Dealers of Lightning (7 page)

In
contrast to the rigid protocol of the brass,
ARPA's
civilian chiefs left
Taylor undisturbed to fashion his
program as
he chose.
What
he chose
was distinctly an extension of
Licklider's.
Interactivity, time-sharing,
graphics: All of Lick's pet preoccupations became Taylors.
He
also
shrewdly adjusted Licklider's management style to fit his own extroverted
personality.
He
would visit his grant recipients several times a year, but
not solely to hear the researchers' obligatory progress reports.
He
was
engaged in something more like community outreach, developing new
teams, nurturing up-and-coming young researchers, cultivating an entire
new generation of virtuosi.

Taylor had spent enough time in academia to know that the most inter­esting intellectual ferment took place well below the stratum of full pro­
fessor. "Who are your youngest faculty?" he would ask around. 'What are
their ambitions? Where are the most impressive graduate students?" He
tracked down the young prodigies, captivating them with the peculiar
charm of a leader who seemed inclined more to listen and encourage
than to dictate. An impresario needs a company to put on stage; Taylor, in
one sense, was holding auditions. In another, he was reproducing one of
the most meaningful moments of his own past. The persons he sur­rounded himself with would get there not by chance or breeding. They
would be chosen, selected . .. adopted.

In his first year as director Taylor organized one of what would become
an annual series of nationwide IPTO research conferences. On the sur­face this filled a troublesome communications vacuum. Here were the
top talents in the field of information processing, all working for ARPA on
essentially the same problems. Yet most did not know each other except
by reputation. He made the conference an annual affair, held each year
at some different and gratifyingly "interesting" place. One year winter
might find the group skiing at Park City, Utah; the following year would
bring them to New Orleans in time for Mardi Gras.

Not that the purpose was chiefly to play. Rather, it was to build a net­work of people mirroring the one he would soon propose for comput­ers. Lifelong professional and personal bonds were forged at these
events. They would start the day with a communal breakfast, followed
by several hours of discussion in the morning. They would eat lunch
together, then were set free until dinner, another communal affair. The
day ended with further colloquia.

The daily discussions unfolded in a pattern that remained peculiar to
Taylor s management style for the rest of his career. Each participant
got an hour or so to describe his work. Then he would be thrown to the
mercy of the assembled court like a flank steak to a pack of ravenous
wolves.

"I got them to argue with each other," Taylor recalled with unashamed
glee. They went at the intellectual roughhouse with the scientists
unemotional candor, oblivious to everything but the substantiation of
truth. "These were people who really cared about their work. They
weren't
interested in politics, they weren't interested in impressing any­
body,"
Taylor said. "If they thought that something
I
was saying was dead
wrong, they'd just as soon tell
me
as not. They'd just as soon tell one
another as not if they thought
they
were wrong.
And in the end these
people, all of whom were pretty bright,
got to
know one another better."

Of
course he had another
motive in fomenting
tire intellectual free-
for-all.
It
was the best way he knew of
gleaning
what they were up to.
He
could have asked any of his principal
investigators
to sit him down and
explain their work, it was true,
but
he
figured
the chances were slim that
he would even know the right questions
to
ask. Better to let his experts
challenge each other while
he watched, like a
platoon leader subjecting
his men to the rigors of field
maneuvers to see
whether they will break,
and
if so, where. "This way
I would get insights
about their strengths or
weaknesses that otherwise might be
hidden
from me," he said.
"If
diere
were
technical weak spots, they
would almost
always surface under these
conditions.
It
was very, very healthy."

But it
was not to be personal. Impugning
a
man's diinking was accept­
able,
but never his character. Taylor strived
to
create a democracy where
everyone's ideas were impartially subject
to
the group s learned demoli­tion, regardless of the proponents credentials or rank.

The
same principle governed the
once-a-year ARPA
conferences
Taylor established for graduate students. Faculty members and even
Taylor himself were barred from
these
meetings. Barry Wessler, not
long out of grad school himself,
was delegated
to supervise, receiving
no instructions other than "to get
people
together and make something
happen."

At
the first session the group piled
on an
unfortunate wild man from that
backwater, the University of Utah,
named
Alan Kay.
Kay
had stepped
forth in a public session to pitch his vision of a computer you could hold in
your hand.
He
had already coined a name for it: "Dynabook," a notebook-
shaped machine with a display screen and a keyboard you could use to
create, edit, and store a very personal sort of literature, music, and art.

"He
was crazy," Wessler recalled. "People greeted the whole idea with
disbelief and gave him a very tough time.
He
painted this picture of walk­
ing around with a computer under your arm, which we all thought was
completely ridiculous."

Taylor, meanwhile, was fully occupied in finding ways to push for­ward the frontiers of interactive computing. At the time, this meant
advancing the technology of time-sharing because there was simply no
other way to pay for the enormous computing resources an interactive
system demanded.

During his tenure Licklider had steered most ARPA funding to time­sharing projects of a certain majestic scale, such as MIT's vast Multics
program, whose aim was to design a system capable of supporting 300
users at once. Taylor encouraged his contractors to embrace smaller-
scale projects as well. One of these, an effort at the University of Califor­nia at Berkeley called Project Genie, was based on the principle that not
every university could afford the multimillion-dollar General Electric
645 mainframe Multics required. Instead Genie aimed to design a sys­tem for no more than ten or twenty users. If such a small machine could
be widely distributed, Taylor reasoned, time-sharing might actually reach
many more users than Multics could ever deliver.

Genie's host machine was the SDS 930, which was made by an entre­preneurial three-year-old company in Southern California named Sci­entific Data Systems and sold for only about $73,000. The 930 was a
popular entry in the commercial market, where it was widely admired
for its exceptional speed, excellent reliability, and large storage capac­ity. The Genie team demonstrated that with a nominal amount of new
hardware and clever reprogramming the versatile 930 could be turned
into a small-scale time-sharing machine. Then they proposed that Taylor arrange for Scientific Data Systems to bring out the modified 930 as
a commercial product.

Taylor gave the idea his enthusiastic endorsement. He invited Max
Palevsky, the founder and chairman of SDS, to the Pentagon for what
ripened into a memorable encounter. Palevsky, offered a government-
funded research prototype on a silver platter, turned him down flat.

The forty-two-year-old Palevsky was an executive whose heartfelt con­fidence in his own business acumen had been reinforced by his com
pany’s extraordinary success. (In 1965, the year of his encounter with
Taylor, SDS had earned more than $5 million in pre-tax profits.) But
Palevsky had no use for time-sharing. SDS made its money selling small-
and medium-sized computer systems for a niche market of universities
and aerospace firms. To Palevsky the technology of time-sharing
appeared too elaborate by half and the commercial market a black hole—
even IBM, the bellwether performer in the computer industry, had run
aground trying to implement a time-sharing strategy of its own. "It was a
very sophisticated scheme," Palevsky recalled, "and I just didn't think
there were that many sophisticated customers around."

Equally strong-willed, Taylor and Palevsky conceived an instant
mutual antipathy that had them at cross-purposes from the start. Tay­lor tried to draw Palevsky's attention to a terminal in his office linked
directly to the reconfigured SDS 930 chugging away on the other side
of the continent. Palevsky had not the slightest interest. Taylor judged
Palevsky's knowledge of computing shallow and retrograde. Palevsky
grew equally impatient with this government bureaucrat whose techni­cal exuberance clearly outran his commercial sense. Beyond that, he
announced that the work done by the vaunted Genie team was "full of
holes."