James L. Pelkey
Recorded: October 13, 1988
CHM Reference number: X5671.2010
James L. Pelkey Collection: History of Computer Communications
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 2 of 13
you get to be responsible for this and how did you come to be prepared to deal with that issue?
contributed to this misconception that -- the prime purpose of the AlohaNet was to connect the outer
campuses of the University of Hawaii, or in fact to really provide any operational facility at the university.
Even today, about 15 years after the project, we occasionally get groups that come in from Japan who
want to see the AlohaNet in operations, and I have to tell them that it really hasn't been in operation for
over ten years. The original project had, as its goal, research more than operational needs of the
university. I think it's possible that we may have justified the research, as researchers often do, by
pointing to its possible applications for the University of Hawaii and for other areas that had difficulty with
telephone communication of data, but certainly the goals of myself and the other people who were
involved in the project were research goals rather than operational goals. I should say, though, that we
did put up terminals on he other islands, never as part of operational networks but, in so far as the Aloha
was not an operational network, but we had terminals as demonstrations on the big island south of Oahu
and on Maui and throughout other locations on Oahu too.
Pelkey: So you had multiple locations so that you could demonstrate the viability of this mutually
launching out the network and it working?
extend the range by repeaters, Aloha Repeaters.
engineer. I was trained as an EE and I taught EE at Stanford for six or seven years
Pelkey: Prior to
Abramson: Well, I spent one year after Stanford at Harvard, but I was really teaching EE there rather
than computer science, and then I went to Hawaii.
Pelkey: What year was this?
Abramson: Twenty-two years ago, I believe. It was '65, '66, that's about right.
Pelkey: When did you get interested in this particular problem? How did it come to you?
Abramson: The way it came to us is that when I went to Hawaii, of course it was taking a pretty big
gamble. Hawaii was and is no Stanford or Harvard, and at that same time, there are other competing
advantages of working in Hawaii that are important to me, in spite of the fact that the intellectual
stimulation at Stanford and Harvard are also important. There was very little research going on at the
University of Hawaii, and I realized that we would have to start some kind of research project. At about
the same time, within the first year after I was in Hawaii, the Department of Defense started -- or sold to
Congress -- a fairly large research program -- this is 'large,' now, for the '66, '67, '68 time frame -- to, if I
want to put it more bluntly than DOD did, to support developing universities. That is, it was research
support of the project type specifically addressed to trying to bring universities such as the University of
Hawaii into the first rank. We realized -- myself, Wes Peterson, Ned Weldon and others -- realized that
this would be a good opportunity for us -- these are all people who went to Hawaii at about the same time
-- to obtain the resources, the funding, that we needed for this kind of a research university. So we said:
"Ok, what would be a good project for us?" Now, my background was communication theory, but I was
just getting interested in computers. I, in fact, was teaching some stuff on computers at Harvard that
year, the first time that was done in the general education program. Wes Peterson had done a lot of very
creative work in related areas and Ned Weldon and so forth, and so we cast about for a research topic
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 3 of 13
we were all communication theorists, we knew, we thought, a good deal about communications, that the
telephone system, especially then in Hawaii, was inadequate for data. Hawaiian Telephone, at that time,
was an independent company. It has since been acquired by GTE, but it had, I think it's fair to say, I think
most people at Hawaiian Tel would agree, they had pretty poor service, sort of part way between North
America and Europe in quality. You know how that is. So we put all this together, our interests and
experience, capabilities in communications, the technology seemed to be going towards computers, and
said: "Well, communications for computers makes sense." The telephone system appeared not to make
sense at that time, especially in Hawaii, and we thought we had something which was intellectually
stimulating and a package that we could sell to ARPA, and in fact, we did. That's how it all started.
ARPA, after I heard about the program, going to Washington
Pelkey: Right, and who was -- was this the IPTO office you went to?
where I had come from the University of Hawaii, rather than from Stanford or from Harvard. I probably
was a little touchy about that.
Pelkey: This would have been in '66?
Abramson: '66, '67, something like that. The program was called the THEMIS Program.
Pelkey: And this is the whole program of DOD to fund the universities.
