November 8, 2001

A Paternity Dispute Divides Net Pioneers

By KATIE HAFNER
http://www.nytimes.com/2001/11/08/technology/08NETT.html

FEW inventions are immune from claims and counterclaims
of precedence. Thomas Edison was embroiled in a number
of patent disputes. It took years for the Wright
brothers to secure their rightful claim to
aircraft-powered flight.

Now a dispute is churning around credit for a modern
scientific breakthrough: packet switching, the
technology that breaks all data that travels over the
Internet into discrete bundles that are then sent along
various paths around the network and reassembled at
their destination.

Aside from computer scientists, few people had heard of
packet switching, much less its 1960's origins, until
the 1990's. Then the Internet moved from academia into
the home and the office, and histories of its
development began to appear.

It was about that time that Leonard Kleinrock, a
computer scientist at the University of California at
Los Angeles, began to stake his claim to having been the
father of packet switching. In 1996 he set up a Web page
on the university's site that describes him as the
"Inventor of the Internet Technology" and credits him
with "having created the basic principles of packet
switching."

Until Dr. Kleinrock began making his case prominently,
two others — Donald Davies, a British expert on computer
security, and Paul Baran, formerly of the RAND
Corporation — were widely recognized as packet
switching's inventors. In the early 1960's, Mr. Baran
outlined a packet- switched network that would make
communications less vulnerable to attack or disruption.
Dr. Davies, who first applied the word "packet" to data
communication in the mid-1960's, later built a small
packet- switched network.

Still, some found Dr. Kleinrock's claim persuasive, if
belated. This year, for example, he was among four
laureates who shared a $500,000 prize for their roles in
creating the Internet. The prize, administered by the
National Academy of Engineering, cited Dr. Kleinrock for
publishing on packet switching as early as 1961.

But Dr. Davies has struck back, in a paper published a
year after his death at age 75. "In Leonard Kleinrock's
work," he wrote, "I can find no evidence that he
understood the principles of packet switching."

The assertion has riven the ranks of Internet pioneers.
It has also left the two sides in the awkward position
of arguing over an eminent colleague's final testament.

"The Internet is really the work of a thousand people,"
Mr. Baran said. "And of all the stories about what
different people have done, all the pieces fit together.
It's just this one little case that seems to be an
aberration."

The dispute is all the more remarkable, those who knew
Dr. Davies say, because he was known for being humble —
to a fault, some say — about his own accomplishments. In
addition to his work on computer networks, he was a
computer security expert, working with a small
development team headed by Alan Turing, the British
mathematician and cryptography expert who broke German
codes during World War II.

"Donald Davies is one of those men who brings to mind
Harry Truman's observation that `you can accomplish
anything you want in life provided you don't mind who
gets the credit,' " wrote John Naughton, author of "A
Brief History of the Future: From Radio Days to Internet
Years in a Lifetime" (Overlook Press, 2000). Mr.
Naughton's history and others like it (including one
published in 1996 by this reporter) credited Dr. Davies
and Mr. Baran with having invented packet switching.

Dr. Davies's last word on the subject came in a paper
published posthumously last summer in The Computer
Journal, a publication of the British Computer Society.
However self-effacing by nature, Dr. Davies was blunt in
rejecting Dr. Kleinrock's claim, though the issue may
require splitting the finest of technical hairs.

For years Dr. Kleinrock was known for his work in
queuing theory, a branch of mathematics that studies how
long people and things spend waiting in lines, how long
the lines get and how to design systems to reduce
waiting. Dr. Kleinrock said in an interview that
although the two fields are different, "it turns out
that queuing theory is the perfect tool for analyzing
the way packet networks behave."

In his paper, Dr. Davies found fault with Dr. Kleinrock
for not describing message switching beyond a single
node, or queue, in his 1962 doctoral thesis. "Following
through such a scheme would have made an interesting
study," Dr. Davies wrote. "But Kleinrock does not go
beyond the single queue."

Dr. Kleinrock disputes that assertion. "My entire
dissertation is devoted to data networks which consist
of connected sets of single nodes," he said.

And more generally, Dr. Kleinrock called the Davies
article "shocking." He and Dr. Davies had exchanged
cordial e-mail in the months leading to Dr. Davies's
death, Dr. Kleinrock said, and he had no idea that Dr.
Davies was writing the paper.

Dr. Davies was adamant that the paper not be published
until after his death, said Michael Davies, his son. "He
felt that if it came out during the latter stages of his
illness, he wouldn't have the strength to get in a
debate about it," Mr. Davies said. "But he felt deeply
that he had to set the record straight."

Yet perhaps counter to Dr. Davies's intentions, instead
of helping resolve the matter, his paper has simply
helped galvanize each side.

Lawrence G. Roberts, who in the late 1960's designed the
Arpanet, the precursor to the Internet, has been a
prominent backer of Dr. Kleinrock's claim, at least in
recent years. His honors include two that were shared
with Dr. Kleinrock — the National Academy of Engineering
award and last year's I.E.E.E. Internet Award from the
Institute of Electrical and Electronics Engineers. (The
latter award, unlike the academy's prize, also
recognized Dr. Davies and Mr. Baran.)

