Tom Dimock
Recollections, 1962 - 2000
High School, Fall 1962
I started programming in 1962 or 1963. It all started because my father came home from work one day and he was going to this seminar on this new Cornell computing language and asked if I want to go. I said, "Sure!" So we went to it and the next day I wrote my first computer program and I've never looked back. I figured the ACM article about CORC must have been 1963 and the CORC Manual is dated 1962 so it was either the fall of '62 or it might have been the fall of 1963. That would have put me in about the 9th grade. So I started real, real, real early. At that point most 9th graders weren't aware that there were such things as computers. Jim Holman was a math teacher at the High School and he decided he wanted to do a computer science course. So Eric Mintz and I of course said "Pick us, pick us!" What he ended up doing, he really didn't know that much and we both had been messing around for awhile so we ended up sort of taking the course and teaching him and he'd teach the rest of the class. I think he might have gone to some special training for math teachers to give them some intro into computers. (Rudan recalled that Holman had attended an Introduction to Computing Course that Dick Lesser and Rudan had given in the early 1960s.) He used to use a manual instruction book which I do not have a copy of, and I'd love to find a copy of it. It had some hypothetical assembler language that they used on some hypothetical computer. In the second half of that year, I had gotten so far as to teach myself Fortran from McCracken's book, a great, great textbook, and so in Fortran I wrote a complete emulator for that hypothetical machine and assembler language and I wrote an assembler for it. He was still using that 4 or 5 years later and teaching the course out of that same book and was using that so that the students could write programs and run them (on the Cornell computers). That same year what he did, which I appreciated much a lot more later than I did at that time, was he took a weekend job as a computer operator at Cornell. He did that mainly to get us access to the computer room. That was back in Rand. I'd go in there and submit jobs but when he got the weekend job as a computer operator, I'd go over and spend all day Saturday in the computer room programming the 220 and 1604 through the console.
We got into all sorts of trouble. Once, I overwrote the Fortran master tape with a write ring in it, no less, something that wasn't supposed to be possible. I remember Skip Williamson, (one notable troublemaker), and we used to do terrible things to him. Skippy used to come down to Upson terminal room with these two boxes full of binary cards. The card reader was just painfully slow in reading binary decks. One day he comes in and plunks them down on the counter. Sandy Pastore was working the counter at that point. We had this all pre-arranged. I went over and started talking to Skippy and distracted him while she took his two boxes and replaced them with two boxes of junk cards. So then he says "Are you going to load those or what?" She takes them and puts them in the reader where we had taped back the stop so it would shoot them to the other end of the room where we had somebody else over there with a cardboard box collecting the cards as they flew out of the reader. Of course, those binary cards were not sequenced so if they got mixed up they were (a total loss). Skippy just about died. (Rudan recalled Skip cleaning the clogged plotter pens for the plotter on the 160A by going over to Chemistry during his shift and cleaning the pens with some strong compound, which totally removed the chrome finish on the pens. Dave Pulleyn did not appreciate this initiative!)
We had a little assembler routine we wrote on the 1604, which a Fortran program could call and turn off the clock. So we'd write these great number crunching programs and the first thing they would do was turn off the clock. At the end of the run they would turn the clock back on so we'd get billed for 15 seconds when we'd been running for 15 minutes. It stretched the accounts right up. Biata caught us doing that and he noticed the discrepancy and raised hell. That was when Bessel got out his soldering iron and put a real time clock on the 1604 so we couldn't turn it off any more. This was all during high school when I was doing that. I wasn't working just messing around. Playing with the unit record equipment. Figuring out how to make the 407 and getting the wires not quite right so that all of a sudden it would start spewing out paper at high speeds. All sorts of stuff. I got pretty good on the 407 and making it do magic stuff. This was the standalone 407 in the unit record room and not the one attached to the 220.
I wrote my first program on the 220, a tube machine. Bessel was a grad student then and I walked in one Saturday and he said, "Don't step on my data link!" He had a paper tape running out of the 1604 paper tape punch, across the room and through the door into the reader on the 220. He was running networked machines. It was the first time I saw machines networked although with a little bit of latency. He was doing a two- phase compiler-compiler and the first phase was running on the 1604 and the second phase on the 220. The 220 was so much slower that it wouldn't get ahead trying to read the tape.
