On Working with Brion
The office in the locker
I once worked for IBM’s research division in Yorktown Heights, New York. It was late 1979 and I had just finished a microprocessor design at Motorola in Austin, Texas. The experience at Motorola had been my first exposure to politics in the workplace, so that was one reason I was ready to leave. I interviewed six or seven places and got offers from all of them (I must have hit a window of critical shortage in engineering). In any event, I put all the offers into a giant spreadsheet with advantages, disadvantages, and characteristics such as weather and made a decision. As an aside, this was long before the publication of How We Decide, by Jonah Lehrer, a book that explains in detail how adding to the range of things considered in making a decision can result in poorer selections, but I’ll leave that for an essay titled “Don’t Interview in the Spring.”
I was a computer guy, so I thought IBM’s research division probably had the smartest computer guys on the planet. I couldn’t believe my luck in fooling the company into making an offer, so I just couldn’t turn it down. I’ll have more to say about this in “Bozo Contours.” In any event, I moved to New York in the fall of 1979. I thought I was tired of working on microprocessors (that would later turn out not to be true), so I chose a project working on a fiber-optic serial channel for the IBM/370. I’d be working with Brion Shimamoto and he and I would be working for Marty Sacks. I don’t remember much about Marty except that I have an impression of him as humorless, insecure, and a possible stand-in for Danny DeVito—very short and very round, but without the smile.
IBM provided the best of everything to its researchers. We had terminals connected directly to high-speed mainframe computers, free parking, private offices, travel money to attend conferences, and an excellent cafeteria. Actually, it was an outstanding cafeteria when I arrived (I think the division hired real chefs), but then the busybody do-gooders got their teeth into the situation and the cooks had to quit using salt and butter. I’m sure we’re all better off than if we had been allowed to continue making our own choices based on manifestly unhealthy notions such as personal preferences.
After about a year, Brion and I tired of working for Danny DeVito and cut him loose to work on our own project (this is something you could do inside the research division). We decided to do a paper design of a microprocessor in order to write a textbook on the topic. It was our thinking that there was a rapidly growing field in something called design automation that lacked a roadmap; there were hundreds or thousands of programmers beavering away building software automation tools to support integrated- circuit design. The problem was that the workers were programmers and not design engineers and they weren’t working from a specification. What was it they were automating? I couldn’t tell, but it wasn’t integrated-circuit design. I thought I could help by writing a textbook on the design process; then the design-automation fanatics would have a roadmap to work from and could actually automate the processes that engineers used in design. We did write the book (Microprocessor Logic Design), and it was used as a university text, but there didn’t seem to be any design automation jobs that depended on knowing what was being automated.
Enough of the boring work-related stuff, let’s get to the fun parts of working in IBM research.
Most people working on projects together want offices close to colleagues; Brion and I asked for offices in the same location, but facing onto different aisles. The Yorktown Research Center is a partial toroid, with the offices embedded in interior spokes. Offices a spoke apart face each other on the aisle. The office has room for a desk or two, bookshelves on the wall, and lockers at the back of the office. Brion and I had back-to-back offices. We sealed the door of one office and posted a formal warning sign similar to the one shown nearby. We then removed the inside back of the locker and thereby created a walkway between the two offices. We could enter “Brion’s office,” step into the locker at the back of the room, and then step out into the “restricted office.” Generally, Brion worked in the front office and I worked in the back office.
I was working away one day in the back office and had no idea what Brion was doing, but it was 11:11— lunch time—so I stepped into the locker and out the other side. Brion was talking to some hapless visitor in his office when I suddenly stepped out of the locker. I think Brion’s visitor fell out of his chair: I’m sure that of all the offices he had been in in the building this was the first time he had ever seen anyone emerge from a storage locker.