While they were graduate students and researchers at the
Massachusetts Institute of Technology in the 1960's, Dr.
Kleinrock became a close colleague and friend of Dr.
Roberts. Inveterate casino gamblers, the two scientists
together tried to beat the odds at blackjack and
roulette with various clever methods. They remain good
friends.

Throughout the early 1990's, when asked, Dr. Roberts
pointed to his own work, as well as the work of Mr.
Baran and Dr. Davies, as the major influences on his
thinking when he designed the network. It wasn't until
Dr. Kleinrock began to lobby for recognition that Dr.
Roberts changed his mind and not only began citing Dr.
Kleinrock's work but also took up his friend's cause.

In an interview, Dr. Roberts said he erred when the
topic of Internet history first arose in the early
1990's. After thinking more about it, he said, he
decided that Dr. Kleinrock did indeed influence his
thinking when he designed the Arpanet.

"I didn't differentiate Kleinrock heavily until it got
to be an issue," Dr. Roberts said. "I realized I had
left out what he had contributed to what I was doing.
All of us started realizing we were part of history, and
we had to make it clear."

Now, if either man sees Dr. Kleinrock omitted from a
citation, each is quick to correct the record.

For his part, Dr. Kleinrock acknowledges waiting until
1996 — and says he did so then only because he saw his
contribution eclipsed by the claims of others.

"You might find it hard to believe, but I tend to be a
modest person," he said. "I become less modest when I
see people taking away the things I had done."

Even without the claim to packet switching, Dr.
Kleinrock's roster of achievements — listed on his home
page at www.lk.cs .ucla.edu — is impressive. In 1969 his
lab at U.C.L.A. became the first node on the Arpanet and
the official center for network performance testing.

An entrepreneur, Dr. Kleinrock was a co- founder of
Linkabit, an early computer consulting company, in 1968.
He is also founder of Nomadix, a three-year-old computer
networking company in Westlake Village, Calif.

Dr. Kleinrock said the controversy had put him in an
uncomfortable position. "I don't like this role of
having to defend the things I've done," he said. "It's
more than pride of ownership. If no one had tried to
claim it, it would be different. But if someone tries to
take something away, that's a proactive move, and that
requires a response. It's not peaceful coexistence.
History should get it right."

At the same time, Dr. Kleinrock said he believed that
Dr. Davies and Mr. Baran deserved credit "on a number of
fronts" — Dr. Davies for coining the term "packet" and
for building his experimental network. Both should be
credited with coming up with the idea of slicing
messages into discrete packets, albeit after he did, Dr.
Kleinrock said.

Dr. Roberts, who is now chairman and chief technology
officer of Caspian Networks of San Jose, Calif., a
company building a new generation of Internet switching
equipment, said the appearance of the paper diminished
his respect for Dr. Davies.

"He was a brilliant man," Dr. Roberts said. "I just
thought he was an immense figure, and this cuts my
opinion some, because he knows that this is not the
truth."

Dr. Roberts was particularly surprised that Dr. Davies
chose to focus on one small aspect of the technology —
the dividing of messages into small pieces of equal size
and sending them to several intermediate nodes on the
way to their final destination.

"The question of who invented which piece may be very
hard to track down, but there's a lot more to packet
switching besides this little teeny piece," Dr. Roberts
said. "We can argue about it, but my contention is we
all were thinking about it."

But another of the Internet pioneers, Mr. Baran, a
Davies ally, said he thought Dr. Davies had made a
professional effort to be fair.

Mr. Baran said he exchanged extensive e- mail with Dr.
Davies about Dr. Kleinrock's work in the months
preceding Dr. Davies's death, though he knew nothing of
the Davies paper until Dr. Davies sent him a draft early
last year.

"Davies was a very gentle guy," Mr. Baran said. "What he
wanted to do was make absolutely sure he really
understood what Kleinrock was saying. He really bent
over backwards to make sure he wasn't misinterpreting
anything Kleinrock wrote."

And for every supporter of Dr. Kleinrock there is an
equally opinionated detractor. Dr. Willis Ware, a
scientist at the RAND Corporation who was Mr. Baran's
boss there in the early 1960's, shares Dr. Davies's
skepticism of Dr. Kleinrock.

He and others suggested that Dr. Davies was moved to
write the paper as a point of honor. "It was clear from
his paper that Davies believed that Kleinrock was
stretching the truth in a way that didn't seem
appropriate," Dr. Ware said.

Honor is not to be underestimated in the world of
scientific achievement. "Honor is an important part of
the reward system, because other rewards flow from
that," said Dr. Charles Bazerman, a professor of
education at the University of California at Santa
Barbara and an expert on Thomas Edison. "So priority
becomes a crucial issue, because it's the most basic
reward."