresses, not
attached to any connection. But no, I did not invent the term ÒIP,Ó nor had I
heard of it until almost 1980. I think Vint Cerf
coined that term, probably some time in the mid- to late 1970s, unless someone
can actually cite any record to the contrary.
And yes, VNET
always had this. ThatÕs why VNET had Òinstant messagesÓ while the ARPANET did
not; it was pretty much impossible to implement any Òinstant messageÓ facility
with the ARPANET, because there were no Òdatagrams.Ó Those Òinstant messagesÓ
were a natural for VNET, it was pretty much central to the design of the whole
thing.
The
Canadian Connection:
Before my
VNET, there was pretty much nothing in IBM for communication, and certainly no
networks. SNA never appeared, really. It was a concept, you see, and a really
flaky one. Sure, IBM trotted out a bunch of junk named SNA, but it was all there earlier under other names, and none of it was around
when VNET appeared in 1973.
The
ways VNET made money for IBM (and everyone else) are innumerable, they were everywhere. Everything, all the
time. I could go through a long list, but my fingers are getting a bit
tired. IÕll offer an example involving Canada for you. Some time back there
Canada had a postal strike. I got a phone call from someone in IBM Canada, who
had heard a rumor that a computer network existed, and wondered if that was
true. I said yes, itÕs true. This person told me that they had a system where
weekly software updates on tape were mailed from the USA to Canada, and so they
couldnÕt get them. He wondered if this network could help. I said sure, and
asked him if he had a pencil and paper. He did, so I told him what to type, it
was real easy. They tried it, and voilˆ, it worked perfectly! Problem solved. And you know, once they
found out how to do it, they never went back to the mail.
This reminds me
of the timeÉ. Preparing for the IBM VNET PRPQ release, we had a little group
from IBM Toronto doing the documentation. I had to go see them, and I had with
me a big computer tape of the software as it was then,
around 1974. I went through Canadian customs at the airport where I showed the
guy the tape, and he asked me what it was worth. I asked him if he meant the
commercial value of the tape reel, or the value of what was on it. He replied,
the value of what was on it, the VNET software. Well, that was before it was
released, so there was no actual market value. I was stumped, and there was a
line of people waiting behind me, so I decided not to pursue it with him. I
just told him $100. He said fine, and I had to pay him $10. I said fine, and
gave him the $10. Of course, I knew perfectly well that for many months before,
all sorts of such stuff was flying over VNET from the USA into Canada with no
duty at all!
Revising
history: why was VNETÕs role suppressed?
My best guess
is that it has to do in part with ARPA/DARPAÕs pride. They were real proud of
themselves for the ARPANET, although I donÕt think they had much reason to be.
They had failed, after all, to bring computer networking to the wider world
that didnÕt have the budget of DARPA and the US government. The ARPANET *never*
made money, it *cost* money. There were several attempts to turn it into a
commercial enterprise, one by Larry Roberts himself, and all of them failed,
because *it couldnÕt make money*. And IÕm sure DARPA was pretty embarrassed
when the universities worldwide started organizing their own computer networks
using my software, or other software copying my approach to it. It was doubly
embarrassing for DARPA that the network design had come from IBM!
Remember also,
when IBM executive management (specifically IBM Vice President of Engineering,
Ted Climis) found out about VNET, they were the last
to know of course, they ordered an investigation to learn how it happened. My
own local management were terrified about that, but I
thought it was hilarious and just great. So I went way out of my way to help
them investigate, and when they got done and submitted their report to the top
executives, I got a copy and I almost fell out of my chair. They found the
reason for VNETÕs success! It was money, of course, they figured it out! And not only that, they made an
estimate of the amount of money VNET was making for IBM, and it was at least in
the hundreds of millions of dollars! *Then* the IBM executives *really* hated
VNET, because they obviously couldnÕt get rid of it, and it wasnÕt SNA. Damn.
DARPA has built
this fiction that the Internet arose out of whole cloth from DARPA only,
because thatÕs how theyÕd like it to read. And they *should* have done it, but
they failed, and I did it first. And with IBM suppressing VNET and them
scrambling to get back in charge, thereÕs your history. So I think it makes
good sense why they canÕt remember finding out about VNET long before the
Internet appeared, thatÕs the way they want it. Seems obvious to me, it was all
a coincidence, you see É right? Only the dates are wrong.
Bronze Placque
The date on
which my own full-blown networking version of the VNET software was publicly
released was April 15, 1977. Believe it or not, this is on a *bronzed plaque*
of the original IBM release letter, which my management gave me at the IBM
Cambridge Scientific Center!
At that point,
VNET became publicly available, and right after that followed all of the
networks that actually used it (BITNET, NETNORTH, EARN and others) as well as
additional networks (UUCP/USENET and FIDOnet) that
adopted pretty much the same design approach. ARPAÕs TCP/IP Internet didnÕt
appear until at least three years after that, and I think it was actually more
like four years.
Interview
reprinted without permission after many attempts at contacting the Interwiee and the Interviwer
moshix
April 2020