Abramson: It was, I don't know, in the tens or maybe a hundred million dollars, or something of that sort,
to provide university research support for second rank universities. I came in to talk to Bob. I don't know
how I had been referred to him, some mutual friend in Washington, and Bob said: "Oh, yeah, we're
building this great thing called the ARPANet." No nodes were up at that time, and I said: "Oh, what's that
all about," and he went into some detail about it. I recall Bob was very blunt about 'what the hell is the
University of Hawaii going to try to do on this?' I recall being
Pelkey: A bit defensive.
conversations about it, and Larry too, and I think we all got a pretty good laugh out of it after the fact, but I
was pretty offensive at that time. I was probably more sensitive than I should have been. In any event, I
recall some of the remarks he made and some of the remarks I made about the whole thing that weren't a
very good way to start off any relationship. Never the less, we all seemed to rise above that, and I think
that's a pretty good indication of Bob and Larry's capabilities, and they ended up providing a very large
amount of money to the University of Hawaii, larger than the university had in any other project, to
perform what we said we wanted to do. That's how Aloha started.
part of the funding project?
think it's either the most important idea or the idea we finished with, because I think what happened is that
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 4 of 13
over a period of two years, perhaps, realized that what was important about what we could do with radio
was that it was a broadcast medium. We could do this multiple access thing that led to CSMACD, and
that was what was important. In fact, I recall, in some meetings to justify the project when we were still
more hung up than we should have been on radio, in discussions saying some to the effect: "Well, we
don't have to be limited to one channel per user. We could do some more efficient things with radio." We
still didn't have the full idea of what we were going to do, and I know that Bob and Larry both took up on
that and said: "What?" and I had to admit I didn't know exactly what, but I said: "Something much more
sensible for radio can be done here than assigning a single channel for every user in the network. That's
crazy. That won't work."
were aware of their developing paradigm of passing data that has this bursty nature, and that you wanted
to do something more than set up a channel in that domain? Just like in your world, it was better than
having a frequency assigned for every user? There were similar sorts of conceptual problems.
open as I could be in trying to understand them, and vice versa. So yeah, they were providing those
ideas. I don't know at what point anybody realized that there was a commonality of ideas there. It's hard
to say, but again, it didn't come, at least to me, in any blinding flash.
became obvious or that you became aware of the fact that you had found something connected to this
Abramson: I recall a meeting at the University of Hawaii, without Larry or Bob -- and Bob was leaving
the organization at that point -- but I recall a meeting where precisely this point was discussed. We said:
"Look, we can't assign one channel per user. We want to think about -- although we may never build it - -
we want to think about a system with hundreds of users, something practical for that. You can't have
hundreds of channels. Now what can you do for that situation?" I and others were aware of the spread
spectrum and multiple access through spread spectrum at that point, and the idea of simply transmitting
the data in bursts was sort of a natural one. I don't know how, it just sort of came out, and once that had
been suggested, it seemed awfully attractive. We mulled it over for a while, and I recall thinking about
should we do this or should we do other things, and very quickly we said: "Gee, that's going to be easy to
implement. It looks like it's going to work out well." We had no theory at that point, and I recall, at the end
of a meeting, saying: "Ok, let's go in this direction." I recall a meeting at which we made that decision.
Pelkey: Was there a sense of either frustration before this meeting about what you were going to do or
how you were going to solve the problem, or had the process of how you were going to implement
followed fairly quickly after the funding and you had just been workman-like about it, leading to this idea
coming up? Had there been a period of struggling with how to do it?
of that. I think that was pretty nice. Things fell into place very -- it was a good idea, it was a simple
design, and I think Bob Metcalfe had made the same point. There's a lot to be said for simplicity other
than just the fact that you can build it and maintain it more easily. There's a lot for conceptual simplicity,
because it leads to other things. It puts you in command of the ideas, and that's important. So I don't
really think we had any difficult times like that unless you want to consider the details of the
implementation. I know the engineers I hired were bright guys, and they had very sleepless nights, some
sleepless nights I'm sure, but I was insulated from that because I didn't know enough about it.
