1994 TSI: The Second Crisis

Posted on May 24, 2021 by Mike

The I in TSI comes to stand for Incorporated. Continue reading →

This entry requires quite a bit of background.

When we were still living in Detroit, Sue Comparetto founded TSI Tailored Systems as sole proprietor. I helped her occasionally in the early days, but for the most part she did it alone. She never had any employees or, as far as I know, a business plan. She inherited a handful of accounts from her former employer. At first she had an office in Highland Park, a small and dangerous city surrounded by Detroit. Then, when TSI somehow obtained an IBM 5120 computer, she set up shop in the spare room in our house in Detroit.

Having the computer in Detroit allowed me to learn BASIC. Having access to the programs and listings from AIS, the company that wrote most of the software that Sue supported, allowed me to learn how business programs could be structured. We were self-taught. I had taken exactly one college-level programming class at Michigan in 1966¹; Sue had none. Neither of us had ever taken an accounting or marketing class. In fact, neither of us had ever even sold or helped market anything.

When we moved back to Connecticut, Sue registered TSI as a partnership. We worked together, but we never really agreed on who was responsible for what. I considered myself much better at programming than Sue was. I therefore expected to do the bulk of the coding (including software for TSI to use) and for her to handle nearly everything else. The way I thought of this was: she does the phone stuff; I do the computer stuff.

The first additional task that I felt obliged to take over was marketing. In Detroit Sue had never needed to find new clients. She was given a bunch of them, and she hoped that IBM would provide her with additional leads. When we moved back to Connecticut, however, we lost the ties with the Detroit IBM office, and it was difficult to make new arrangements. We had only a few clients and lousy credentials.

I copied company names and addresses from the Yellow Pages.

We scrambled to get a few custom programming jobs. I did nearly all the design, coding, implementation, and training. I pulled together a mailing list from phone books at the library and wrote letters to businesses that I thought might be interested in systems designed for our clients. We never made a lot of money this way, but it did generate some business. Eventually, IBM also gave us some leads.

We hired a receptionist/bookkeeper, Debbie Priola, and a programmer, Denise Bessette. The former freed up time for Sue almost immediately. The latter consumed quite a bit of my time for a couple of months, but eventually she helped a lot. Unfortunately, she decided to return to college and cut back on her hours at TSI. More details about the early years of TSI can be read here.

Both Sue and I found most of the decade of the eighties to be enjoyable but frustrating. The programming was fun and very challenging. Almost all of TSI’s customers appreciated our approach. However, we never came up with a good way of monetizing our efforts. The ad agency system, GrandAd, did better than the “anything for a buck” approach that we had been forced to use in the beginning. However, our market was effectively limited to agencies that were within driving distance and were too large for a PC system. In that reduced market, it was difficult to make enough sales to get by. Eventually there were so few reasonable prospects remaining that a change in strategy was essential.

I was convinced that our future lay in selling AdDept to large retail advertisers across the country. There was no real competition, and there seemed to be a good number of prospects.

I don’t think that Sue agreed with this change in focus. She had always favored local businesses over large corporations when purchasing something, and I am pretty sure that she also preferred dealing with smaller businesses over dealing with corporate executives. The fact that both of our first two AdDept clients declared bankruptcy and left us with tens of thousands of dollars of noncollectable invoices reinforced her attitude.

Sue had always been a night person. I was the opposite. I always was out of bed by 5AM or earlier. I usually became very sleepy around 9:30PM. I then took a shower and read a few pages of a book in bed. I was almost always asleep within a minute or two of turning off the lights. I stuck to this routine for decades, and I still do in 2021.

At some point in the eighties Sue developed a sleeping problem. She liked to watch late-night television, but she almost always dozed off in her chair. She slept very fitfully, waking up with a start and then falling back asleep. This went on for a long time—months, maybe years. Finally she went to a doctor. He prescribed a sleep study. It was not a surprise that it confirmed that she had sleep apnea. For reasons that I have never understood Sue was reluctant to purchase and then use the sleep machine. The models sold in those days were big, expensive, and ungainly. Even so, breathing well while sleeping is critical to good health.

I suspect strongly that this long period in which she was not getting enough oxygen when she slept impaired her performance at work and elsewhere. She regularly came in to the office late—very late. She was late for appointments. She missed appointments all together. The books were never closed on time. She repeatedly put off providing the accountant with tax information, even though the company’s operation was not a bit complicated. There were many other issues, but the worst thing, from my perspective, was that she made employees call the people with whom she had appointments in order to make excuses for her.

To the best of my knowledge none of the people whom I listed relapsed even once.

In 1987 or 1988 Sue gave up smoking. At almost exactly the same time, Denise did, too. So did Patti Corcoran, Sue’s best friend, and, halfway across the country, my dad. This was like a dream come true for me. I had never taken a puff, but for years I had worked in smoky offices and had taken Excedrin for headaches. When TSI’s office was declared smoke-free, my headaches went away forthwith, and they never returned.

Sue, in contrast, had a very difficult time quitting. She put on quite a bit of weight, which amplified the sleep apnea problem. She was also more irritable at work and at home.

I must mention one other factor: Sue never throws anything away. Okay, if it has mold on it, or it is starting to stink, she will discard it. Otherwise she stuffs things for which she has no immediate use in bags or boxes.

When I first met Sue, she was renting one room in the basement of someone’s house. It was not cluttered at all. She seemed to have no possessions except a water bed, a record player, and a few albums. By the early nineties we had a house of our own with two rooms that had no assigned function, a garage, an attic, and a full basement. All of them soon became full of junk. Both of our cars had to park outside because the garage was wall-to-wall miscellany.

