Transient Enthusiasms

Category: Work

I’ve had an interesting work life; I’ve done enough different things that I told my students that I follow the Brownian Motion Theory of Career Development.

  • Cumberford Martinique

    Cumberford Martinique

    The Cumberford Martinique came from the fertile minds of the Cumberford brothers; Robert was the designer and James handled the business side.

    I became involved through the friend-of-a-friend-of-a-friend. My friend had worked in England when currency export was blocked, so he used his earnings to buy an old Bentley; I volunteered to change its water pump. This was while I was maintaining and developing the software that HBS used to teach financial modeling. “He likes cars and does financial modeling” traveled through the links, and I was asked to visit their company and perhaps do some work for them.

    Of course I dressed up; wearing my best three-piece suit I knocked on the people door next to an overhead door. After a couple of minutes the overhead door rose and a bearded man in dirty jeans leaned under it and asked what he could do for me. I said “I’m looking for James Cumberford”; he said, “That’s me”. I ducked under and was shown around the facility and introduced to some of the people working there.

    The People

    James had been in sales most of his life. He had sold diesel motors for boats, and later owned a yacht delivery service. He had also been a Sloan Fellow at MIT, which is where the friend-of chain started.

    Robert is James’ older brother. He had been a car designer and journalist. His first design job was at General Motors, where he had been Principal Stylist for the 1958 Corvette. He had subsequently designed cars for many other companies, and was widely know and respected in the industry.

    Marilyn was in charge of dealing with the government – navigating the many regulations that affect all car companies.

    Chappy helped Robert produce the drawings for the car. He was the oldest of the group and had started and run a drafting school in Detroit. He once questioned Robert’s statement that the intersection of two services was a straight line – so he spent two days developing the intersection on a series of drawings. (I’ve never seen so many lines on these drawings.) Eventually, he proved it to be a straight line.

    Peter did everything requiring hands and thinking that didn’t fall into a standard job. He was the one who figured out how to produce the wood-composite fenders that are so distinctive.

    The Car

    At this time old-looking cars were popular – the Clenet is an example – these were a modern version of coachwork on recently-purchased cars with the body removed. While visually striking, their additional weight hurt performance.

    Picture of the Cumberford Martinique at the Detroit Grand Prix before people arrived.
    In Detroit, at the Grand Prix

    The Cumberford brothers had a better idea: They would make a car that looked like a french car of the late 1930s with the best-available technology. Here are some of the highlights of that technology:

    • Engine and transmission from the BMW 733
    • All-hydraulic suspension from the Citroen CX
    • Cast aluminum chassis to the same specs as the F15 by the same caster
      • The pieces were held together by bolts that broke off when they were at the right torque so they couldn’t be over- or under-torqued, very trick!
    • Wheels by North American Rockwell
    • First use of TRX tires by Michelin

    Only Robert Cumberford could have made this possible; no one else in the world would have been permitted to purchase and use these parts. Even he could only do so because the car was to be much more expensive than anything made by BMW or Citroen. The planned retail price was $125,000; this was a lot of money around 1980 when a new Ferrari 308 sold for $45,000.

    The Plan

    It was never intended that there be a lot of these cars; the overall logic came from a statement from someone with decades of experience advertising cars: “You can sell two hundred per year of anything; there are always enough people who want to be the only one at the country club.” So we planned to sell the Martinique for a few years and then produce a car with more modern bodywork. This strategy was designed to avoid conflicting with the suppliers while making fine cars.

    Driveability

    Most of the driving feel of the car came from the chassis and suspension. The chassis was very stiff because of the size of the chassis components and the rigidity of cast aluminum. This is important because it allows the suspension to work as designed. The Citroen hydraulic suspension was made to very close tolerances: the suspension components used needle roller bearings. This, combined with 2.5 turns lock to lock steering left no play at all. So Citroen had designed means of reducing the steering preciseness to the level that the driver preferred.

    Even the springs were unique: They were spheres pressurized with nitrogen; it was that pressure that determined the suspension stiffness. But because they were spheres, they provide an additional adjustment that is very difficult with mechanical suspensions. Normal metal springs have their stiffness increased the same proportion for each spring movement. Translating: if the spring stiffness increases 50% with a displacement from 0 to 1 inch, then 4 to 5 inches will also increase stiffness by 50%. The Citroen suspension wasn’t limited by this: The pressure determined the stiffness of the first inch’s compression, but the diameter of the sphere determined the stiffness of the 4 to 5 inch compression. This allowed a very comfortable ride on smooth roads while stiffening over large bumps. Other cars try to achieve the same effect with rubber bump stops, but it’s not the same.

    Driving The Car

    When I drove the car the first thing I tried out was the self-centering steering. Regardless of what the car was doing, if I let go of the steering wheel it returned to center with the wheels straight ahead– but without tugging on my hands to do so. Then I tried some corners, both bumpy and smooth. It stayed flat and controllable, regardless of how hard I pushed it. The chassis felt like a race car with all body seams welded and a roll cage– because of the stiffness. Just a great car to drive. I also received the first speeding ticket in a Martinique, 80 in a 65 – although I was actually going faster – the officer was kind, and probably intrigued by the car.

