Meet The Programmer Who Helped Put The First American In Space

Mercury Project rocket first American in space

May 5th was the 50th anniversary of the first U.S. manned space flight.

The astronaut was Alan Shepard, and it was a huge deal back in 1961 — the Russians had beaten the U.S. with the first satellite in 1957, and again with the first manned space flight a month before.

Computers played a huge role in that flight, and Art Cohen was the main mathematician behind them.

As the leader of IBM’s Space Computing centre in Washington, D.C., Cohen’s team was responsible for all computing support for every Project Mercury flight from 1961 through 1963.

Cohen is now 83 years old, and he retired from IBM in 1988. But he gave us an interview as part of the company’s 100th birthday celebration this year.

Things were very different back then, as he explains:

  • The two computers that directed the first space flights filled a large room, but only had 1/100,000 as much storage space as today’s least powerful iPhone. They also had the processing speed of 3Mhz — about the same as an early 1980s vintage PC.
  • All data about the space craft — position, velocity, and so on — was tracked by radar on the earth. The only data sent directly to earth was medical data from sensors connected to the astronaut’s body.
  • Data traveled from the radar stations to Cohen’s team at a blistering fast pace of 1,000 bits per second — about 1/5,000th as fast as today’s wireless connections.

Here’s a transcript of the interview, edited slightly for length.

Business Insider: What was the computer system like that you used in the Mercury Mission and how would you compare it with today’s PC or a smart phone in terms of power and storage? 

Art Cohen: We were at something called the Goddard Space Flight centre, which was in Beltsville, Maryland. That was the central communication hub for Mercury. And we supplied all the information to the Cape, to mission control and so on.

What we had there were two 7090s. They were second generation computers, transistorized. And at that time NASA was so concerned about transistorized – would it be reliable enough? – that they made us have a duplex system. They felt that tubes were more reliable than transistors because they weren’t sure about transistors.

We had two of these things right next to each other. It took up a whole room. And by room, I mean a large room — with tapes and DASD and storage units, and so and so forth — a very, very large set up.

We had 32,000 “words” of storage — 10-digit words of storage — that’s what we had.

[Ed: This is equivalent to about 150 kilobytes (kB) of storage on today’s computers, which store data in different-sized chunks. Most hard drives today are measured in gigabytes (GB), which are 1,000,000 kB each. Wikipedia articles on the IBM 7090 and 36-bit word length have more background.]

Now you’ve got things you could hold in your hand that are more powerful than what we had. With tremendously more storage. But compared to the world I lived in … in ’52, we worked on something called the CPC, the Card Program Calculator, and we had 34 — get this — 34 10-digit storage units. 34 units of 10 digits each. But we solved all kinds of problems with that. It was a challenge, but we did it.

So we thought this was gravy. I mean, we had 32,000 units of internal storage — that was pretty good.

We had pretty sophisticated programs for Mercury. All kinds of mathematics — orbit mathematics. Orbit mechanics. Differential correction, sliding wire techniques, short arc techniques, re-entry, knowing the positions of all the radar sites, acquisition. We told them where to point the radars and we told the ships where to go for landing.  So it was a lot of complicated stuff that we did we those 32,000 pieces of storage. It’s kind of remarkable when you think about it, isn’t it?

MR: What was the rate of processing?

AC: [After checking] The 50,000 transistors could read and write a the rate of 3 million bits per second.

[That’s about 3Mhz, which is about the same as the Tandy TRS-80 — one of the first personal computers introduced in the early 1980s. Today’s fastest processors are over 5Ghz, and benefit from multiple cores and other advances as well.]

It was slow. In those times, it was very fast, and that’s why we used the 7090.

At that time we were always worried — can you pack integrated circuits close enough, because the heat would build up so badly that there’d be a limitation on how much you could do, how fast you could go.  And we obviously broke that sound barrier, right?

BI: You had to get that data in there, process it in real time, and then send it back.

AC:  Yes, you’re absolutely right. IBM had built up in Kingston something called the 7281 which was called the real time channel.

By the way, the data that was coming in from down the Atlantic missile range — hold onto your hat — was 1,000 bits per second. The fastest data we had for Shepard’s flight was 1,000 bits per second coming in — that’s position data coming in and we also were receiving velocity data. 

[By way of comparison, Verizon’s 4G wireless network is at least 5,000 times faster.] 