George Strasser was one of the people doing weird stuff. Were you there for the "Biata Simulator" incident? Strasser was working midnight to six and he would shut the machine down at six and then Biata would bring it back up at eight. One day Biata comes in and starts the machine up and it just went bananas on him. It ran him ragged. Changing plotter pens and inks, mounting tapes, demounting tapes, this and that and the other thing, putting special paper in the printer, and then about 10 o'clock Biata was really bedraggled, raving in typical Biata fashion about what a horrible day it had been so far. Then about 10 o'clock the computer stops and the console starts up and says, "Dear Bill, you have been a victim of the Biata simulator. You're screwed. George!" or something like that. Then the 1604 went down. Biata raised Holy Hell. After that George went to work for Tom Gold and then for the Securities Exchange Commission when he left here. I had lunch with him his last day at Cornell before going off to the SEC when he asked the question, "I wonder if one person could cause a stock market crash?" I figured that was good enough reason to stay out of the stock market for awhile. His name popped up a couple of years back but I can't remember the occasion. He was a character and he did all kinds of weird stuff. (On) that graveyard shift the machine was idle half the time and he wrote the typewriter piano keyboard where he could play the piano on the console and it would play it back on the speaker. All sorts of stuff.
After high school -- mid-1960s
When I finished High School I became a student at Cornell in the Engineering School. That was a mistake, I should have never been there. At that point they had absolutely no experience with students starting Engineering School who had computer background. So they made you take the Intro Course. It was crashingly boring. So I got a horrible grade in it. That was a disaster. I didn't do too good at Cornell. I flunked out after a couple of years. I never made it out of Basic Studies. In my sophomore year I had the car accident and I dropped out of school.
Air Photo Studies: That was Thanksgiving and I came back some time in the spring. I moved back into the fraternity and got a job over at Air Photo Studies working for Don Belcher's group as their programmer. A guy named Ron Shelton was the operational manager over that whole thing. They had just gotten a contract to do a natural resource inventory of New York State and need somebody to take care of managing all that data and stuff. They hired me hobbling around on crutches, a wild-eyed hippie at the time. I worked for them right up until I went full-time with MSA when they hired me. That whole project was a lot of fun. I was writing Fortran mostly and did a program called PlanMap, which was a printer-plotter for maps and had quite a sophistical modeling system in it. I sort of had to invent spreadsheets to make it all work. It produced line printer printed maps using the printer characters. It had some pretty sophisticated modeling capabilities and I did a couple of generations of that in Fortran and I was sort of pushing Fortran beyond where I really wanted to go. So one day I told Ron I needed to do the new version of it in PL/1. Bless him, he paid me to learn PL/1. PlanMap4 (or IV?) was written in PL/1 and it was the final version of it. It was still running about 10 or 15 years later in Albany. I don't think they upgraded the plotting.
NPD: Somewhere along in there I ended up getting a college degree, I got a Bachelors from Empire State College, the non-residential program. I did a lot of adhoc work for various people doing Statistics. I worked for Francese from NPD (National Planning Data) for quite awhile doing data stuff. There's a funny story there. His stuff involved reels and reels of Census tapes and so I'd be running these jobs where you had to have 100 reels of tape mounted one after the other. Of course, the operators hated doing that, they just hated it. So what I would do is show up out at Langmuir at 10 o'clock at night and I'd mount all my own tapes. The guys didn't care as long as it meant they didn't have to do it. Well, I was out there one night about 3 o'clock in the morning, about halfway through this huge set of tapes, when the operators were John Skope and Terry Norman, I think, were the operators on that night. That was a crew! They had both gone upstairs to play Ping-Pong and I was the only one in the computer room when Pulleyn walked in! Oh my God, you had thought I had set off a bomb in there. Holy Mackerel did that generate trouble. Pulleyn held that against me for years.
CIT
Systems: I worked for various grad students doing Statistics stuff. Just freelancing wherever I could. I finally got a job in Systems for a little while. I worked for Systems for 3 weeks in which time I wrote them a manager that managed the password data set for MVT. Cogger hired me and I was working with Bob Lent who was there at the time, Shames was there, that whole group and Herbie was there. I managed somehow to create the password dataset without having the password to the password dataset. It was equivalent to locking the keys in the car. Herbie had to bring up a VS1 partition to crack the password dataset to fix that one. That (job) was on a little contract basis.