TSI’s headquarters in Enfield was nearly as bad. Sue’s very large office was the worst. Strewn about were boxes and paper sacks full of correspondence and memorabilia. Her desk was always completely covered, and post-it notes were everywhere.

In the rest of the office stood several file cabinets. Of course, every business must retain records, and one never knew when the company might get audited. It was also critically important to maintain good records about contacts with clients and prospects, and our business, in particular, needed up-to-date listings of programs, which we had by the thousands. So, we had a lot of important paperwork.

No more mainframe announcements, please.

However, in TSI’s office could be found many other things, which by any measure were totally useless. One day I undertook to throw away the announcements that we constantly received from IBM about its products. These documents formed a stack about four feet high. 90 percent of these missives were about mainframe products. There was absolutely no chance that we would ever work with any of these machines. Even the remaining ones (all of which I intended to keep) were seldom of any value because the information might have been contradicted by a subsequent notice.

Sue asked me what I was doing, and I told her. She immediately got very upset and even started to cry. She just could not stand for anyone to make the decision to discard anything that she considered hers. I realized at that moment this was a reflection of a very serious problem. I put all the notices back in the file cabinet.²

1994: It was the best of times; it was the worst of times.

The business was finally taking off. Our new salesman, Doug Pease, was demonstrating that he was ideal for the job. The nationwide retail recession had ended. The retail conglomerates with money (or credit) were gobbling up smaller chains, and in most cases this worked to our advantage. We were approaching a position in which we need not ever worry about competition. Most of us were working very hard, but we were getting new clients, and it was exciting.

The problem was Sue. She was hardly involved in any of this at all. Her behavior was becoming really unprofessional. Doug complained about her often. She kept hiring assistants, and they kept quitting. I could not find out where we stood financially because our books were so out of date.

On a couple of occasions I was stretched so thin that I asked Sue to take trips to clients for me. I did not think that technical expertise would be involved. I just needed someone to find out what the users needed. The first one was to Macy’s East in New York. Sue never told me what happened, but the people at Macy’s told me years later that they had made voodoo dolls representing her and stuck pins in them.

The other trip was to Foley’s in Houston. Sue flew all the way there and then realized that she had brought no cash. Her credit cards had all been canceled by the issuers. Fortunately, she had a checkbook, and Beverly Ingraham, the Advertising Director at Foley’s, cashed a check for her.

In May of 1994 Sue and I took a very important road trip to Pittsburgh. We met with Blattner/Brunner, an ad agency (described here), and Kaufmann’s, a chain of department stores (described here). Both of these sessions went quite well. When we returned to Enfield, I was required to spend a lot of time working on the proposal for Kaufmann’s. It was the most complicated and difficult one that I had ever done, and if I did not do a good job of analyzing and estimating the difficulty of each element, we could suffer for this for years.

So, I asked Sue to follow up on Blattner/Brunner while I was working on Kaufmann’s. Sue had been there for the session in Pittsburgh. There was no one else I could turn to. She completely fumbled the ball. I was quite angry, but I knew that it would do no good to nag her about it.

On the other hand, I appreciated the fact that she was the founder of the company. These opportunities never would have happened if she had not started the ball rolling back in Detroit.

The day finally came when I just could not take it any more. I told her to go home and not to come in to work any more. There was no argument and no tears. She told me that I was making a big mistake and just left.

No one else thought that it was a mistake.

Within a day or so I approached Sue with the following arrangement: TSI Tailored Systems Inc. would be registered as a Chapter C corporation.

I would be president and have 55 percent of the stock, and Sue would would be treasurer with 45 percent. We would hire a new accountant to handle the corporation, and the bookkeeper would report to me. It would be my responsibility to make sure that the books were closed on time, and the taxes were paid on time. I would also do our personal taxes. We would fund the corporation with the difference between our accounts receivable and our accounts payable. If it needed cash (as it did a few times), I would loan as much as necessary to the corporation at a reasonable interest rate.

Sue was not happy about it, but she agreed to this. She did not even argue about the salary amounts that I set.

Our new accountant’s name was Sal Rossitto². He guided us through the transition. He advised us to set up an Limited Liability Company³, but I insisted on a true corporate entity that issued stock to its owners.

Setting up the new corporation was fairly straightforward. We had to open a new bank account. I found it to be a fairly simple matter to close the books every month within a day or two of the end of the month. We also set up a 401K with matching funds, a profit-sharing plan, and a good health and disability insurance plan from Anthem. None of this was difficult.

I am not sure who took over handling of the payroll after Sue left. TSI eventually hired Paychex to do it. Denise collected the time cards from the employees and submitted the requisite forms to Paychex.

Our accountants loved our Nov. fiscal year. They could work on our taxes in a less busy season.

I made one very good decision. We set our fiscal year to run from December 1 through November 30. We paid bonuses and made contributions in November. This gave all the employees the entire month of December to spend or save for tax purposes.

Dissolving TSI was a much more complicated task. Sue and Sal met often over the course of several months to unravel issues in the partnership’s books. I remember, among other things, some kind of ugly situation with regard to sales tax in California regarding the way that the installation at Gottschalks occurred. At the end of this process Sal confided to me that he now understood why I wanted to set up a real corporation.

We also ordered new letterhead. Ken Owen worked with me on the logo. I eliminated the stripes and the lean of TSI. The color around the TSI was pure blue. The colors to the left of that block went from a very light blue gradually darker almost to pure blue. The effect worked better on the computer screen than it did when printed.

