Helping the U.S. Federal Aviation Administration --
Ha Ha Ha

Computer History Vignettes

By Bob Bemer

I have mentioned elsewhere my acquaintance with Ernest Baynard, who was at the time of this story (circa 1970) Chief of Staff for Congressman Jack Brooks (TX). As Chairman of the Government Operations Subcommittee, Brooks had oversight for many departments. The Post Office was one. The Federal Aviation Administration was another. There lies the story. One of ineptness and insularity.

I had gotten interested in the Air Traffic Control System because my secretary, Pat Opre, was a great private pilot. The FAA was in as much trouble those thirty years ago as it is now.

Baynard and I had much respect for each other. He discussed the FAA problems of the time with me, and introduced me to the transponder, a basic element of the system. It was a little gizmo in an airplane, with 4 dials each settable from 0 to 7. Someone in the cockpit set it to the value most recently radioed by the ground controller, who said something like "squawk 2455". That was how the airplane got its identity for the ground control system -- both local and national. They called it the "4096 transponder", 4096 being the decimal equivalent of octal 7777, the maximum number one could set.

A major problem was that there were more than 4096 airplanes existing in the U.S. airspace, so the numbers had to be continuously changed to fit within the local area. A permanent number, akin to the tail number, was not possible.

I'm not going to go into the full spectrum of problems here. Just two. Perhaps later I'll flesh out the remainder.

The Transponder Value Limitation

The FAA was going to a new system. $18 thousand millions worth at that time. With inflation and the usual underguessing that's probably near $60 thousand million by now.

I asked if the "ping - ping - ping" of the existing radar scan was reliable. Very -- they said. So I asked why not make it go "ping - pong - ping - pong"? I found in San Francisco a $25 switch that could be attached in each plane to make it do this. Going immediately from a capacity for 4096 airplanes to a capacity for over 16 million. The tail number could be used, and pilots would never have to fiddle with the transponder. Except maybe to switch to a number that said they were being hijacked, of which an awful lot was going on at that time. Permanence also gave less room for error.

It appeared that my solution was too inexpensive. It might cut down the money budgeted for the FAA.

Software Efficiency

Somehow I got invited to a meeting of the National Academy of Sciences, where help was asked in making the existing hardware cope better while the FAA figured out what new system to devise and buy to really fix it. I recall that (then) Commander Jan Prokop was an attendee who knew what he was doing.

One question I asked was about computational method for trig functions. When you're vectoring in on a moving object in space you're going to compute an awful lot of sine and cosine values. I asked whether they had compared the times for doing this by (a) table lookup or (b) polynomial approximation. Which method was faster? They didn't know. "Well why don't you find out?", I asked. They replied "How?" "Simple. Do it by both methods and compare the elapsed times". Again, "How?" "Take internal clock times at start and finish, for both methods".

It turned out that not even ONE of their systems had an internal clock. IBM was charging a flat $100 for one, and they had declined to buy it. For want of a $100 clock, citizens, would you want your airplane to crash? Of course I thought of the horseshoe nail and the lost battle.

Back to History Index            Back to Home Page