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 5 of 13
could apply it? In fact, it took a while for us to apply it. One of the reasons was certainly that -- my
impression, and I think this was common in a lot of people, was that to do something different with radio
communications means that sooner or later, you're going to have to fight the FCC, and I didn't want to do
that. I was faculty, a professor, and I truly felt I had no capability in that kind of area and I wouldn't do
very well at it, so I really couldn't see myself as trying to shake up the FCC and have them change their
rules. That meant that I was thinking of operating under the existing rules, and Aloha wouldn't allow you
to operate under existing rules. As a research project, it was quite interesting, but to look further to
operational and commercial systems
that a particularly big event to the group of you that had been working on this?
that to something to the effect that we're marking the success of the first phase of the Aloha project, and
everybody went: "Ok, now what's the second phase?"
Pelkey: Let's enjoy the first one before we go on to the second. Do you recall when that was?
Abramson: I could give it to you. I would say -- there was a paper that I wrote. Let's see, that was
minute, let me think now.
conference was in '70. I can get that date for you.
can give you exact dates.
was presented where?
that out too. I recall that it was one of the first conferences that I went to from Hawaii to the east coast,
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 6 of 13
a killer." I got used to it after that. It was an early one.
because he was looking for a project to work on, and that paper was seminal in terms of his thinking,
causing him to think about doing a ring network. It was the same IEEE meeting.
Abramson: Where was that in Georgia?
Pelkey: He just remembered that
although presumably Larry Roberts knew what was happening.
rather than industrial, so we weren't trying to keep it a secret. It's just that nobody was looking for an
awful lot of research to come out of Hawaii at that point.
Pelkey: What was the reception to the paper?
Abramson: Well, things don't get received with 'oh yeah, that's the way to go' sort of reactions ever. I
just have never seen that happen, no matter what the ideas are. Good ideas get accepted and
recognized and appreciated by a process of accretion, I think. Somebody looks at it and says: "Gee,
that's pretty good, but if I did this to it," and other people change it a little bit, and sooner or later it starts
affecting the way people think about certain kinds of ideas, and I think that's what Aloha was able to
achieve. It made people -- or let people -- think in terms of broadcast architecture, and that was a big
setting free from the telephone company.
Pelkey: Now, in talking to Bob, he tells me the story of Steve Crocker's apartment. You probably have
heard him tell the story.
reached over and picked up what turned out to be these proceedings with your paper in it. He started
reading your paper and he was infuriated with this notion that people just sit there and type, and he said:
"People don't do that. If people don't get response, they stop typing," and that would significantly change
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 7 of 13
that prepared because his PhD dissertation had just been rejected, and he was looking for some math to
beef up his PhD thesis that was on the ARPANet, and then he went to Xerox. When did the two of you
first come in contact?
project. Now, I don't know -- in fact I think I looked this up when I wrote my paper -- we somehow ended
up paying some trivial amount, maybe it was expenses or something. The project, at that point, was
probably near its peak. That is, we had, I would say, about 40 people or so, and either because I'm not a
good manager or because that's the way I manage, but whatever it is, I tend not to try to keep in touch
with everything that's going on in the project. I did at that point, the Reagan form of management. It has
its good and bad points. Now, Bob was introduced to me through Dick Binder. Have you run across his
graduate student. He had some industrial experience, and Dick eventually became the person
responsible for the protocols -- what are now called protocols; at that point, we though of it as software
and programs -- and the software structure of the network. I don't know how, but somehow he got in
touch with Bob Metcalfe, and then I found out that Bob Metcalfe, who I just knew as some graduate
student from, I think it was MIT or Harvard at that point, was coming out and wanted to work on the
network, Dick Binder wanted to work with him, and that he had done some interesting things, and I said:
"Great," but I didn't make that decision, I don't think. Bob came out and he was working more or less
independently, but with Dick, I think. He certainly wasn't working very closely with me in research,
although we had a number of conversations, and I took a look at his thesis and I was quite impressed.
written and I read that, and I made it a point to talk to him because I recognized that -- I though this was
some good stuff, and he had some very nice ideas about our network that I hadn't really thought out
myself before. That's basically what happened in Bob's case. He was out
Pelkey: This would have been in '72?
did he say?
me to talk to him on a few occasions. We went out to dinner a couple of times, that kind of interaction,
and he was rather closed-mouthed about what he was doing with respect to EtherNet, and I didn't want to
push him on that.
wanted to clarify exactly what was happening -- what had happened. According to Bob, and I think this
was the case, as soon as he finished his thesis he was hired by Xerox, and then, of all things, he got
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 8 of 13
about a year ago, I said: "Well, that shows that you were a hell of a salesman even in those days."
a significant impact upon his thinking, in terms of
Abramson: Well, it was analyzed in his thesis, to a much greater extent than we had analyzed it. He
cable is something we had never thought of.
saying: "If there is collisions and people know there are collisions, what should the behavior be?" IE,
there should be this back-off.