And then later we improved it — it was a big deal to get it up to 2,400 bits per second. And the radar data was coming in much slower than that.

All that data was coming in fast enough so that the people down at the Cape were able to make decisions.  We drove all the displays at mission control and they were able to make decisions about retro fire, separation from the booster, and all this kind of stuff.  All that was done with the data that we provided to mission control.So it was good enough.

NASA Project Mercury Tracking System
A schematic of the Project Mercury Tracking System that Art Cohen worked on.

BI: How was the data getting from the rocket back to earth? AC: We had radars that were actually tracking that thing. They were watching it. And that data was then being sent over this 1,000-bit-per-second line.

MR: So there wasn’t any positioning data coming from the spacecraft to you? 

AC: No. The position data was coming from the radars which were watching the craft. 

MR: Was there any way for Alan Shepard to communicate?

AC: We had voice communication.

MR: How did that work? 

AC: I suppose radio. [In 1962, John] Glenn orbited the earth. Whenever he got near a radar site, there was an astronaut there with voice contact with the astronaut. And then, all of sudden he went over the horizon, and then they lost contact. So, there was a short period of time when there was contact.

It was voice communication to find out how things are, and also, they had telemetry data coming down which was like medical stuff. He was hooked up, and that came to us and we also sent that down to mission control.

They couldn’t bounce signals off of satellites at that time. It was all direct. What they could see is what they could get.

MR: When you see some of the consumer technology today, do you see any technologies that came out of the space program in the 1960s?

AC: You name it.  First of all, we had all the radar sites, and all these sites in the United States where we knew exactly what the latitude and longitude were. That was done by people who were on Mercury. Not IBM, but other vendors who were doing that — and they did a great job. And that of course helped GPS.

Guys who worked for me then went on to were in charge of the airline reservation system. The SABRE and the Passenger Airline Registration System, PARS, came out of that.

The people who developed the air traffic control system, which now has to be replaced from what I understand.  But the one who was in charge of that was a guy that I hired.

Computers are the silent partners behind medicine, particle physics, you name it.  

BI: When the mission was happening, what were you most concerned about? 

NASA Project Mercury Tracking System

AC: We had a great team of people — and I developed a system where we communicated very well — whether it be the launch system or the orbit or the re-entry or whatever. That was number one.

Number two, this system was tested at the sub-system and at the systems level very, very well — very, very carefully. And we used simulation techniques to simulate orbit or launch.  We had data put in so we would think that it was actually real data, and then they would be able to indicate that there’s an abort, and we would get an abort signal or an interrupt, and we’d find a landing zone for that thing.

We had the duplex in case one computer broke down, we had tested out that we could switch from one to the other and it was self-correcting. Even though each computer might be getting data a millisecond sooner than the other, we worked it out so it would all smooth out nicely.

The only thing is that we were a little anxious about were the astronauts. We had met these astronauts. I had never met Shepard, but I met [Gus] Grissom and [Deke] Slayton. The booster, sitting on top of an 83 foot Redstone rocket, with this little capsule and a guy who’s strapped into it, and a tremendous amount of rocket fuel that’s going to send the guy up into space going 5,000 miles an hour, suffering 11 Gs of force. So there was a lot of anxiety about this, but it had nothing to do with the computing.

By the way, this thing started May 2nd, and even before that, we had been prepared for the countdown, then there were weather delays, things happened. So it didn’t go off until the 5th. We were sleeping on the ground. We had people who brought in mattresses and sleeping bags and I was sleeping on top of the console of the 7090 I did damage to the 7090. IBM never fined me for it, but I dented the vent.

MR: How many people were working on your team?

AC: I had about 100 people. We had machine operators, we had systems programming, and they were terrific. I mean, that was key. I mean, the monitor that was handling the real-time channel. But they were really systems people, not necessarily mathematicians…say, off hand, about half of the group [were mathematicians].  At least 50.

MR: What do you think about the state of space exploration today? Is it going private?

AC: NASA can give you a better answer about their plans. Of course, right now the budgetary problems are crippling what they want to do.

I don’t know exactly what the private people — if it’s just tourist stuff, you know, that’s private enterprise, we can’t really talk against that. But I would think that the government will have to still be involved with any concerns of looking into deep space.  I think that’s important work that I don’t think you would be having private entrepreneurs doing that kind of work. I think that’s important, and I hope NASA will stay in that business.