Then, all of a sudden, out of a job again. Right then was when Al Doniger got drafted and so he promptly enrolled in rabbinical school and disappeared with like zero notice. He was the sort of central programmer on the payroll system and they were looking for someone right then who knew PL/I. So I walked in Charlie Evans' office and said, "Hey Charlie, I hear you are looking for a programmer?" Right then Dave Bessel sticks his head in behind me and says, "Charlie, I wouldn't touch this guy with a ten foot pole!" I says, "Dave, with friends like you, who needs enemies!" Charlie just says, "Oh, I don't know. Coming from Dave that's a pretty good recommendation."
Payroll: They hired me and the first thing I worked on was the check-writer program. Maybe it was Pauline Shih, one of the oriental grad wives who had been working on it, who left for the other end of the country or something, as her husband had graduated. So, I took that over. I opened up the folder that I was handed and in it was a flow chart in Chinese. That was it, that was the total documentation! I said, "Charlie, you got to explain to me what this is supposed to do because I'm kind of at a loss here?" He did his hand waving and I went off and wrote code. That was just panic time from A to Z. We again had, as with the 1998 PeopleSoft HR/Payroll installation, a non-negotiable arbitrary deadline. We came right down to the deadline and I had gotten pneumonia and I was home just sicker than a dog. The phone rings at 3 o'clock in the morning, it's Hollenbeck saying, "Look, we're really in trouble here and can you come in." I says, "You got to be kidding?" He says. "No, I'm not, we're not going to get it up here!" I said, "Yeh, I'll come in, but I'm not going into the computer room." I was pretty sure that I would die if I went into that air-conditioned computer room. I went out and ensconced myself in my office and I had people just running in all directions. I was just debugging, debugging, debugging, and sending people off with stuff to do. Dry running that again and then I take a look at another program that was failing. I sat there for 36 hours, well over 30 hours for sure, a long time, and we got the payroll out with remarkably few errors. I went home and collapsed and slept for about 30 hours! The group involved was Hollenbeck, Charlie Evans, Gale Sprague, Libby, I think, and Al Seliga. I worked on Payroll for awhile more.
Library: I did a whole bunch of things. I did the Library systems for awhile. That was kind of amusing because that was all Gale Sprague code. Lord save us from some of his code. He used to write his comments but he never looked at what he was doing. He had this alarming habit of having his fingers get off one key he was a touch typist and they'd get off one position on the keyboard and all of a sudden his comments would dissolve into gibberish. It would go on like that for line after line of just total gibberish. (Even if translated by reshifting the keys) his comments were not that useful. It was the definition of spaghetti code and it had a lot of problems and it just didn't work very well. There was this variable creatively named "n", which was a global variable and it was just spattered all across all the modules of the Acquisitions System. Nowhere was there any definition of what it did or what it meant. Finally, it was causing me too much grief not to know what it was and so I sat down and spent three solid weeks trying to figure out what it did. At the end of three weeks I came to the final conclusion and convinced myself that it did absolutely nothing whenever it was tested it always had the same value. Without asking anybody I just stripped it out of all the code and it worked fine and nobody knew I ever did it.
That was about when we hired Bill Turner. The first thing I had him do was sit down and go through every module. That was written when PL/I was a real new language and they used "redefines" to layout the 461 character acquisitions record. I had him basically take coding sheets and stick them together for 461 columns wide and go through it and map out all the "redefines" from every module and find discrepancies. They were all over the place. It was a wonder it worked at all. In different modules the fields were off from each other by a byte! It was just insane. We cleaned all of that up but it had this nasty habit when it dealt with encumbrances. It used encumbrances when you ordered a book and when it came in and the bill got paid you reduced the amount and got rid of the encumbrances. The net amount of the encumbrance account would go negative all the time. One processing time a whole bunch of them went negative and they called me in and said, "Look, you just have to get this fixed." We just couldn't make it come out right and so I sat there and worked and worked at it and found a bug that could be causing it and fixed it. It went negative again. I found another bug. I found 6 bugs that could be causing the problem before I found the one that was causing the problem. It turned out that the root of the problem was the special utility program we had to reset the encumbrances after they had gotten screwed up, was itself wrong.