For me the most important thing was to reestablish blue as the company’s color. It started with a light blue as shown at the top of the page, but over the years it had somehow evolved into something that was more green than blue. I hated it.

The next few years were boom years for TSI. I worked my tail off, and my travel schedule was a killer. I didn’t care. We had finally turned the corner, and the future looked very bright.

Life at home, however, was very difficult. Sue was obviously unhappy. She probably thought that I intended to dump her. I still loved her; I just did not want to work with her any more. I was quite sure that the company would do better without her.

displayed no interest in finding a job. This surprised me. She had had quite a few jobs since I met her. She really liked a few of them. She could summon up a great deal of enthusiasm about new projects, and she loved meeting new people. I could think of several occupations that she would fit very well.

Instead, she leased some space in an old office building in a questionable part of downtown Springfield, MA. She then fixed it up and rented it out to dance teachers who needed a place to give lessons. I don’t know how much of our money she lost on this venture. I am not sure that she even kept records of it. She certainly didn’t ask my opinion about it.

On weekends we still drove to Wethersfield to visit our old friends, the Corcorans, regularly. That helped quite a bit.

At one point Sue awarded herself a vacation. She drove to New Orleans to see a guy that she knew from high school who was into social dancing. She stopped at some other places along the way. I never asked her about what happened on this trip. When she returned she did not offer any details.

Eventually things got a little better. After the trip to Hawaii (described here) in December 1995 the situation became more tolerable for both of. At least we had some money to spend and save for the first time ever in our relationship.

1. The course that I took as a freshman at U-M taught a programming language that was unknown outside of Ann Arbor. It was called MAD, which stood for Michigan Algorithm Decoder. We wrote our programs on 80-column punch cards.

2. Perhaps you are wondering why I gave in without an argument. It was because I recognized quite early in our relationship that Sue was expert at playing the “Why don’t you …? Yes, but …” game described by Eric Berne in his best-selling book Games People Play. A pretty good write-up of the “game” is posted here. This is also the reason that I did not press her about the sleep apnea.

2. Sal Rossitto died in 2002. His obituary is here.

3. The purpose of an LLC is to protect the “members” from being personally responsible for debts and obligations undertaken by the company, but it is not as completely separated as a true corporation.

1972-1974 Connecticut: Working at Hartford Life

Posted on January 13, 2021 by Mike

My short career at Hartford Life. Continue reading →

The Hartford has two adjacent buildings. I worked in the tower.

Needless to say, I spent the first half of my first working day at the Hartford¹ filling out forms. Then I was told that I would be working in the Group Department. My supervisor for the afternoon was the woman who kept score at basketball games. However, the next day I was reassigned to the twenty-first floor, the home of Life Actuarial, Individual Pensions, and Special Risk Underwriting.

The insurance world had advanced somewhat while I was in the Army. The hordes of clerks with huge Fridens on their desks were still there, but there were a few electronic calculators as well. In 1972 electronic calculators required electricity, and cost as much as the Fridens, about $1,000 each, roughly half the price of my car, Greenie! The companies took strong measures to keep them from being stolen.

My specific assignment was to assist Mike Winterfield,² an actuary who had joined the Hartford when his old company (in St. Louis, I think), which specialized in variable annuities³, was purchased by ITT and folded into the Hartford. He had a vacation planned, but first he needed to submit a business plan for the VA product to his bosses at the Hartford.

My recollection is that ITT required a return on investment (ROI) of 15 percent for all of its subsidiaries. So, the algorithms that Mike developed for the business plan fixed the return ratio at .15 and juggled the other assumptions to make it work. There were no spreadsheet programs available yet in 1972.⁴ To evaluate the assumptions Mike designed an accountant’s worksheet with about eleven columns and I don’t know how many rows, probably one for each year. I only helped worked on this for a few weeks; I don’t remember many details. He would change one or two assumptions, I make all through the calculations to determine how much capital infusion the infant product needed to reach the desired rate of return.

These columnar worksheets were ubiquitous at insurance companies in the early seventies.

Someone else would check my work and put little red dots beside each number that had been verified. This approach, of course, meant that all of the clerical work in the department had to be done twice, but when programs were written to replace the accountant’s sheets, nothing replaced the red dots.

Finally, after many iterations everything seemed to be in order. The plan was submitted, and Mike went on vacation. A day or two later I had the dubious distinction of being quizzed tin a three-way phone call about the plan. The interrogators were Don Sondergeld,⁵ Vice President and Actuary, and Bob Goode, a high muckety-muck of the Hartford. They asked me several questions that stumped me. I disclosed what I did know, which was limited to how each cell on the worksheet was calculated. I am pretty sure that this information was not what they were looking for. Before I hung up I had to wipe off about a quarter cup of perspiration from the receiver for the telephone, which I shared with Sue Comparetto,⁶ the only clerk with an electronic calculator.

I have always had an aversion to phones. My VA experiences did not help.

The only other duty that I remember from my time in the VA area was calculating proposals for salesmen. They told us how much money prospects wanted to invest and when they wanted to receive the annuities. I, of course, had no clue what the market would do in the interim. So, we made an assumption of a specific interest rate, perhaps 6 percent. The sales agents would occasionally call to complain that a competitor’s proposal assumed a higher return that yielded a higher annuity amount or lower premium. I tried to explain that any company could in theory use any interest rate, but the Hartford’s policy was to use the same rate on all proposals. The salesmen often heard the first part, but not the second. They would try to pressure me into giving them a better quote. I never gave in, but once again the phone got drenched with sweat.