Saving the trashcan
The cleaning people at IBM Research didn’t go into the private offices to pick up the trash; as a consequence, if you wanted your trash emptied, you put the trashcan outside your door when you left for the day. The problem with this scheme was that putting your trashcan in the hall gave access not only to the cleaning people, but also to anyone else that might want a trashcan. We lost a half-dozen trashcans to opportunists. What to do…
We got a trucker’s chain and a couple of padlocks and chained our trashcan to the door. That was good for about a day—we were summoned by our second-level manager (a big deal in a hierarchy-conscious environment). It seems the supervisor of the cleaning crew had gone to the lab director to complain about the chained trashcan. We meant no offense to the cleaning crew, but rather to the sticky fingers of our colleagues, but there was no explaining this in any satisfactory way; the chain had to come off. From then on, we stocked a dozen or so trashcans behind the door and took to more direct methods of offending our colleagues.
See my diplomas and awards
I remember two parodies of our pompous colleagues. On one wall of Brion’s office, we had mounted what looked like the usual citations and diplomas posted on numerous offices throughout the building attesting to the brilliance and accomplishments of the offices’ occupants. Close examination of our wall showed our citations to be matching, forged high-school diplomas, one for each member of the project lined up like fireplace Christmas stockings. I also had posted a framed employment-rejection letter from IBM’s typewriter division and a letter of reprimand from IBM’s chairman (a story for another time). We were directed to remove both of these. We appealed the decision, but to no avail. The diplomas stayed.
Library Copy: Do Not Remove
We got into rubber stamps and our own formal reporting documents in parody of IBMs formal processes and procedures. One rubber stamp we had said: “Library Copy: Do Not Remove.” It was pretty much identical to a similar stamp that the library used. We made liberal use of this stamp on the many professional magazines and other publications that came through the office. A casual glance at our trashcan would reveal marked, torn, and discarded documents and magazines carrying the library’s putative admonition stamp.
For IBM External Use Only
In parody of documents designated “For IBM Internal Use Only” and “IBM Confidential,” we had a stamp that said: “For IBM External Use Only.” We used this stamp on nearly all of the Micro/370 working documents. That made it easy to turn away potentially interfering busybodies: “I’m sorry, but the contents of our project are designated ‘For IBM External Use Only.’ While we are free to discuss project developments with anyone outside IBM, no one on this project is authorized to disclose anything substantive to another IBM employee.”
The best chapter in the textbook
Brion and I started a microprocessor design project so that we could write a textbook covering the design method and have concrete examples to point to. During the course of the design the project morphed from a pencil-and-paper design into a real chip design: we decided that we’d have better credibility with a real design than with a paper-only design. In addition, we felt there was a real need for a IBM/370-based microprocessor inside the company.
One problem with publishing a textbook on the design method was that internal review committees and a host of managers in the division’s hierarchy would have to approve the release of the material. We knew at the beginning of the project that this would be a challenge and took it into account in the outline of the textbook. In addition to dry, detailed chapters on such things as hardware flowcharts and how a microprocessor works, we included a chapter outlining the tribulations of project including our dealings with management and with political infighting with rival projects. All of the reviewers focused exclusively on the content of this chapter, probably not even bothering to read the technical material. We won approval to publish the book by agreeing to leave out the chapter that we had included from the beginning with the expectation that that would be the result. It is a shame that that chapter didn’t get published as its insight may have been as worthwhile at the technical content.
I told this story to friend Don Gaubatz about twenty years after the book was published. His response: “I think I have a review copy of that chapter. I’ll find it and send you a copy.” He did, too.
Opportunity to meet the speaker
Research division was like a major research university without students. We had a regular train of researchers interviewing or just visiting colleagues and speaking about their research. There were so many of these that the schedule for these talks was published weekly and ran to several single-spaced pages. That’s a lot of talks; you could spend all of your time keeping abreast of what other researchers were doing. It was tempting and distracting. It took some investigation, but we were eventually able to reliably tell which talks to attend. If, following the abstract and the speaker’s bio, the announcement said: “opportunity to meet the speaker,” we knew that was a euphemism for “snacks will be served.” A highlighter and three minutes with the schedule each Monday morning and you had the week’s talks sorted.