Responsibility Center Analysis: I also wrote the Responsibility Center Analysis for Jack "O" (Ostrom). That is when they had a whole new approach for doing accounting at the University and make every unit responsible independently. That's when I really wrote a spreadsheet program before there were spreadsheets. That thing was a spreadsheet. It worked really well but I had real mixed feelings about the whole thing because it caused the University to start doing some really bad things. It was really weird. Like if the Arts School needed a course taught by someone who had legal expertise but since the internal charging rate for a Law School person was so arbitrarily high because of the system, they would instead go outside and hire somebody from Ithaca College. In their attempt to optimize the use of resources internally they were actually globally "pessimizing" them and forcing us globally to make very bad decisions.
Disk space charging: At one point I got the task of trying to rationalize the charging for DASD space. I actually did a pretty good job. I really did go and try to figure out exactly what it was costing us to have a megabyte. I got the whole thing finished and proved that basically we were overcharging for DASD by at least a factor of 4. The whole thing got trash-canned. In classic Cornell fashion they were using DASD to subsidize something else.
Student Record System: The SCT project for the Student Record System? That one was a sweetheart deal between Hank Vaughan and his buddy who owned SCT. If I had any doubts about that, I was on the testing team and I was testing the Registration system. They had run their acceptance tests and all was fine. Well they ran their acceptance tests on a total student body of like 30 people. So I ran it on 300 people then 3000 people and it was not looking very good at all. I went into the code and it turned out it was not only not linear but an "n squared" situation. When run with our entire student body it would not complete within the meantime to failure of our computer. There is no way. It would literally take multiple days and we couldn't run for multiple days without our computer crashing. So I wrote that up in my report. You know what Hank's reaction was to that report - he took me off the testing group! We proceeded another six months before it became obvious to everybody that the system was not going to work and the whole thing went down in flames. Byron McCalmon, the then Registrar was heavily involved in the project. But it was Hank who protected the project because he was a college classmate and a good buddy of the guy who ran SCT. There was definitely some relationship there even though he crucified him when the project failed. Well nobody came out of that project looking good. We lost a lot of money on that project, roughly $300,000 to $400,00 went down the tubes even though it seems a pittance these days.
Traffic: Hank and I never got along very well. He never, never, ever, ever, believed I was doing things in the best interest of Cornell and he always thought I was doing things for my own benefit. It got really aggravating sometimes. He'd do such incredible things. I don't remember who it was, the Head of Traffic at the time (Bill Crissey?) wanted a new traffic system. This was when we were still in Rand Hall. So, I got put on the project of designing a new traffic system. This was going to be an online system using Hazeltine 2000s or Mod 1 terminals. It was our first attempt at writing any kind of online system. There was basically nothing out there, no tools available for doing this, so we were inventing it all from scratch. I came up with a neat database design, which used VSAM files and incremental files to get speed. It actually ended up that a grad student from Computer Science came over and studied what I did and went back and got his Master's degree basically on that design for his Master's thesis. I had Al Seliga and Chester Williams working for me. The estimates I had done came in way too high. There was no way Vice President Gurowitz, I think it was, was going to approve it. So Hank and the Head of Traffic cut themselves a little deal on the side quoting a cost that was considerably less than it was going to cost and Hank would just eat the overrun without any complaint. We went ahead with the project. We were three weeks away from implementation of that system; it was basically done, when the Head of Traffic took another job. A new guy comes in, Bill Wendt actually, and starts going through the status of existing projects. Nobody told him of this deal and "Boom" he goes to Gurowitz and raises hell and of course Gurowitz is completely in the dark on the whole thing. The "Shit" rained down big time and Hank got called on the carpet and I guess he got chewed up pretty seriously. He comes back to Rand Hall, storms into my office and starts screaming at me and blaming the whole thing on my incompetence. This was one of two times that I totally lost my temper at work. I just blew up. I mean, everybody in Rand Hall, all three floors, heard that fighting. It went on for about half an hour just screaming at each other. The next day he came in and apologized. It was the only time I ever heard him apologize for anything. The whole project got canned and never went live.