Hartford's insurance companies had hundreds of these. — Hartford’s insurance companies had hundreds of these.

For some reason I kept getting moved around. My recollection is that in my first six weeks or so at the Hartford, I sat in five or six different grey steel desks. All of the actuarial students were rotated from one area to another, but nobody moved as much as I did.

My next stop was in the Individual Pensions Department. The name was a little misleading. The purchasers of the product were small businesses, not individuals. For some of the prospective customers it worked out better to buy individual policies for each eligible employee than one group policy.

My first assignment there was to take over maintenance of the program that was used to calculate proposals for the pension plans. An actuary named Fred Smith had been doing this, but the bosses had more important assignments for him. The Hartford offered several varieties of them from single-premium fixed interest to variable annuities. The proposal program could use any interest rate to generate the premiums for a fixed-benefit plan or the benefits for a fixed-cost plan.

This program ran on a computer that was located not at the Hartford but at Kaman Aerospace. The Hartford had its own computers, of course, but their use was jealously guarded by the Data Processing Department. Many actuaries were quite capable of doing the programming, but it was very difficult to get access to the mainframes. So, the Actuarial Department rented time from two different companies. Kaman had an HP 2000A; Tymshare had a DEC PDP-10. Kaman was cheaper, but Tymshare had more features. We used interpreted BASIC on both systems.

Each punched row on the tape contained one byte (character) of information. Fred Smith could read these tapes!

We gained access to the remote systems using a teletype machine connected to a phone line. Our data was stored on paper tape. I think that we stored the programs on the remote hard drive, but I might be mistaken. The teletype had the ability to take information from the keyboard and/or the tape reader. Output was printed on an 8.5″ continuous roll in 10-pitch Courier at a breathtaking ten characters per second. The unit also including a tape-punching device. Because it made a lot of noise when printing or punching, the teletype was isolated in a very small room, maybe 4′ x 8′.

Later a second unit was added. It did not have a tape device, but it printed at thirty characters per second using a round disk. It was not a “daisy wheel”; the disk was metallic and was perpendicular to the page. People came from every floor and from the Hartford’s other building just to watch that baby hum. To many it seemed magical that it could print so quickly and accurately.

The pension proposal program was in good shape when Fred turned it over to me. There was an up-to-date listing, and Fred had inserted a lot of comments. I had to learn BASIC, but a thin handbook of the commands was available. BASIC was similar enough to MAD, the only language that I had ever coded in, that I mastered it fairly quickly.

The first thing that I did was stupid. I removed all the comments to make the program more efficient. In my defense:

I documented what the program did did on a separate document referencing the line numbers.
Programs in those days were so slow that removing the comments actually did make it slightly more efficient.
My one and only programming class was six years earlier, and Господин Muchnik never taught us about documentation.
I had no training for this job.
Shut up; I already admitted that it was stupid.

Patti Lewonczyk⁷ was in charge of producing the proposals. A girl named Paula, who was, I think, fresh out of high school, did most of the data entry. One time something went wrong. I don’t remember what the problem was, but I fixed it easily. However, I had to ask Paula to run the program again for one plan—perhaps a fifteen-minute job. She got very upset and started crying. Evidently she thought that I was yelling at her. I told her that she did nothing wrong, but she was still very distraught.

My second big project in Individual Pensions involved the reports that were sent annually to the companies that had purchased one of these products. A company in New York had developed a software program to produce these reports. A group of clerks led by Carolyn DesRochers filled out coding sheets with the current census information for each plan a month or so before the anniversary. This information was supplied to the Hartford by the customer.

How, you may ask, did we get the coding sheets to the software company? Someone from the Hartford transported a box of them every day to the bus depot downtown. A Greyhound bus then brought them to New York where someone from the vendor picked them up. The vendor’s software program then processed the data and created the report on attractive formatted paper and sent it to Hartford in a box on another bus. Someone from the mailroom picked it up at the bus station and delivered it to Carolyn.

Dave McDonald’s⁸ title was only Secretary, but that indicated that he was the top man in the department. He summoned me to his office one day to tell me that for some reason the program to produce the annual reports was no longer producing accurate information. The clerks had been observed whiting out the numbers printed on the reports by the computer and inserting corrected numbers using typewriters. This required a great deal of time and produced ugly reports that reflected badly on the company and especially the department.

I Iinterviewed Carolyn and a clerk or two about it. I also talked with someone knowledgeable at the vendor. It turned out that the vendor only allowed for a few choices of interest rate when setting up a policy. This caused no problems for a while, but by 1972 interest rates had started to rise, and the Hartford was quoting (and selling) plans with higher interest rates than the program could handle. The proposal program that Patti’s group ran calculated the interest using standard compound interest formulas. The report program, however, did not calculate interest; it looked it up on tables that someone had populated when the system was written. Evidently the designer of the program did not know the actuarial formulas, and no one had foreseen the significant increases in interest rates. I asked my contact at the vendor if their programmers could fix this. He was not sure that anyone would know how to address it.

I reported back to Dave McDonald all that I had discovered. He was gobsmacked. I never found out what was done about it, and I did not ask who had approved the purchase of a service with such a severe and obvious flaw in it. I designed thousands of programs over the subsequent five decades. I took some wrong turns, but I never made this big of a mistake.

The other task that I remember concerned lapse rates. In determining a reasonable premium for any life insurance product—all the pension plans had an insurance component—it is important to account for the possibility that whoever pays the premium might allow the policy to lapse. In general, there are quite a few lapses in the first year of life insurance policies and much lower rates later. On these policies there were inordinately high lapse rates for the first five or six years. I was asked to examine a sampling of the policies to determine what the causes were.