    Robert in dark hat, me in light hat, at Detroit Grand Prix to show off the Cumberford Martinique

    What Happened?

    The company had orders (and deposits) for more than one hundred cars. But additional funds were needed to fill the orders. We wrote a prospectus and began the money-raising process. Then John DeLorean was arrested on drug-trafficking (he was later found innocent); all the prospectuses were returned to us within a week. We had thought that DeLorean would go under because, in their first prospectus they had a graph of how many cars that would sell at various prices to justify their forecast unit volume. Later versions of their prospectus kept that volume with much higher prices, and the graph was no longer included. So we were hoping that we could be funded before it happened.

    The Cumberfords tried to continue for a few years, but were never funded. All deposits were returned and all bills that could be paid were paid. Both Robert and James went on to other things.

    April 8, 2020

  • Diagrammatic Guide Signs

    Diagrammatic Guide Signs

    Research on diagrammatic guide signs was part of my first job after college at a consulting firm named Serendipity, Inc. My first task was the same for every project: we called it “getting smart”. It consisted of two steps: (1) Read everything we had that might relate to the project objective; and, (2) write down any thoughts that came while we were reading. The company had learned that those first thoughts almost always contained elements of a solution. We came out of this first task with lists of questions and ideas.

    Serendipity, Inc.

    Our objectives were simple:

    1. Find which diagrammatic sign designs worked best on Interstate highways
    2. Determine if they would work and where they would work best. Earlier studies showed that most accidents (and most fatal accidents) were in the vicinity of interchanges, with the highest concentrations near major interchanges. So we focused on this.

    Our task was to find a way to test guide signs indoors. Our approach was:

    1. Tell the test subject(s) their destination
    2. Project a picture of a road, so they could see how many lanes there were
    3. Project a sign on a screen for a short time, typically a second – based on the thought that drivers were busy near interchanges
    4. Ask the test subject(s) to say what lane they should be in – based on the assumption that last second maneuvers could cause accidents and that early lane positioning was desirable

    We moved from asking (for a single subject) to some buttons to be pushed that we recorded, so we could test a group of subjects. This worked, so we received permission from the Smithsonian Institution to use their facilities and access their visitors as test subjects.

    Federal Highway Administration

    After the project was completed I left Serendipity. After I worked for a few months at another human factors firm, I was hired by the Federal Highway Administration – who was taking over the project and wanted someone who was involved in the earlier work. I was the only junior person, so I was hired.

    FHA’s job was to coordinate more comprehensive testing. Their first step was to validate the results. To do so they set up two instrumented cars. These had a small projection screen mounted next to the rear view mirror and a slide projector in the back seat. The testing was similar to the laboratory work:

    1. Tell the test subject their destination
    2. Project a picture of a sign at the appropriate time
    3. Take notes on the car’s movement and the driver’s heart rate
    Example of sign projected in the instrumented car, showing lanes with dotted lines.
    Example of sign projected in the instrumented car

    There were three groups of test subjects:

    1. Ordinary people, 25-60
    2. Retired ministers
    3. Airline pilots

    Everything went well, and there was only one incident: One of the instrumented cars was loaned to us by Ford; they had instrumented it and used a gold-plated steering wheel to collect heart rate and – for public relations purposes – had used a convertible. That car was following a truck when a chain broke and the truck’s load – pressurized acetylene tanks for welding – began to fall off the back, just in front of the instrumented car. The driver – fortunately an airline pilot – weaved among them and dodged any that would have landed on the car. His heart rate didn’t change while this was going on – only afterwards!

    FHWA Testing On The Road

    The instrumented car tests indicated that diagrammatic signs had real potential, so we began the final tests: We – the four people at FHA on this project – would set up two test signs and develop methods of data collection and analysis. Simultaneously, an RFP (request for proposal) was sent out to select a firm for a more comprehensive test on the I495 Beltway around Washington, DC.

    The two test signs were an issue: we couldn’t find a company that would install them on our schedule as all local companies were booked for months. So we decided to install them ourselves. First, we needed a “before” study. The only approach we could follow on short notice was super-8 video cameras, run at one frame per second. We planned to use four, so we bought six. These were set up on tripods to show the traffic upstream and downstream from each sign. My job was to place the cameras and their operators, and to check back hourly and provide food, drink, and relief. After a few days the first camera broke; I’d already found a local firm to repair it so I took it there and asked them what had happened. They told me that the camera wasn’t really designed to run for long periods at one frame per second; that each camera should be rebuilt every 100 hours to be reliable. Well, that’s why we were doing this; to find out what would happen. I looked at industrial grade cameras, these were much more expensive and couldn’t be rebuilt locally. So we stayed with the consumer grade and planned to buy more.