Abramson: Really, what Bob showed, and this was graduate student-itis -- I've seen this many times --
is that it doesn't make a damn bit of difference. I think that Bob would say that he agrees, but it does give
you a PhD thesis, but it doesn't make a damn bit of difference, and you show it thematically, which I
guess is one of the things that was a problem on his first version of the thesis.
project approved, it was really Bob Taylor and Larry Roberts was a program manager primarily
responsible for the ARPANet, but shortly thereafter Larry succeeded Bob as director
the Director of IPTO, I believe, and Larry was deputy, so he was naturally in line for the position.
Furthermore, I believe it was known then that Bob was leaving and Larry would be taking over, so Larry
was really moving into that.
money to this project." Your contact with Larry, the way you describe it, was unique in the sense that it
was more than just money. He personally took an interest in your project and got involved in the project
at some level.
candidate, he would have gotten a PhD for what he did. He doesn't need another one. If he had been
faculty, he would have published three or four papers for what he did. He was an active, innovative,
imaginative contributor to the whole project all the way through.
he said: "Hey, Norm, I think we can double the throughput of an Aloha channel by this clever idea I have
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 9 of 13
(unintelligible) you got into, but anyway it's a theoretical result which Larry came up on
Pelkey: By slotting you mean
Abramson: Putting time -- setting the packets so that they're synchronized in time to some fixed time
base. That turns out to give you a theoretical increase of a factor of two in the capacity, but has some
other problems. It was an interesting theoretical problem.
it's a good idea. It's used in several places now. He came up with the right answer, theoretically, of how
the slotting would affect the capacity, and he went through a proof which I found very difficult to follow,
and I recall looking at it, being convinced, but I couldn't follow his proof, and I went back and proved it
another way, and that was one occasion. Another occasion -- I think it was at a meeting that he and
Steve Luchesik, who is now the chief scientist of Northrup and was head of ARPA at that point, came out
to Hawaii, and he proposed something called 'Capture Effect.' Actually, it was something known before in
other contexts, but he said: "If you use this something called 'capture effect,' again you can increase the
throughput of the channel, and you can get these results." I recall thinking about it, going away and
coming back to him later and saying: "Oh yeah, you get these results. You get these other results, too."
So neither Larry nor I is the kind of person who writes long detailed proofs and revels in the mathematics
of it. I think we both are the kind of researchers who like to think about ideas and hopefully prove
something or other, but not to get
ideas and common sense, and I think we like to work at the boundary of those, whatever it is. We worked
that way on a lot of those thought.
Pelkey: It must have been a fun interaction.
Abramson: It was a lot of fun. It was very stimulating.
Pelkey: To have someone in that capacity who was coming to you and introducing ideas, it must have
Abramson: Yeah. That's really the way a good university should operate.
Harvard or MIT, and to have someone like Larry who was coming in and turning up the heat every once in
a while, or introducing some new ideas, must have been for you, professionally, rewarding.
Abramson: Yes, it was, and we appreciated that an awful lot. We appreciated the funding too, because
none of it would have gone on without the funding.
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 10 of 13
all of a sudden get a phone call two weeks before December 17 saying: "Get your site prepared." What
happened when you received that phone call?
Abramson: I don't know. I think I sort of half expected to get that phone call.
Pelkey: Did you really?
Abramson: I knew the way DOD worked.
Pelkey: That story will be in the book, for sure, because it's just so fantastic.
Abramson: No, that's absolutely true. He was called out and I was just sort of waiting around reading
things, and I saw this stuff on the board -- great way to put it on the agenda.
rather easy way of working together so it wasn't a problem. My suspicion is that he knew I had done that,
because it was in my handwriting. It just stood out, and he said: "Well, that's a pretty good date. Do it."
Pelkey: A lot of people visited -- first of all, by this time, ARPANet is, in your community, it's well
understood what was happening and that it was working, but the AlohaNet was also working.