I discovered that most of the alleged lapses were not caused by the customers’ failure to pay the premiums. Instead, the sales agents were arranging “attained-age conversions”. The agent went to an existing customer and told them that the Hartford was now offering a new product that would be a better deal for them. He would request a quote from Patti for converting the plan. This was done by lapsing the policies on the old plan and issuing new policies based upon the current age of the policyholders. The agent was right. The customer who converted received either lower premiums or higher benefits.

Everyone (except the Hartford) profited from attained-age conversions.

It was a win-win-win-lose situation. The customer and its employees got a better plan, and the agent got a first-year commission on the same plan. Since the first-year commission on many life insurance policies is more than 100% of the first annual premium, this was a real windfall for the agents. Moreover, since the Hartford had four or five individual pension products, an agent could pull the same trick more than once. Quite a few agents had been doing this for years. The Individual Pension products were yielding good sales but no profits.

I know that there were fairly high level meetings that involved Dave, Paul Gewirtz, the actuary in our department, and the Sales Department. The sale people were adamant that this technique should not be taken away from them. Evidently some of the agents who used it were very influential. The problem was not resolved by the time that I left, but in the end my understanding is that the Hartford (and most other companies) stopped selling these products.

The A model of the HP 2000 could support 16 users, the F could manage 32. Rumors claimed that some employees on the twenty-first floor had found the built-in game programs, including a nifty one in which two people could call plays for opposing teams in a football game.

At some point someone put me in charge of the computer room and the computers. This did not involve much responsibility, but no one was suspicious if I spent time in there during slack periods. I wrote a program to produce sheets for a football pool. Each week I selected twenty competitive college games and five programs. At the bottom of the page I listed my Bottom Ten college teams. I think that I charged $1 per entry, and the person with the highest score won the whole pot, which was between $20 and $30. I might have gotten in trouble for this, but the spirit of “Don’t Ask, Don’t Tell” prevailed. I won the jackpot once.

On paydays there was also a pool that was run by Wayne Foster. At first employees bought tickets, and the person whose ticket was drawn won the prize. I bought a ticket every week, but I never won. Sue Comparetto won the pool once, and she tried to use the money to buy a round of drinks at the Shoreham Hotel, the site of the customary Friday night get-together of employees from the twenty-first floor. Since she was almost certainly the poorest person in attendance, the offer was rejected and, in fact, roundly disparaged as contrary to the spirit of the payday pool.

After Connecticut instituted a lottery, the rules were changed so that half of the pot was used to buy lottery tickets. The lucky winner got both the remaining cash and the tickets. I never participated in this, and it surprised me that any actuary would have anything to do with lottery tickets, which have a worse payoff than the numbers racket.

At some point Don Sondergeld assigned a special project to Tom Corcoran and me. The company that marketed the software for the APL programming language wanted to sell the Hartford the rights to use the language on the mainframe. The plan was to install in the handful of departments in which actuaries worked expensive APL terminals that were connected to the mainframe. A certain portion of the computer’s disk space would be allocated to the APL users. Something may have been done about sharing memory, too.

The advantage was that the actuaries would gain access to the computing power of the mainframe without being subject to the rules of the Data Processing Department. In those days tying to get new software projects approved and implemented by the DP people was an incredibly time-consuming task. The standard joke was that if you wanted to know how long a project would take, double the estimated amount and use the next higher unit. So, one hour meant two days; six months meant twelve years, etc.

APL programs require far less (in terms of number of characters) of computer code to make the same calculations. However, the APL representatives were unable to give us an example of a project that was relevant to Life Actuarial or Individual Pensions that APL could handle significantly better than Fortran or BASIC. It would be totally inappropriate for the two big projects that I had worked on. Both required extensive use of string variables. In APL a string of characters was treated as an array rather than a different type of simple variable. To me this seemed like a fatal flaw. Tom was more enthusiastic than I was. He liked the simplicity, and he had heard good things about it elsewhere.

I am pretty sure that the Hartford never purchased the product. This project was probably doomed from the start; my specialty was debating on the negative. Selling me on anything is very difficult.

The bills from Kaman and Tymshare produced a strong reaction. The Hartford bought its own HP 3000 computer and charged the Operations Research Department, not the Data Processing Department, with its management. A guy named Gerry Schwartz was put in charge of the new machine. Its operating system supported both BASIC and Fortran. The connections were over telephone lines using teletypes. Users in several departments brought over some simple BASIC programs that they had been using on Kaman’s HP 2000, and things went quite well for a while.

After my stint in Individual Pensions I was rotated to work for Jan Pollnow. My primary responsibility was to produce a monthly report, which my predecessor Chris DesRochers⁹ (husband of Carolyn), had been generating using a set of columnar accounting sheets.

I decided to write a BASIC program on the HP 3000 to replicate Chris’s sheets. As I added more code it got slower and slower. Eventually it brought the system to its knees. I had to set it off to run when I left at 5:00 and hope that it finished before I arrived fifteen hours later.

Gerry Schwartz blamed this on BASIC. He said that I should convert this new program to Fortran. Since I had never used Fortran, it took a while to do this. The conversion helped a little, but it still was horribly slow. Then Gerry suggested that I use “the segmenter” on the Fortran code. This involved breaking the program into pieces. This was a new concept for me, but at least there was printed documentation. I don’t remember any details of this implementation, but segmenting did improve the performance to a level of barely tolerable.