    After finishing the “before” data collection we spent a week installing the new signs ourselves. We rented and put up scaffolding. The new signs were delivered to the sites, stacks of 4×8 plywood, each marked for where it should be installed. We replaced the old signs with the new signs. I’d never worked on scaffolding before, and didn’t like heights. But I got used to it. It took a full week in the hot sun, but we got it done. And we spent another two weeks collecting “after” data.

    During the data collection I felt like everything that happened on the few miles where we were working was my responsibility. I drove up and down it dozens of times a day – frequently stopping to deliver food, drink, film, cameras, and people. Any time a car stopped on either side of the highway, I stopped to help. By that time I knew every store and repair stop within miles. Once, it was a truck with two appaloosa in a trailer – beautiful horses. After I helped them they gave me their card and asked me to visit sometime. I’m still sorry I didn’t.

    Contractor Testing

    This was in the fall; the contract was expected to start early in the next year. They would collect “before” data, change about a dozen signs, and collect “after” data. The sign change was scheduled for the summer months to include as many tourists as possible. The winning company also did something that we couldn’t: they planned to use tape-switches on the highway to measure traffic by lane. They also planned to put some of them on the lines near the interchange that cars were not supposed to cross. All of these were to be connected with wires so that the data could be recorded. While this was much superior to our camera approach, they also planned to operate cameras – primarily for backup. Everything went as they planned and they submitted their final report on schedule.

    Results

    There were also studies done by eight states; in total, there were about twenty studies of diagrammatic guide signs. From these, there were three major findings1:

    1. More time is required to read, understand and react to diagrammatic signs in comparison with conventional signs with the same number of legends…
    2. Drivers have certain expectancies as they drive along a highway. Because exits are usually to the right … A left exit violates the expectancy of both exiting and through traffic. It is at such exits where unusual maneuvers are required that diagrammatic signs benefit driver
    3. Benefits from diagrammatic signing … In other words, because drivers decide earlier on the appropriate path to reach their destination and position themselves accordingly, there is less indecision at the gore proper [where the roads split].

    Several years after the final report, there was a study of accidents and fatalities at the changed interchanges. These found a 20% reduction in both at the recommended locations for diagrammatic signs: left exits from Interstate highways. Overall, this was an unusual research project because it encompassed multiple sign testing methods ranging from the laboratory, through instrumented cars, on-highway, and before/after accident and fatality records. While my part was small, it provided a standard that I have remembered.

    April 3, 2020

    1 Pages 4-6, Diagrammatic Guide Signs For Use On Controlled Access Highways, Report No. FHWA-RD-73-21, December, 1972.

  • My first Teaching Gig

    While I was working on my doctorate at Harvard Business School I took a one semester job teaching a graduate finance course one evening a week at Babson College, a business-only undergraduate and graduate college in Wellesley. This is the story of the first night of the course.

    The course began in January; I had prepared carefully – I thought – as I wanted to make a good first impression. (As Babson was unofficially associated with HBS, having many of its graduates as faculty, teaching ratings were important.) I had assigned a case in the class notes, but no one had even read it.

    Note: A case in this context is a description of a situation in an organization that requires a decision. Its length can range from a single page to thirty or more. A finance case includes financial data, which requires analysis – generally arithmetic and thinking. The case I had assigned was probably about a dozen pages. My plan was to use elements of the decision for very short lectures on finance.

    I tried to turn the session into a lecture, based on the case and my short lecture planning, asking the class to take ten minutes to read it. My problem was that I hadn’t considered that I might have to do this, so I did it poorly. I stumbled through it until the break, it was a three hour class with a fifteen minute break in the middle. The break finally arrived and I walked outside to think.

    “What to do” was my problem. I wanted to get in my car and drive home, but I knew I couldn’t do it. So I had to go back in. I thought about the material that the class would need to know to analyze the cases for the following week. Normally, the assumption for a case is that the student would do the needed reading in order to analyze the case. But I decided that a lecture on it was the only thing I could do that might regain some ground with the class. But I had to tell them what I thought was going on, so that they could understand why I was doing what I was.

    The first thing that I did was to apologize. I told them that this was my first course to teach, and that I hadn’t considered what I would do if they hadn’t gotten my request to prepare the case. I felt that this explained the situation without any hint of me blaming them, which I felt was important to get them working with me. Then I told them I was planning to lecture for the remainder of the evening on the material for the next week and asked their help in putting an outline on the board. They did help, expanding some of the points. And the lecture seemed to go well.

    At the end of the evening I described the decisions needed in the cases for the next week, and how the things I’d lectured on might relate to them. Finally, I told them I was looking forward to the next week and I hoped that they were too. I was very happy to find that most were there the following week.

    I learned a lot from this experience. I feel that I dealt well with a bad situation by: first, admitting that it existed; and, second, by getting others involved in fixing it. Then, I learned that I should avoid that particular situation by better planning for the first class: never again did I assume that material was read – far less, prepared – for a first class, not even by graduate students.