From Harvey

Thanks for sharing this. It's fun to look back now. I've added John Dawson to the email chain.

It's unfortunate I didn't get your first message a little sooner. I left AT&T Labs April 30 and trashed lots of old files, including correspondence with Microsoft from that era. Here's some things I recall off-hand:

• Steve Ellis and I did several trips together, hunting for applications to be ported to the 3B2. When we visited VisiCorp in San Jose, we were amazed by their cutting-edge open-plan office arrangement. (Let me assume that was meant to be sarcastic. "Open plan office arrangement" is a euphemism for 'cube farm'. Or, even worse, the cube farm without any walls, not even waist-high. And it was hardly a new concept in 1983. My first two jobs -- at IBM in 1961 and Bell Labs 1962-'66 -- my own desk was in an open plan office. Bell Labs had ten people in the office with no intervening walls. My IBM office had several acres of desks with no walls.)
• When we visited Lotus Development, one of their officers (I think Jonathan Sachs) picked us up at Boston Airport in the kind of messy and worn Volvo station wagon a grad student might drive. In light of Lotus' fantastic business success, this was unexpected. When we mentioned it, he explained that his new Jaguar was waiting at the dealer, but he'd been "just too busy to pick it up." The Lotus execs were certainly interested in AT&T's deep pockets, but their developers were shocked to learn that we wanted Lotus 1-2-3 to be re-written in C and to run in a multiuser multiprocess environment. They wanted us to give them the 3B2 processor instruction set, and the only way they would port Lotus 1-2-3 was as heavily-optimized assembler with total control of the machine. (A software developer would get immediately what Harvey is saying. Anybody else would not. I started to write an explanation for non-developers, and quickly devolved to, "Don't even bother trying.")
• Logitech at that time was primarily a mouse company, but they had the rights to a spreadsheet that was (IIRC) written in C. When we visited, they had a gaggle of developers working on other projects around a single large conference table, with several copies of "The C Programming Language" (including one in Japanese, which we thought was portentous). (Either 'portentous' or 'pretentious' would be appropriate here.)
• An independent developer in Utah, Dick Kreutzer, had a pretty good spreadsheet called UltraCalc. It written in C and compatible with UNIX. He had visions of becoming wealthy, and was something of a pain to deal with, but I managed to get a contract and we actually offered UltraCalc on the 3B2. Meanwhile, somebody at Western Electric negotiated a separate and more expensive contract with Kreutzer for the same rights that I had obtained, illustrating the company motto: "At AT&T, communication is our business; not our policy." So, Kreutzer may not have gotten rich, but at least he got paid twice.
• The 3B2 was (IIRC) originally targeted at the financial community. Of course, the ability to compute interest is central to this market sector, requiring floating-point calculation. When Western Electric delivered the 3B2, their FP co-processor was still in development. Modifying the UNIX kernel to include a FP emulator was-- for some electropolitical reason-- anathema to Western Electric, so the 3B2 limped along performing floating-point calculations with maximum inefficiency in user space. When it came to computing interest, a vanilla 8088-based IBM PC could mop the floor with a 3B2. Wall Street was seriously underwhelmed. (I don't remember that story. I remember one about Intel's Pentium floating point co-processor, which was a bug on the released chip. That sounds to me worse than a late chip. But history shows otherwise for the only score that counts. The Pentium was a commercial success anyway.)
• A story not directly related to the 3B2, but illustrating the state of AT&T's legal organization, was my first patent. In 1981 I developed an algorithm that used a speech detector as input and calculated the correct times to play "audible backchannel responses". The AT&T patent attorney insisted that software simply could not be patented, and refused to file an application. I re-submitted it 20 years later, and it was granted

From Steve

Thank you for sharing. I loved it! It is fascinating, well-written and funny. It brings back so many memories of an exciting year of my life at Bell Labs, when I found skills I never knew I had and managed to do things that I never thought I could do. I do remember a great sense of accomplishment and "never having a dull moment." While my memories of that year are not as coherent or as detailed as yours, I do have some comments and anecdotes (some of which may be apocryphal) that I want to share with you. I also want to say "Hi" to Earl, who was one of the best teachers I ever had in my career. I agree, Earl was a wonderful guy to work for. He did good things for my career on at least three occasions I can think of. (And that's not counting my education in contract law.) Perhaps the most significant was pushing me back to my computer technology roots when my other managers were pushing me away from it and towards management skills. As it worked out, I was happier as a techie; Earl saw that and restored my confidence and competence as one. And a few years later, I left management, returned to being an engineer, and never looked back.