I spent a great deal of time getting this to work, but it made the monthly task much easier. The program ran without problems for a few months, but then one month’s data produced erroneous results. The problem was easy to fix, but it was a black eye for the department, and Jan called me on the carpet for it. As usual, no one checked my work.¹⁰

A few months before my departure Jim Cochran was brought in to our area to work with me. I made sure that he understood both the technical and operational aspects of the program before i started my next adventure.

* * *

I retain a few other vivid memories of my days working at the Hartford. People with a “coffee cart” came around every day at about 10 a.m. to sell coffee, donuts, and other pastries. This prompted a coffee break that became longer and longer over the weeks. Eventually the bosses warned everyone not to turn it into a floor party

I cannot remember the exact time that the buzzer rang to indicate that it was closing time. I wager that most of the clerical employees can. A large portion of the clerks, almost all of whom were female, would have all of their materials put away and their belongings gathered well in advance. The mad dash to the elevators when the buzzer sounded was somewhat comical.

Most of the clerks in Life Actuarial were supervised by Bob Riley. Almost all of them had Fridens on their desks, and they used them all day long. The machines were not 100 percent reliable. Every once in a while one would go berserk while performing division and start making awful noises. I am not sure why the clerks themselves were not allowed to unplug the calculators, but Bob always rushed over to take care of it. They generally seemed to work OK after they were plugged back in.

There must be a mountain of those Fridens somewhere. I wonder where it is.

Making copies of documents in those days was a process. At the time photocopiers were both rare and expensive. To my knowledge there was only one in the entire tower. It was manufactured by Xerox, and everyone called it the XM¹¹ machine. Someone would carry the document from the department to the floor on which the machine resided. The courier would stand in line near the operator of the machine. Sheaves of documents were presented one at a time to the operator, who passed judgment on whether they were worthy of duplication. For those that qualified copies were made, and both sets were given to the courier.

It might have been possible to use interoffice mail to do this, but there was no telling when (or even if) the documents would be returned.

I searched the Internet for pictures of photocopiers from this era, but I found nothing that approached the bulk of the one used at the Hartford (as I remember it).

1. I did not realize this at the time, but the Hartford was owned by ITT, which was then the prototypical conglomerate. Under Harold Geneen the company acquired all kinds of businesses. When Geneen left, the ITT spun off the Hartford.

2. In 2023 Mike Winterfield was still an active member of the Hartford Bridge Club. I played with him at a tournament once, as described here.

3. The purchaser of an annuity agrees to pay an amount—all at once or in installments. The party selling the annuity adds interest and deducts expenses and profit before paying it back in installments when the purchaser reaches a certain age. In variable annuities the interest rate is recalculated each year based upon the performance of the stock market or another index. Insurance components complicate the calculations. In the early seventies few actuaries were familiar with this product.

4. VisiCalc was released in 1979, but it only ran on an Apple II. Lotus 123 did not appear until four years later! Excel was introduced in 1987.

5. In 2020 Don Sondergeld is retired, but he is an active member of the Hartford Bridge Club.

6. There is much more information about Sue here and in subsequent entries.

7. Patti and Tom Corcoran married while I was coaching debate in Michigan in the late seventies. They had two children, Brian and Casey. In 2021 they both live in Burlington, VT, with their respective families. Patti died in 2011. My tribute to her can be read here.

8. Dave McDonald retired from the Hartford in 1995 as a Senior Vice President. He died in 2011. His obituary is here.

9. Chris died in 2013. His obituary can be read here.

10. These four words will probably be on my tombstone: Nobody checked his work.

11. XM stood for Xerox machine, but everyone still added another “machine” on the end when talking about it.

1967-1969 Part 1: U-M Classes

Posted on September 23, 2020 by Mike

Classes in the middle years. Continue reading →

To the best of my recollection this was my undergraduate academic schedule at Michigan:

Freshman	Fall 1966	Math 195	Russian 101	Chem 103	Latin 350
	Spring 1967	Math 196	Russian 102	Chem 106	Greek 101
Sophomore	Fall 1967	Math 295	Comp Prog	Econ 101	Greek 102
	Spring 1968	Math 296	Comm Sci	Sp Persuasion	Greek 201
Junior	Fall 1968	Math 395	Econ Intl	Sp Debate	Greek 202
	Spring 1969	Math 396	Topology	Homer	Thucydides
Senior	Fall 1969	Probability	Statistics	Psychology	Euripides
	Spring 1970	Sp Study	Linguistics	Russ Lit	Anthropology

The classes for the first semester of freshman year (fall 1966) are discussed here.

The classes for the second semester of freshman year (spring 1967) are discussed here.

The classes for the second semester of senior year (spring 1970) are discussed here.

Here is what I remember from the middle years by subject:

Math: The three-year basic sequence (195-6, 295-6, 395-6) started with a review of calculus. The last semester covered, I think, the calculus of several—more than three—variables. It was during this last class that I realized that I had absolutely no interest in being a mathematician. I could no longer visualize what was being talked about. I am sure that it is valuable for things like string theory, but it was not for me.

In fact, I was lucky to last that long. These six classes were all taught by professors, as opposed to graduate students.¹ We were expected (but not required) to do the problems at the end of each chapter of the text. I looked at them, but I never did them. Our tests asked for proofs rather than solutions to problems. The material became more and more difficult as the course numbers got higher. I familiarized myself with the concepts, but I did not actually learn them. That was not my goal; my goal was to pass the tests.

