As a teenager I was very interested in computer programming. I wrote mostly BASIC on a tiny Sharp pocket computer and assembler code for a simple Z80 development board. My mother had a friend whose husband worked at DEC (Digital Equipment Corporation) in the „Artificial Intelligence“ department. Through this connection, in summer 86 (I was soon going to be 17 years old) I got the opportunity to work a summer long as an intern at DEC Zurich.
In the 80’s AI meant things like LISP machines and expert systems. Therefore I was assigned the task to create a demonstration program for the new OPS5 expert system language. First I got a VT52 terminal and an account on an older VAX 11/780. But soon it became clear that I used too much of its processing time and memory and other users complained. So my mentor organized a brand new microVAX based VAXStation for me. At the time an insanely expensive machine with a huge b/w high resolution monitor just for myself!
After a few weeks I had a first prototype to show. I had built an expert system which was able to make a weather forecast based on weather observations (clouds, temperature, air pressure etc.). I called it VAXfrog (download: part 1 - description, part 2 - code) and my mentors had a close look at my application and the code. They were rather happy as the code seemed to do something useful. But one of them complained that my solution was „not symbolic enough“. This was actually true, as some of the input variables (like air pressure) needed to be continuous variables. And other inputs were converted to discrete (integer) variables. The forecast propagated forward through several fixed time intervals.
If I think back, my solution had indeed little in common with a typical expert system of the time. Today we would call something like this a „deep neural network with manually adjusted parameters“.
But my mentors suggested that I made a presentation in front of some managers and also technical staff nonetheless. In the following day I created an overhead projector presentation built on a system of translucent pieces of plastic foil which could be put on top of each other in various combinations to illustrate the inference process.
When I started my presentation in front of this highly authoritative audience a few days later, I first greeted everybody and looked around. There was only one woman in the group and I found her extremely attractive. Unfortunately I probably stared a fraction of a second too long at her. And my fate was sealed. When I looked in her direction again soon afterwards, I noticed that she smiled at me. She smiled in a way no woman had ever smiled at me before. This was most confusing and it became difficult for me to focus on my presentation. I started struggling with my overly complicated system of foils and I really did not manage to get my message accross to the audience. When I looked at the group I saw only faces of puzzled people trying really hard to understand what I wanted to explain. It was a complete disaster. And then I looked at her again. She still looked straight into my eyes and her smile was saying something like „so so, it’s me you want, cute young boy. I see it clearly. Do you really think you could get a woman like me?“. And my confusion got worse and my presentation too. I also started to feel anger rising inside me because I thought that she was intentionally ruining my presentation and maybe even my reputation as a developer (and she even seemed to be enjoying the whole thing!).
My mentor saved me after several painful minutes by suggesting that I should drop my foils and just answer questions. In this way I somehow managed to survive the remaining few minutes of my presentation until it finally ended.
At least I knew now what really matters in the life of a man. This fine woman taught me the difference between end goals (in this case a desired woman) and instrumental goals (like to be able to program a computer) in just a few minutes.
Follow me on Twitter to get informed about new content on this blog.
