This segment is from pages 36-40 of an excellent history of Seymour Cray and his career.
The Supermen: The Story of Seymour Cray and the Technical Wizards Behind the Supercomputer Charles J. Murray (Author)
...
The purchase of ERA was a curiosity to the engineers at the old
Eckert-Mauchly facility. They were unsure what ERA did. They'd
heard something about ERA making cryptologic equipment for
the navy, but they were unclear how the objectives at the St. Paul
glider factory fit in with those of Philadelphia. Most of the engineers
in Philadelphia were openly disdainful of the St. Paul crew.
Eckert told Norris he thought ERA was in "the Dark Ages." In
Philadelphia they referred to the St. Paul facility as "the factory."
When that reference trickled back to the engineering crew in St.
Paul, it rankled them. The factory? It was true that the ERA crew
referred to their own facility as "the glider factory" but when
Eckert-Mauchly engineers said it, it had a distinctly different
feel-as if it pictured a band of steelworkers building automobile
crankshafts. Whether it was reality or paranoia, or a combination
of both, the ERA crew began believing that Eckert-Mauchly engineers
looked down their Ivy League noses at them. Every slight
was remembered and repeated. Someone had referred to them
as "the farmers." The farmers. Of their state-funded, prairie school
educations, someone else had coined the term "Moo U." With
each slight the gap between the sister divisions grew deeper.
If the Philadelphia group was disdainful of ERA, however, the
St. Paul band returned its condescension in trumps. True, they
grudgingly admitted, the Eckert-Mauchly crew was more innovative.
Some of the features on the UNIVAC I had required astounding leaps
in engineering, but the crew's record of reliability
proved that they were far better scientists than engineers. That
was the one element of justice in it all. In a perverse way ERA was
gratified that the Philadelphia theories weren't always successful.
ENIAC had been plagued by downtime, and UNIVAC I was lucky
if it could sustain a ten-minute run without going down due to a
bad vacuum tube or some such thing. In contrast, ERA's Atlas
had an extraordinary run of five hundred hours in which it required
only sixteen hours of unscheduled maintenance. The
technological gap that emerged between the two groups centered
on the practical versus the theoretical. It was Ivy Leaguers
versus the farmers, and the debate never ceased.
What galled the ERA engineers about the debate was that the
theoreticians from Philadelphia never seemed to acknowledge
the importance of the practical side. It was as if Philadelphia's job
was to explore theory, leaving the issues of building workable designs
to lower forms of life. Eckert bluntly told Norris that ERA's
creative capabilities simply didn't exist.
The tension between the two groups reached a peak when Eckert
visited ERA to discuss new ideas for a commercial machine
called the 1103. The engineers in attendance were quizzical at
first, wondering how Eckert could discuss a design that was so
close to completion. When he proceeded to spend an entire
morning and part of an afternoon lobbing brilliant ideas at them,
the ERA engineers were appalled. They sat in stony silence on
their metal folding chairs, wondering how he could possibly expect
to incorporate new theories in a nearly completed machine.
Eckert, who was known for his blustery ego, sensed their reticence
and exploded. "What's the matter with you people?" he
barked. "I've been standing up here all day throwing out new
ideas, and no one's said a word." But the farmers sat quietly, eyeing
him with their stony midwestern stares. Afterward, they privately
confided to one another that Eckert's outburst served as a
metaphor for all his company's problems: Brilliant as they were,
the engineers in Philadelphia wanted to keep the design unbuttoned
until the last tense moment, and in the process Eckert-Mauchly had
eventually ruined their reliability.
When Norris took the helm as general manager of the UNIVAC
Division, he learned that the differences between the two operations
went beyond petty name-calling; they reflected profound
disparities in philosophy. "The difference," he told the Philadelphia
group, "is that you people run a laboratory, and ERA runs a
business." The principals in Philadelphia were interested in raw
speed and didn't particularly care about reliability. Worse, they
had put themselves in a financial tangle by bidding less than one-fifth
what it cost to build the UNIVAC I. Later when building a
computer called the LARC, they would spend $19 million on a
project for which they had bid $2.85 million.
The navy had forced ERA to toe a financial line by setting up a
cost-plan-fixed-fee arrangement and then planting itself on the
premises in the Naval Computing Machine Laboratory. With Eckert-Mauchly
now answering to Norris, the intimidating farm boy
from Nebraska, the brilliant theoreticians from Philadelphia
were forced to toe the financial line, too.
In mid-1953 Bill Norris asked Willis Drake to go to Louisville,
Kentucky, for two weeks to oversee installation of the world's first
on-site commercial computer, the UNIVAC I. For an engineer
who'd begun his career in St. Paul, as Drake had, it was a strange
request. The UNIVAC I was designed and developed in Philadelphia,
and in a sense it was a winner in the race to be the first commercial
machine - and a St. Paul employee was being asked to
oversee its installation. But Drake understood that the walls between
the two organizations must one day collapse, and he
agreed to go.