I developed several techniques for pretending to prove that formula A is equivalent to formula B, which is the format for most, but not all, mathematical proofs. It would be best to come up with an actual proof, but this is not like horseshoes or hand grenades. Coming close can count for a lot. I thrived on partial credit.

The first technique is known as proof by induction, a legitimate and powerful method absolutely essential for proving some mathematical concepts. It even has its own Wikipedia page. Any time than something countable—an integer or even a rational number—was mentioned in either formula A or formula B, I would used proof by induction. It was always good for a half-page of paper, and I never got less that seven out of ten points. Occasionally, it actually produced the required proof.

The technique that I invented and perfected was the Wavada Squeeze. You will not find it on Wikipedia yet. Here is how you do it: Start with formula A, and in a column beneath it make as many manipulations of it as you can think of. For example, you could multiply and divide by something: A=((x+3) x A)/(x+3). After ten or so manipulations, you end up with something quite different from what you started with. On a separate sheet of paper perform a different set of manipulations on formula B. Once in a while one of the two will actually show the way to a proof, and that’s great. However, even if neither does, you can maximize your partial credit by writing the the manipulations of B in reverse order beneath the manipulations of A. This will give you a “proof” that A=B with twenty or so legitimate manipulations, and one horrible leap somewhere in the middle. This was usually good for at least five out of ten, which was much better than nothing.

I made the mistake of taking one graduate-level class, Topology. I ended up with a C, which is a terrible grade in a graduate class. I remember entering the classroom on one Monday after a debate tournament. Everyone was busy writing their names on “Blue Books”, which meant that an exam was about to take place. This was news to me. Evidently, it was announced in a class that I had not attended. Usually I tried to make sure that I had a friend in classes to warn me about such events. This class, however, was composed of graduate students in math. None seemed approachable. I got one right out of nine. Showing up clueless in a classroom full of people with Blue Books appeared frequently in nightmares over the decades.

I am embarrassed to say that not only have I forgotten the names of these teachers, but I also have only vague recollections of any of the concepts, never mind the details, of what I supposedly learned. I never considered graduate school in math.

Actuarial Science: I passed part 1 of the actuarial exams in May of 1969. I took two classes from Cecil Nesbitt², a very famous actuary, in the fall semester of 1969 with the half-hearted purpose of passing part 2—probability and statistics. I was not a bit impressed with the quality of the other students, when compared with the guys (and girl) in my math sequence.

My attendance in both classes was extremely spotty. I did not do any of the assigned problems. Nevertheless, I thought that I had passed part 2 in November, but i was wrong. The exam was offered again in May, but I was not upset enough about this failure to study more in the spring session. The result is described here.

Cecil Nesbitt’s Wikipedia page is here.

Greek: For the first two Greek³ classes the teacher was H.D. Cameron.⁴ I can clearly visualize the professor who taught the 200-level classes as well, but I am embarrassed to report that I cannot remember his name. He also taught the Thucydides class.

My clearest recollection of those last two classes in the sequence involved the final exam. With less than one hour remaining before the Greek final, I realized that I had misread the exam schedule and studied for a different subject. The exam for that subject was a day of two later. So, I ended up taking the Greek test with practically no cramming.

Thank goodness it a Greek course was the one to which I gave insufficient attention; I got an A anyway. If I had neglected the other subject (I forget what it was), I would have been in a world of hurt. This kind of thing was also a subject of many future nightmares.

In all honesty the university should not have allowed me to take these four classes. My high school Greek class was good enough that I could almost certainly have placed out of Greek. However, there was no placement test for Greek at U-M. I am thankful for the lack; I received four easy A’s.

The Homer and Thucydides classes were more difficult. All of the other students in both classes were graduate students who took nothing but Greek classes and spent all of their daylight hours in the classics department. I was lucky to survive.

I enrolled in a course that translated the plays of Euripides. I had to get special permission from the professor to join. However, after a week or two I dropped the class. It was just too demanding for my schedule.

If I had applied to graduate school in 1970, I would have tried for a masters in the classics. These were the only classes that I enjoyed the most.

Speech: I took only one real speech class, persuasion. It was taught by Bruce Gronbeck.⁵ The other two speech classes were gift A’s to compensate for the time that my participation in debate consumed.

I went to the persuasion class every time that I was in town. I gave all the speeches and turned in the assigned paper. I found the material presented boring, but the speeches were quite interesting.

I have several vivid recollections. We talked about some of the speeches as a class. In one situation I remarked that I thought that the speaker had seemed sincere. Prof. Gronbeck asked me why I thought that, and I was stumped. Nobody else really had a better answer. As Jean Giraudoux famously noted, “The secret of success is sincerity. Once you fake that you’ve got it made.”

Jim Pitts, the captain of the basketball team, was on the roster for the class, but he never attended, not even for the assigned speeches. Prof. Gronbeck told Dr. Colburn about this, and a few days later Jim appeared, and he was prepared to give a speech. No speeches were scheduled for that day, but he was allowed to deliver his speech, which was about abortion.

Jim took the floor with about twenty 3″x5″ cards held in very large hands. The use of a deck of cards in a speech is always a bad idea unless you plan to demonstrate how to throw them. Jim’s first line was “The chances of getting an abortion in this country are 1. Slim and 2. None.” On the fourth card he could not read the handwriting of whoever wrote the speech. A few cards later the cards were in the wrong order, which always seems to happen. When he finished the speech, Prof. Gronbeck thanked him, and no one else said anything. The class then continued as usual.

That was Jim’s last speech, but he never missed any basketball games that year.

