Through the much of the 1980’s and early 1990’s, Cambridge-based Thinking Machines was ahead of its time. As innovators in parallel computing, they developed a massive, 65,536 processor supercomputer known the Connection Machine. Visually, it made Cray’s distinctive look seem like a piece of outdated furniture, and was even stunning enough to star as the “impressive blinky-light server” in Jurassic Park.
Of course, that’s just about all it was good for. The Connection Machine was an AI researcher’s dream that no AI research lab could afford. Its inability to run FORTRAN – and every other programming language aside from a specialized Lisp dialect – made it pretty much useless for business and scientific purposes. Its baffling inability to even do floating-point operations mostly guaranteed that no one would buy it. But, hey – who needs customers when there’s lots of money from daddy!
In this case, “daddy” was DARPA (Defense Advanced Research Projects Agency), and not only provided Thinking Machines with millions of dollars in funding, but brokered and subsidized many of the Connection Machine sales. As more money poured in, even more money poured out. CEO Sheryl Handler – whose background included interior design, landscape architecture, city planning, and third-world resource planning – built a lavish office that included huge open spaces, an enormous marble archway, and even a plush cafeteria staffed with its very own gourmet chef.
While all of this so far is chronicled in Inc. Magazine’s 1995 article, The Rise and Fall of Thinking Machines, reader Andrew Garland wrote in to share his own experience with Thinking Machines. During their heyday, he interviewed for a software engineer position.
I arrived at Thinking Machines at 9:00AM sharp. Having not gotten a detailed scheduled from the HR person, I wasn’t quite sure what to expect, and just assumed that I’d meet with one or two people and then be on my way. I do remember being impressed by their substantial building and the meticulously designed lobby.
The first interviewer that I met with had never seen my résumé before, so I sat patiently as he read about my background. He said that I had great experience, and wondered if I had any sort of experience on massively-parallel, hypercubic-based computers. Seeing that only Thinking Machines programmers would have such a skillset, I said that I hadn’t designed for or programmed such a machine, but certainly was interested in learning more.
On that note, he perked up and talked about the amazing things their computer could do, and detailed the particular project that he was involved in. The project was a bit abstract, so I asked how it could be applied for business computing purposes. He scrunched his nose and scoffed at the very notion that I’d ask such a question.
My second interview of the day was almost identical. I met with a programmer who had never seen my résumé before and was more interested in talking about his current project than my technical skills or knowledge. When 10:50AM rolled around, he walked me over to my third interview, but made a brief pit stop at a Coke machine.
“Ya see this,” he said with great pride, pointing to the magnetic strip bolted to the machine, “we actually hacked the machine to accept employee access cards as payment. It’s pretty elegant; a request is sent via Ethernet to a server, which then verifies it and runs an automatic charge against that employee’s account.” Though I was quite impressed by the creativity, I did wonder where they found the time to do that.
By my third interview, I finally got the hang of things: I quietly sat while the interviewer read my résumé for the first time and then listened to him talk about his exciting computing project. While I can’t remember what his project was, I do remember what his response was to my question about its application in science or business: “errr, I’m not sure; I guess I don’t really understand why you’d ask that.”
My third interview ended at 12:30, and I was asked to stay for lunch at the building’s impressive cafeteria. After that, they’d talk to me again after. I was assigned to an escort who showed me where to go, and he had lunch with me. We briefly chatted about the company some more, and the interview process.
“Five interviews in one day,” I said, “I guess that’s a good sign, eh?”
“Well, yes,” he hesitantly replied, “they usually do this for everyone though.”
“Oh. So what happens after today, if things go well and all?”
“They’ll actually call you later tonight,” he explained, “and invite you back tomorrow for another day of interviewing.”
“Wow! Heh,” I said jokingly, “I hope they don’t have a big unanimous vote or something!”
“Well, err,” another hesitant response, “actually… you will need to be interviewed by all of the development staff. There’s twenty five or so. Then, they’ll meet in the auditorium to discuss you as a possible new member, and any no votes would end your consideration.”
“Yeah,” he said, “I guess they’ve always hired this way. Even the CEO has to vote you in.”
After we finished lunch, my escort took me to the fourth interview, which began and ended just the same as the ones before it, and the fifth one after it: lots of talk about their pet project which, to me, did not seem a natural application of the technology or make any business sense at all.
I was done at 4 PM and happy to walk out into the sun and stretch. I didn’t look forward to the next day.
Thankfully, I received a call later in the evening from the HR person. She said that I didn’t seem to be a good fit for Thinking Machines. Reflexively, I asked why.
“Well to be honest, Andrew,” she replied, “the developers didn’t think you were a true believer. We have some really incredible things here, and you just weren’t sufficiently impressed by the technology you could have worked with.”
As it turned out, Andrew wasn’t the only non-believer. Just as DARPA was about to send more barrel-loads of cash to Thinking Machines, The Wall Street Journal rained on their “subsidized sales” parade. That led to an embarrassed Bush I administration, which led towards an end of support from daddy.
With the impressively inept Sheryl Handler at the helm — the CEO who prioritized things like publishing a cookbook with recipes from their cafeteria instead of, say, trying to sell their increasingly useless Connection Machine — Thinking Machines quickly sank and filed for bankruptcy a short two years later.