How I got into computing

With an undergraduate degree in Astronomy and Astrophysics, a lot of people ask how I got into computing. Now I’ve worked with a lot of “almost physicists”, so the pattern of failing to become a physicist and ending up in computing is actually pretty common. But the story is amusing enough to tell.

Back when I was young, and dinosaurs roamed the earth, computing was pretty new, and the industry was small. There were small number of mainframes, kept in glass rooms, guarded over by system programmers who knew obscure magical incantations. I had just about no interest in them. I wanted to be an astrophysicist and discover what made the universe work. Along the way, I discovered a few things:

Physicists were really smart. Not “I can do the New York Times crossword puzzle” smart, but genius smart. While they were not all quite that smart, the concept here is Dick Feynman smart. If you’re not familiar with Feynman, he was the sort of guy that people wrote books like Genius about.
Most physics undergrads, and a lot of physics grad students end up not doing physics. Instead they end up using what is a good broad based technical background to do other things. Things mere mortals can do.
Unlike physics, in computing, mere literacy often passes for genius. Just reading the manual page puts you ahead of the crowd.
There are a lot more jobs in computing than physics.

The discovery about computing began with an interesting event. At the time, I was a Junior at Penn. I’d taken one introductory computing class, writing simple programs in WATFIV and run in batch on cards. That was the extent of my computing background. I think at the time, my programming experience was limited to about 300-400 lines of WATFIV, typed on a IBM Model 29 keypunch.

At the time, I was taking Stellar Structure, and we had spent two or three classes working on a gray model of a stellar atmosphere. Our professor, Dr. Binnendijk, concluded with something like this

So, now we have four equations and four unknowns, so we can solve it. Unfortunately, it is transcendental, so we must approximate it. When I was a young boy in Holland, my professor and a staff of 12 would take many weeks to do so, using numerical methods. They tell me there is a machine downstairs that can do it in a few seconds. I would like it done for Tuesday.

At that point the ten or twelve of us were given accounts on the 4341 in the basement of DRL and that was that. We all spent the weekend learning CMS, XEDIT, and programing in IBM FORTRAN. At the same time, we went rummaging through the Math-Physics Library looking for algorithms to do Numeric Approximation of Differential Equations. If I recall correctly, I ended up using the Runge-Kutta method. Basically we were doing exactly the sort of thing that only thirty years before, the computer was invented to do, right across the street from the Moore School, where ENIAC was built.

Lots of time was spent in the basement of DRL that weekend. We all experienced a number of “hey look what I figured out” eurekas. Somehow we all managed to get things done. And the rest, as they say, is history.