Prof. Gronbeck wanted to hold a one-on-one debate, and I volunteered. My opponent, who was from Germany, chose as the topic the reunification of her country. She was against it. It wasn’t much of a debate. I had to go first, and we only got one speech each. She made no attempt to refute anything that I had said.

Communication Science: For some reason in those days the academic department at U-M that dealt with computers was called “Communication Science”. I took two courses. The first was the introductory programming class. My instructor was Господин Muchnik, whom I already knew as a student in my first-semester Russian class. We learned to program in MAD, which stood for Michigan Algorithm Decoder, a programming language used nowhere else in the world.

There were no terminals. We had to record our programs and our data on eighty-column IBM cards. This required use of a machine to punch the cards. Each card corresponded to a line of code or a row of data. There was no backspace or delete key. If you made a mistake when you punched the, you had to eject the card and throw it away. The university had a dozen or more of these machines for students, and they were located on the other side of campus. If they were all in use, you just had to wait or come back later.

When you finally had your program and data punched, you inserted the deck of cards into the pre-compiler, a small machine that found syntax errors in MAD programs. When you fixed all of the errors that it found, you were ready put a rubber band around your deck of cards to submit a job to the computer. The operator would accept the job, assign it a number, and give you a card with your job’s number and a phone number for a recording that announced the number of the job that the computer was currently on. It might be hours or even days befor the announced number was greater than the number of your job. When it finally happened, you walked back to the computing center to retrieve your output and your deck.

The first few times that you did this, the output consisted of a list of errors. If there were only a few errors, you would wait for a keypunch machine to be available, fix the errors, and resubmit the deck. This process went on until you either despaired or were euphoric when you finally got some results.

For our final project we could do any program that we wanted. The only requirement was that it be at least one hundred lines of code. My project read in a bridge hand of thirteen cards and output the opening bid using the standard American system taught by Charles Goren. I did it over the Thanksgiving holiday when there were no lines for the keypunch machines, and shorter waits for the jobs. I got it to work well pretty quickly.

Needless to say, I really liked programming, and I was good at it.

The other Comm Sci class that I took was a strange amalgam of language theory and the history of computer languages back to the Turing machine. I did not get a lot out of it.

Economics: You cannot debate as much as I did without doing quite a bit of research into economics. By the time (sophomore year) that I enrolled for the introductory class at U-M I had already read hundreds of articles and several books on economics. The format for the class was lecture plus recitations. I enjoyed the lectures, although my memory of who gave them cannot be correct. I found the recitations tiresome and overly simplistic.

I studied for the final with my debate partner, Gary Black, a senior. I helped him more than he helped me. I was more than a little peeved when he got a low A, and I got a high B.

I took one more economics class. I did not like anything about the way the course in international economics was taught. I put in as little effort as possible. I think that I got a C.

My enduring impression about the field of economics was that it was 99 percent unproven theories. I was right. The next fifty years proved conclusively that most of what I was taught was just wrong. In fact, Daniel Kahneman, a psychologist, won a Nobel Prize in Economics in 2002 for demonstrating that the basic behavioral assumption underlying all economic analysis is nonsense.

Psychology: As a senior I signed up for the introduction to psychology in the honors program. I was allowed to take one class pass-fail, and this was the one that I selected. Grades lower than C were unheard of in honors classes. So, I correctly determined that I could pass this class with virtually no effort.

The teacher was a female graduate student, and she insisted that we understand the principles of feminism. Most of us had never heard of the word. I remember very little else about the class, except that it was full of freshmen with first names like Hill or Twink. Twink told us that in the home in which he grew up no one ever questioned Freud’s teachings. I found that astounding, but I suppose that it is no more astounding than believing in the Book of Mormon or, for that matter, the Bible.

Twink was a fairly big guy, maybe 6’3″. Once I was walking somewhere with my friend, Tom Rigles, who had already heard me tell stories about him. Tom said that he did not imagine him being so big. I casually remarked that I could take him. Rigles scoffed at me. As God is my witness, I could take Twink then, and if he is still alive, I could take him now.

I don’t think that the psych class had a final exam, but we were required to write a paper. I had picked up a book called Games People Play by Eric Berne. The night before the papers were due, I skimmed the book and wrote and typed the paper. It was the third and last paper of my entire undergraduate career

I disliked all the “social science” courses that I took. If you had predicted in 1970 that I would voluntarily take almost nothing but social science courses for six years in graduate school, I would have told you that you did not know what you were talking about.

1. Ted Kaczynski, the notorious Unabomber, was a graduate assistant in the math department at U-M during the sixties. The offer of the teaching position was the main reason that Ted chose Michigan over Cal Berkeley and the University of Chicago.

2. My best friend since 1972, Tom Corcoran, worked as an actuary. One year I was shopping for a birthday card for him. I looked through about twenty humorous cards until I found one that actually mentioned Cecil Nesbitt! I was absolutely astounded.

3. I learned classical Greek in high school and college. Decades later I tried to teach myself modern Greek in preparation for a Hellenic vacation. The alphabet was the same, and the grammar was similar. However, the vocabulary and pronunciation were drastically different.

4. Professor Cameron in 2020 is emeritus in the classics department and curator of the university’s Museum of Zoology.

5. Professor Gronbeck died in 2014. A very lengthy obituary can be read here.

Wavablog

Memoirs, bridge, popes, travel, and other stuff

Tag Archives: MAD

1994 TSI: The Second Crisis

1972-1974 Connecticut: Working at Hartford Life

1967-1969 Part 1: U-M Classes