DeBoer report: The DeBoer report came along somewhere in there. Everybody else left. I was one of the half dozen people who stayed. We just kept our heads down. Actually the Traffic thing may have been close to the DeBoer report. It was somewhere in along there somewhere. It was one of the few development type things that got done at all. Mostly we were just doing maintenance and the systems were getting more and more screwed up. Then of course, Ken King came along. Actually, I did a very funny analysis of my salary over time corrected for inflation. It was really funny because I started out when I was hired as a programmer trainee at I think $8700 per year or something like that. My salary went up gradually and then Hank and I are in the same building and it goes down and then Hank got fired and it went up again. You want graphical proof that Hank and I clashed all the time - that was it. (Rudan noted that in his oral history interview with Hank, Hank expressed admiration for Tom's skills)
Printing reports from APL2: He and I had a very interesting thing about that. Hank loved APL. He decided he needed to show off his technical expertise so he wrote a time tracking system for the department in APL. APL2 had just come out which finally had support for printers; it could actually talk to a printer. So, he had this time system that had no print support. He'd gotten hold of a copy of the APL2 manuals but he made sure nobody else could get hold of a copy and he was going to put in this print support. He was getting absolutely nowhere, he'd spent three weeks on it, and it was not working. On the next Monday he was going to have this big presentation of this system to the department. But he was out of town on the previous Thursday and Friday. So he left on Wednesday actually about noon. So, Wednesday afternoon I went to Art Johnson and said, "Hey, Johnson, what's the password to Hank's account?" He told me. You know Art, he would do anything. So I got on Hank's account and I went across the street to Uris and got a copy of the manuals, from Gimbrone or somebody, and went back and fixed all the print support. So Monday, Hank comes in and he's got this big presentation scheduled for 2:00 o'clock in the afternoon. At lunchtime he's sitting there just checking out making sure everything is working and I walk up to him and say, "Oh Hank, last week while you were gone I fixed the print support and your reports should print." He stiffens right up and says, "How did you get on my account?" I said, "I asked Johnson for the password." He says, "Are you trying to undercut me?" I just said, "I'm telling you now" and I just walked away. I never said a word to anybody. The demo worked, the print support worked. He took credit for it but he backed off me a lot after that. I mean, it was just flat out blackmail. He'd just gotten to be such a pain that I had to do something - "Hank, you're not that good technically, I'm better than you technically, I've always been and I'll always be!" In contrast to Hank, Hollenbeck was the ace at hiring good people and ones who were smarter than him. He was just great at it. Hank was deadly afraid that it would become apparent to someone that he actually had people working for him who were smarter than he was.
Adabas: What can you say about the evaluation committee that selected Adabas as the database management system for Cornell? There was another couple of people involved, but I can't remember whom, even though the project was almost all my work. Jim Quiggle and I wrote the report. That happened when they hired Ken King. He was going to start January 1 and he came up to campus, after they announced the selection in November, and one of his conditions of accepting the position, was that by the time he showed up on campus we would have selected a new database and a 3rd/4th generation programming language for administrative development. One of the things he was charged with doing was doing something about the mess that administrative computing had come to after the DeBoer report. Although I may be off on the dates, I know that Adabas came in in January. Whatever, the selection was done between the time he was hired and the time he came to campus. We did have a bit of stuff running in IMS but it was apparent that that was not the way to go in the future. We looked at all the databases out there and to my mind; there was absolutely no question of selecting Adabas. The only real competition right then was IDMS from Cullinane and thank God we didn't go that way when CA later sucked up Cullinane. One of the main reasons I selected Adabas was that Adabas was very amenable to on-the-fly changes. Our customer number with Adabas was 300 something so they had been around for awhile. It was the most expensive one out there at the time. I was quite surprised when Ken went ahead and approved it, but they had Adabas running where he came from so he was familiar with it and comfortable with the choice.