• Either because Earl chose to start the software acquisition small, with only your group involved at first, or because I and several of the people in my group didn't want to drop what we were doing to become software contract negotiators, I and most members of my group did not attend Paul Hoffman's course in software law. (I had originally written that the whole department attended. I stand corrected, and fixed that in the story.) Harvey was actually a member of my group, not yours, but he volunteered to join the software acquisition team while remaining a member of my group. (I had also remembered that, but it wasn't important to anybody but you, me, and Harvey, and would have distracted from the story.) I, and the members of my group other than Harvey, did not participate in either Hoffman's training or the first round of vendor negotiations that you described. As time went on - probably a matter of a few weeks after you started - I succumbed to Earl's persuasive powers and agreed to turn myself and my group into a software acquisition team. That caused a number of the MTS to leave the group almost immediately, leaving myself, Harvey and Tom Miesner (a TA at the time, but subsequently promoted twice to MTS because of his software acquisition work) as the only contract negotiators in the group. I later added a contractor to administer our 3B machines.
• I was heavily involved in negotiating for a product called "PC Connect" from a small software-development firm in Santa Monica, CA by the name of Locus Computing Corporation. I visited that company several times to meet with Jerry Popek, President of the company and a fairly well-known computer scientist. (Jerry passed away in 2008.) PC Connect allowed Windows PCs to access and share data with Unix-based computers, and we desperately wanted that capability for our 3B2's. Well, at some point after we signed the contract, Jerry decided that we owed him a down payment of $75K in software portation fees, and he threatened to stop working on the port until we paid him the money. Somehow - in that year of software contract negotiation, we came to feel that we could work our way through any obstacle - I manged, within a day, to get high-level approval for a $75K check to be cut by Bell Labs purchasing, payable to Locus, and a ticket to LA. The next morning, I flew out to LA, drove to Santa Monica and met Jerry for lunch at a beautiful restaurant. We had a lovely meal in the open air under the glorious California sun. I gave Jerry the check, we shook hands, and I drove back to the LA airport and caught a late afternoon flight back to NJ. (Something you don't say in this account: did the contact call for the payment?)
• Dave, at some point after you left the department, a meeting took place up in Basking Ridge to review the packaging options for the 3B software that we had contracted for. I remember that there were about 20 people at the meeting, including Joe Timko and his counterparts and their underlings from AT&T Product Management and Marketing and possibly Western Electric. If I recall correctly, the Bell Labs Technical Writing organization was charged with designing the packaging, and they brought prototypes of their proposal to the meeting. As you know, PC software was being delivered at that time in cardboard "boxes" (resembling books, to fit on a bookshelf) that contained the software documentation in a thick 8" high 3-ring binder that slid into an outer shell, with the software on 5.25" floppy disks in a sleeve inside the binder. The Tech Writing team's design included an attractive combination of gray (for the outer shell) and rust - or some similar color - for the binder. They also presented a combination of fonts and "look & feel" for packaging and documentation with an AT&T look. The only problem with the design was that, in an attempt to make our packaging distinctive, the prototypes had a unique feature: there was a small triangular slice cut out of both sides of the box sleeve on the top. Not only did this give our software packaging a unique look, it also made it easier to slide the binder out of the box; other vendors sometimes cut a half-moon notch out of both sides of the outer box to accomplish the same thing. Well, the meeting ran on for at least a couple of hours because neither Joe nor his counterparts were willing to take responsibility for making the decision as to whether to go with the triangular notch on the outer box. Everyone knew that it would present a problem for manufacturing the boxes - even if it was possible to mass produce cleanly, it would add significant cost - but the VP's didn't seem to want to say yes or no. I think that the meeting finally ended without a decision being made. Eventually, down the line, the decision was made to eliminate the triangular cut-out and to go with a standard box.
• Another story that I think is worth telling in detail -- by someone with more inside knowledge than I -- was how late in the process we realized that, when vendors would start delivering software to us, we had no way of testing and certifying it. I do remember that, virtually overnight shortly before the first contracted software was scheduled to be delivered, Bill Schnarr's department was reorganized into a software acceptance team.
• Early in 1984, after we had been acquiring software for a year or so, I was asked to give a status report to Lee Tuomenoksa, Joe's boss, in his office. Joe Timko was in the meeting, but I can't say for sure that Earl and Steve Bauman were there. If they weren't there, I soon found out why. I proudly told Lee that we had signed contracts committing AT&T to paying around $17M to software vendors in software portation fees and guaranteed royalties. The initial look on Lee's face was priceless! I think that Lee might have fired me on the spot, except for Joe's presence and the respect he commanded. It was clear to me that Lee didn't fully understand what we were doing nor why we were doing it, but he accepted Joe's assurance that it was necessary and proper. Of course, within a month or two of that meeting, the AT&T Product Management organization wrested responsibility for software acquisition for the 3B line away from Steve and Earl, and we were all left to find ourselves other jobs within the Labs.

Last changed - 5/27/18