Pelkey: Different scale and so on, but in terms of their coming to fruition as research projects, they came
one went into operation first, I think it was the ARPANet, but I'm not sure.
things were there. It must have been a very exciting period of time, with what you were doing and what
was happening with the ARPANet. There was a new form of communications work for computing
devices, and something new was happening, as opposed to these telephone circuits that were out there.
to get into the outfit that really -- the birth of an industry like that.
the beginning, no, but as things went on and you saw what happened, as EtherNet came on and
ARPANet went on line and the FCC started looking at frequencies to use for these kinds of things, it all
sort of sorted itself out.
at that point.
finding ways to be allowed to connect other things to it
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 11 of 13
me quite a bit was the fact that Larry was able to go through the established procedures, that is he was
able to buy lines from AT&T, to use them in new ways, in spite of, I think at that time, some pretty hot
opposition on the part of AT&T, saying: "This packet stuff won't ever fly. That's not the way you
communicate. You communicate this way." Larry was able to communicate another way, and do it on a
large scale, and do it in such a way that he moved AT&T. Now that's an achievement. Moving AT&T is
not easy to do.
Pelkey: I don't have any other questions about that period of time. Is there something that I haven't
were graduate students in the Aloha project, yes. Bob was not. He was, I'm not sure, a post- doctoral
student. He showed up. He was just a good guy to have around that thing. We never worried too much
about who was in what title. Now, as far as John and Charley are concerned, they were both grad
Pelkey: The University of California, Berkeley.
Abramson: They -- I'm not sure where they came from. Was it Berkeley? They were undergraduates
They both were interested more in the software than the mathematics and so forth. They ended up
working directly for Wes Peterson, both of them.
Pelkey: Were they involved in the project right from the very beginning?
Abramson: Yes, but they were involved, I would say, not at the heart of the project. They were involved,
in that they were paid by the project, but they were doing coding for a timesharing system that was built
by Wes Peterson with an IBM computer. For some reason, both of those guys -- I don't usually interact
that much with the software types, but I did talk to both quite a bit, and I was on their thesis committee,
but Wes Peterson was the chairman of the committee. So, yeah, they were both Aloha graduates.
Pelkey: Were there other Aloha graduates who went on in the data communications industry?
Abramson: Certainly there were, because we had, at the height, maybe 25 people -- 25, perhaps,
anybody else that I would name right off the bat. There are lots of good people who were doing some
nice research these days, but certainly Charlie and John both stand out.
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 12 of 13
that, in terms of getting the message out about what you had accomplished. Then, this research project
stayed in existence for some period of time. I assume there was continuing refinements and graduate
students worked on it and so on, but after a period of time you moved on to other things.
after a while, but we had some problems
Pelkey: Because Bob came in and started getting involved in things after '72, and he had his packet radio
rather than technical. He was more concerned with -- he was concerned with trying to sell this to other
people at ARPA and DOD.
Pelkey: His network design was many people in the field being able to talk to each other.
Abramson: That's ok, that's what we were doing too. No problem there. You mean directly to each
other instead of through a central point? Oh, ok.
characterized the problem
Abramson: There wasn't that much difference. We were interested in that too. That problem is still an
primarily, because there were some hard-liners in DOD at that point. It's hard to imagine now, but we had
problems with graduate students from foreign countries, and one guy in particular from Canada
talking about our Canadian researcher, and the Colonel looked at me and said: "Well, that makes me
uncomfortable," and there was no ambiguity on anybody's part that this was simply not going to fly, and
that was his way of saying it. We just felt it was better to terminate the project at that point, and I think it
was the right time to terminate it from lots of points of view.
Pelkey: So the work that Bob did, in terms of packet radio, which had an impact in terms of
internetworking and the creation of TCP from the NCP and so on, the work that had been done at the
University of Hawaii, what they did built on what you had done, but it took on a different . . .
Tape Side Ends
to other things. Were they involved in data communications?
international area and ended up doing some work with the United Nations, and got involved in a whole
bunch of other things, so yeah. Now, the Aloha project sort of -- the brand name is worth a lot, so we've
kept that. We have a number of industrial sponsors who provide us with a small amount of funding, and
we use Aloha to get that kind of funding.
CHM Ref: X5671.2010 © 2010 James L. Pelkey/Computer History Museum Page 13 of 13
I greatly appreciate your time.
though, whether it's right or not.