When we were negotiating for Adabas the Software AG salesman was a guy named Don France? I think, to this day, he was the most outstanding salesman I have ever met. He went from rookie salesman with Adabas when we started dealing with him and within two years he was executive vice president for sales for all of Software AG. He was just astonishing. I mean, he would come in and if he didn't know the answer to a question, and he knew a lot of them, he would say, "I don't know but I'll find out." You always had the answers to the questions within 24 hours. He was just amazing. One of the things we asked him was, "Would it run under VM?" Without batting an eyelash he said, "Yes!" It turned out they had this sort of renegade group of programmers who had actually gotten it working under VM but it had not been announced as a product at that time. And, so we were supposed to install on January 1 and Don comes up with this tape and we installed it under VM and started doing our development on the Admissions System, the first system we built with it. That was our learning curve and we were supposed to go live on June 1. So we had 5 months to get up on the learning curve with Adabas and Natural and build a system. Insane schedule! So we installed Adabas VM and it worked. It was slow but it worked. We plugged along and we were supposed to go live on June 1 and we went live at 4:00 in the afternoon. We actually made the day, late in the day, but we made the day.
During the development, the whole thing was that Ken had approved running it under VM for development only but we were going to run live under CICS/MVS. So we had these two parallel efforts - we were developing the system under VM but we were also trying to get it all working under CICS. We were just having nightmarish problems. 3270 clusters would lock out for 15 minutes at a time and neither Software AG or IBM could tell us why - it was really bad. About the end of the first week of May with three weeks to go live and we still do not have a production environment. I went to Ken and closed the door and said, "Look Ken, we're in really trouble here. I don't know what to do about it, but, if you let me take it live with VM we can make the deadline." He thought about if for a second and he says, "OK!" So we did! We went live with CMS and it was really slow. It was really slow. Larry Chace got involved and we started looking at it. It was Adabas and we had absolutely no source code of any kind. One afternoon Larry says, "There's something really wrong with the Ada I/O module, I got to figure out how it works!" Starting at noon on that day he wrote a dis-assembler in assembler and dis-assembled Ada I/O, went through the code and figured out how it worked, commented the code, so we had a commented copy of the Ada I/O module, and he fixed it! And vroom, it ran 10 times faster and we literally got an order of magnitude improvement in throughput.
While that project was going on I had gone off to the Software AG conference in Florida. It was the first time I had ever been to a Software AG conference as we were a brand new customer. So, I get down there and I'm looking around for anybody who knows anything about Adabas/VM because we were in the middle of this total nightmare of performance. They finally point me to this guy who's in charge of Adabas/VM and I go over and say , "Hi, I hear you're the Adabas/VM person!" He says, "Yeh, what do you know about it?" It still hadn't been released as a real product. I says, "We have it, we're running it, we're doing our development under it!" "You do! Where are you from?" "Cornell University." There's this long pause and he says, "That's where that tape went!" It turned out that Don France had gone into the development labs and stolen the tape to be able for us to install it, as they still did not have it as a product. So we got along real well and it was fun. We subsequently ended up selling those mods back to Software AG, something which people had told me couldn't possibly happen because they were very not-invented-here type of people at that time.
But that conference was so funny. At the Adabas/VM Special Interest Group meeting a couple of the places that officially had it as Beta-testers were there and so of course, I was there. There were some Software AG people and I didn't know who anybody was at that point in time. I got up and was just bitterly complaining about the performance of this and that and the other thing and I just got into this shouting match with this person from Software AG about what a piece of crap this was, the support was very poor and it was incredibly slow and yada-yada-yada. After the meeting breaks up somebody comes up to me and says, "Do you have any idea who that was who you were screaming at?" I said, "I have no idea, and I don't care because I was right!" "That was Peter Schnell, he owns Software AG!" "OH!" He says, "You have screwed yourself with that company so badly, you will never dig out of that one!" Well it turned out that was absolutely wrong. What happened was that Schnell was being arrogant German and of course he had to be right. Well he went back and found out that he was wrong and I was right. The result was that he remembered me very well and we actually ended up getting along great. We got along fine. Subsequently, they actually referred a couple of consulting jobs to me where people who were having trouble with Adabas/VM performance. They'd refer them to me and I go out and deal with them. We got good performance under Adabas/VM and ran it there for years and years and years. Now Adabas is under MVS but Natural is still running under VM. I was in charge of Adabas for over 10 years. It's solid like a rock.
