edge. And we also thought it was a good recruiting technique in terms of
having people focus on three things we thought were important. First of all,
“Let’s focus in both directions on the people, on the culture, on the environment,
see if that makes sense to you.” Because we asked them some extraordinary
things. We said, “This project is going to take about 2 years, and it’s going
to be a lot of work. Furthermore, we are going to institutionalize it by saying,
‘We need you to work every other Saturday.’”
Livingston: Really?
Gruner: Yeah. “You gotta be here. It’s a regular work day. And the other
Saturdays and Sundays you might have to be here too, but every other Saturday
is a regular work day, gotta be here on time, full day, no monkey business. And
that’s for 2 years.” We told them that early on. We wanted to get people to focus
on that first, and then get psyched up about the project. We didn’t want to
reverse it, where people got really psyched up, “Oh, I want to work on this sexy
technology.” We held that to the very end.
During that 2-year period, where we were working very hard, we had virtually
no attrition. At the time we announced, we had, I think, 40 people in the
company, and I think over the 2 years, the attrition was one or two people. It as
very small, because we did a careful job of filtering people. And they didn’t
come in saying, “I feel like there was misrepresentation here,” or “I
didn’t understand what was going on.”
Livingston: Do you remember any big turning points as you were building
these parallel processing systems?
Gruner: We were using a technology called gate arrays. These were custom
integrated circuits. In our case, we were using technology by Fujitsu in Japan.
They were very expensive to design, very expensive to tool. They had a very
long development cycle. So in building the design, the computer that would use
the first revision of gate arrays was really critical. If we had to go through revisions,
we had budgeted that, but that would make things a lot more expensive
and more complicated. So when our first gate arrays came back, representing a
total investment at that point of probably $3 million, and essentially worked
almost completely, that was a huge milestone in the project. That was a year
before we announced.
So we had hardware that began to work, and then the next step was the
Fortran compiler. Fortran is a computer programming language, and the compiler
is what took a Fortran program, say, written for a Digital Vax computer,