When he arrived in Louisville, Drake was first dazzled by the
accommodations that had been made for the new machine. General Electric,
the first customer, intended to proudly display UNIVAC I
in a special site at its Appliance Park facility. GE planned
for tremendous news coverage - the machine was purchased as
much for public relations as for real computing. As a result, it
built a gleaming new showplace for UNIVAC I: floor-to-ceiling
glass walls, multicolored drapes, huge potted plants, and spot-lights,
all designed to solidify GE's image as a technological
leader.
But the real surprise that awaited Drake went far beyond the
building's decor: much of the UNIVAC I was missing. The UNIVAC I,
dark and quiet, sat in the middle of its vast new showplace,
surrounded by cardboard boxes. And inside the boxes was . . .
nothing. Remington Rand had delivered a box of electronics and
little else. There were no input or output systems, so GE's programmers
would have no way to put in data or take out data. The card-to-tape reader,
the tape-to-card reader, the high-speed printer -
all were missing.
Drake was perplexed. Here he was with only half of a UNIVAC I
to install. A whole machine was a rat's nest of wires and contained
thirteen thousand vacuum tubes. GE had spent millions of dollars
on it and had already orchestrated substantial press coverage. And
after the press coverage died down, GE would expect the machine
to work. Its users were already putting together programs for business
applications: payroll, inventory control, production scheduling.
Drake called Remington Rand's Philadelphia offices, but was
told that none of the missing equipment was ready. For several
weeks he worked with GE managers, explaining the problems involved
in the installation, and in the beginning the managers
waited patiently He pushed the operation dates back, and GE reluctantly
accepted the new dates. Still, the parts failed to arrive.
Drake meanwhile called John Parker, who now worked as a vice
president in one of Remington Rand's New York facilities, urging
and begging for the missing parts. The situation was becoming
awkward, he told Parker. There'd been mountains of publicity.
The press wanted to see UNIVAC I in operation; GE wanted to
run its payroll and production programs. Parker seemed surprised.
"Gee, the card-to-tape was due to ship last Wednesday" he
explained, but there was a last-minute change and they felt it
would enhance your liability [as printed in the 10th edition] if they put that change in before they
sent it. And as long as it's this late anyway ..."
Drake was appalled. Each week there was another problem
with the delivery of the card-to-tape system, or the tape-to-card
system, or the printer. And each week the engineers in Philadelphia
seemed to be tweaking the design just a little bit more. The
original two weeks had now stretched to three months.
Finally GE's Appliance Park manager, whose patience was at an
end, called Drake on the carpet. "Either this equipment is going
to be here on this date, or all this stuff is going to be out in the
middle of the street, and I'm not kidding you," he said. Panicked,
Drake purchased a plane ticket with his own money and boarded
a flight to Philadelphia. He entered Eckert-Mauchly without an
appointment and introduced himself, asking to see the card-to-tape
converter and the high-speed printer. He showed them a letter
from John Parker, promising a delivery date that had long
since passed.
The engineers shot glances at one another, gazing at Drake as
if he were speaking Italian and they'd forgotten their Italian-English
dictionaries. One of them led Drake to a Ping-Pong table
with electronic parts scattered across it. "That," he said, "is the
card-to-tape converter." The printer was even farther behind
schedule. "I don't know where in the hell you got any dates like
that," he said. "This isn't going to be ready for a year." The Philadelphia
engineers, Drake thought, acted as if the input-output
devices were mere details, that they'd already completed the important
part.
Drake hustled out of the Eckert-Mauchly facility and boarded a
flight to New York, where he cornered Parker. He told Parker
about the card-to-tape converter, the printer, and the Ping-Pong
table. He explained the position he was in, the waning patience
of GE management, and the lack of any concrete solution. Parker
listened in disbelief. Then, to Drake's amazement, he responded,
"You're wrong." Parker stood up, ambled across his office, and
pulled a letter from a file, showing Drake that the systems were
supposed to be ready. Drake was incredulous. As far as Parker was
concerned, the memo said the parts would be ready, so they must
be ready. It was as simple as that.
For Drake, it was the ultimate learning experience. The fiascoes
continued, deadlines were missed again and again. Drake remained
in Louisville two years before his two-week project ended.
Ultimately, the machine worked, and GE obtained years of use
from it.
The lessons were obvious: The years at the glider factory,
under the guidance of the navy, had taught ERA engineers how
walk a fine line between evolutionary and revolutionary product
development. They knew how to use research without becoming
researchers. They were engineers, and as engineers they'd
learned to build machines that worked-reliably and on time.
For that, they didn't need a giant conglomerate or a rich parent
company.
The problems at Appliance Park were an example of how far
the engineers at the glider factory had come. If this was the state
of the art in the commercial market, carving out their own niche
wouldn't be difficult.
|