1980s
Tapes: From the 1980s I remember an amazing fight with (Bob) Tucker over tapes. At one point we had to retrieve some stuff off a backup tape and we couldn't read the tape. I started looking into this and I discovered that a lot of our backup tapes weren't readable. So, I starting looking at it and it turns out that all the bad tapes had the same brand name on them. They were from some cheapy company. Tucker had gotten a real deal on these tapes. I was yelling at Tucker and he was insisting that it couldn't possibly be the tapes as they were perfectly good tapes and all that. We got a brand new one of his tapes and I wrote a little program to just write it full of stuff and read it back. I had Dick Schriber from IBM put an oscilloscope on the tape drive and, in a single pass on that brand new tape you could just watch the signal level drop off until about 2/3 of the way down the tape when it disappeared into the noise. The tapes were shedding oxide so badly that you could not go all the way through a reel of tape when the read heads became so crudded up that they couldn't read the tape anymore. I said, "All right, read it and weep! You may think those tapes are OK, I say they aren't. I donąt ever want to see one of them on my Adabas backups again!" So we started buying a better quality tapes for backups.
Changes: We were putting up stuff like mad in the 1980s. It was a very intensive development period. A big new traffic system, student. It was right then when Adabas and Natural came in that we installed the 3270 terminals. That was why we selected Natural. Natural was a painfully immature language at that point, but it had two things that made me say we are going to do this and we'll deal with the problems as we have to. Those were; the access to the database was fully integrated into the language, and the access to the 3270s was fully integrated into the language. Any other language, like PL/I or something, you had to have these imbedded command strings and all that nonsense. I felt if we were going to have all of our programmers be able to write that stuff we are going to have better language support and so we went with Natural.
CUSSP: In that period, Ed Hollenbeck was the director and then Russ Vaught came along. Working with Ken was great. He'd just say, "Make it so!" And off you would go. We got amazing support from him. Then Stuart came in. Stuart had great vision but sometimes it was kind of hard to figure out what the hell he was talking about. He'd been ranting on at Mark and I about stuff and finally I just said, "Stuart, what is it you want me to do?" And he just sort of stopped and said, "Make distributed computing work at Cornell." and he walked out of my office. And I'm sitting there going, "Make what work?" I had no idea where to go with that. But I let it stew around in my head for awhile. Then one day there was an article published in the (Cornell Daily) Sun saying we were going to do touch-tone registration. We never tracked down how that got started because there was no way we were considering touch-tone registration, at all. It was kind of embarrassing because of this article in the Sun, but Stuart said it would really be good if we could give them something, so why don't you guys think about it. Maybe we could do something that was sort of equally amazing but different and we could say they were just confused.
So we thought about that. Mark finally comes up to me in the hall a couple of days later and says; "Do you think we could give students access to their grades?" Hmm! By this time we were starting to have networks then, and there were Macs out there connected up to things. Remember also that Macs came in on Ken's watch. So I thought about it for a little while, not for long, as I was still right there out in the hall, and I said, "Yeah, I bet you we can make that work!" So a couple of us knuckled down and that's when CUSSP (Cornell University Stateless ? Protocol) got invented. That was just a handwaving thing a couple of days later. We built a prototype and it was programmed in C running on the mainframe doing direct calls to Adabas. CUSSP and the client were in Hypercard on a Mac. We got it working in 5 weeks. We put it out in Day Hall - one Mac sitting on a table outside the door to the Registrar's office on the second floor and we had one in Financial Aid, I think. Nobody much saw this second one as the main one was in the Registrar's office. We'd hide down the hall and peek around the corner and watch to see what was happening with it. It had no instructions, it just had a splash screen and you'd have to figure it out yourself. It was pretty amazing. Within 5 weeks, between then and the end of the term over 1/3 of the students at Cornell had used that to look up grades. It was total garbage at a technical level but it worked. It made the user happy. It made them so happy that then we couldn't take it away and so we continued to have to support that horrible lash-up until we could build a little bit more industrial strength version.
Macs: The Macs. That was great! They showed up and we were all just going, "Wow!" Actually what happened was that they brought in a Lisa and showed that off. When I first saw that I said, "Now there's a little computer I can deal with!" Up to that point I had just completely resisted having anything to do with PCs, or Amigas, and whatever else was out there. I said, "Hey, these are a not very smart version of what I've got access to every day. Why would I want to deal with them? That one was different!" If I were ever going to be a salesman, I would want to sell something like that as they were pretty neat. Then the Mac came out and of course we got those. Then the laser printers fairly quickly right after the Mac. The $10,000 Laserwriter. I can recall standing around in the basement of Day Hall watching those first pages come out of that laser printer and everybody just going, "Wow!", because nobody had ever seen a laser printer before. We were using Diablos, Decwriters and fixed font machines. It just wowed people. We then put together the whole Loaner Program where we would put Macs into an office and then take them away if they didn't pay for them. Those printers were the devices that sold the Macs just using Macwrite and the Laser printers. MacWrite spit out Postscript to the laser printers. You could put out beautiful things. I remember Quiggle did the Traffic System so that it spit out Postscript. Oh boy, you didn't want to look at that code.
Mandarin: Then Mandarin started up. That's when I separated ways from doing Software AG and I never went back. After that I was on the Mandarin crew and doing that stuff. That was a lot of fun. Lonny Spung and those guys out of Apple. We did all those road shows. The Apple VITAL concept was way too complicated and I told Lonny that. Lonny didn't understand why I continued to work at Cornell. We got into a deep conversation about that one night down at the Chapter House where we had gone down there and we were drinking beer. He was going on about my working at Cornell where no one appreciated what I was doing. I said, "Lonny, I have a life. Here, I work here for eight hours and then I go and I do other things. If I came out and did what you did, I wouldnąt have a life and I don't want to do that. Thanks, anyway." As far as I am concerned things started going down hill as Mandarin phased out.
CIT's reputation: We tried the team-based movement but that was just another management trend of the month. We went through a lot of that. Koehler was into management trends. I think Koehler was a lot of fun and I really love Dave but he did get into some strange fad-of-the-month things. It was amazing that we were going to try to double of productivity given how high it was at the time. For a long, long time, Cornell has always had an administrative computing staff that was way, way above average. Well we were appreciated, but never at Cornell. Stuart understood it. Ken understood it. There was that one wonderful article in the Sun where they were talking about Housing and Dining and how screwed up they were, written from the student perspective. It had this one line in it which said, "Now I'm not saying that Housing and Dining is actively evil like CIT!" I showed that to Mark and said, "We've got to do something about our reputation out there. This is not good!"
Computing at Cornell has never done a very good job of presenting themselves to the world. We over-delivered and under-priced and thatąs a dangerous thing to get into, because people get to expect a lot more than you can actually deliver. You can only do it for so long before you start burning people out. We've gone through that more times than I want to think about, where we go into these crash modes and we just burn people out. But, that's how we got things done.
Project 2000: Project 2000 was right up there with the DeBoer report era in terms of "Oh Boy". A), I thought the technology was pretty awful from the get-go and B), they formed another database selection committee that recommended Oracle and so of course, we got Informix instead. Basically, the technical people were completely ignored. Everything we recommended, they went a different way. Decision by "executive suite" indeed. Then once we made this commitment to PeopleSoft, they started sucking up all of the resources and all of the people and that was when a lot of Mandarin people just left and went off with Cogger, wherever they went. I was very unhappy at that time but I finally decided that I had better shut-up because they had those layoffs and before I ended up like Bob Cowles.
You know, I don't know and I don't care about Bob's dismissal but it sure as hell looked like he was sacrificed, which didn't matter what the real reason was because how it was perceived was that he was sacked for saying "the emperor has no clothes!" I mean, it shut some of us up who were also saying the same thing. I was still extremely unhappy and it was pretty evident that I was unhappy. At one point, Lambert comes by to talk to me about how I feel about everything that is going down. Judy Hart was with him and I would not been nearly as honest had she not been there. The two of them came in here and we closed the door and I just let him have it with both barrels for about an hour and a half exactly of what I thought of what was going down. I told him how unhappy I was that he had taken an absolutely world class distributed computing development group and destroyed it because there were almost none of us left at that point just gone. I just told him that, "You have put all your eggs in one basket and you have got no safety net left, you have destroyed it." He wasn't really happy at what I had to say. At that point, quite honestly I didn't care if I got canned. I didn't think it was likely with Judy being there. If she had not been there it would have been a lot more dangerous thing to do. I think Lambert left because the whole thing started going sour on him. I think he read the writing on the wall and he decided to get the hell out before it really came apart. His reputation was not going to be enhanced with Project 2000.
(The interview ended with a frank discussion about Project 2000, most of which is not worth transcribing for the record.)
Prepared by John W. Rudan on July 24, 2001