The most important invention of the 20th century? To one who was there, it was disk.

The original IBM RAMAC shipped to its first customers in 1956, and the history of magnetic recording changed forever. The first impact was felt in inventory control, where the direct access of the huge drive saved serious money and time. One of the original team at IBM for commercialization of the RAMAC was Bill Donnelly, an IBM staff member and consultant from 1956 to 1998. Bill recently reminisced with editor-in-chief Mark Ferelli.

How did the RAMAC come about?

I went to San Jose in 1956, just prior to the shipment of the first RAMAC. They had built, I believe, nine test models and installed them in different places around the country--like Zellerbach Hall in San Francisco--as test machines. So I wasn't part of the original development of the RAMAC, but I was there from the very beginning, when they started getting serious about building and testing the machines.

How was it received by the original customers?

Very well--or else we would have killed it. It was really unique, because suddenly you had access to what seemed like an enormous amount of information. Five megabytes--five million characters. That would enable a company to keep track of 50,000 inventory items--or something like that. It was a huge step forward. And the unique thing about it was that it had what we first called "random access" and later called "direct access." You could get to any piece of information in less than a second. Prior to that--when they used magnetic tape--if you were searching on magnetic tape the odds were that you would have to search one third of the tape before you could get to any unique piece of information. It was many seconds, even minutes, before you could find something. Then this machine comes along and in half a second you can get to any piece of information. And there were 50,100-character records. So you could keep records of inventory, customers, accounts receivable, and so forth--and it was called the random access machine, RAMAC. And then the spin-masters got hold of it and changed it to the Random Access Method of Accounting Control, and that became the acronym for RAMAC.

When they had already built the development models, I went to see the 305-a RAMAC. I looked at the signals coming off the file and they looked very erratic. My first impression was, "I don't think that this thing will ever work." But I wound up writing the instruction manuals for maintaining and repairing the machines and also taught some of the first classes on it. I knew the machines forwards and backwards, from writing maintenance manuals and teaching people how to keep it running and how to make it run when it didn't.

What was the biggest technical challenge in keeping them up and running?

Just that it was an extremely complex machine. It had every technology that was known at that time. It had core memory, it had the first rotating disk, it had a start program and front board program--it had extremely complex mechanics. It had a card reader and a card-punch, air compressor and a typewriter. So you had to be an extremely gifted person to be able to maintain this machine, because you had to be a mechanic, an engineer, a logician, an early programmer. To me, it was amazing that we could get the machine to run for any period of time, because it had hundreds and hundreds of vacuum tubes, it had relays, it had critical timing measured in milliseconds, and we knew we were onto something really big because it hadn't been done before. And we also knew that if we could get this thing going, it would create a whole new industry. A lot of people call San Jose, "Silicon Valley". But we called it "Iron Oxide Valley" because for a long time we were a bigger part of the economy there than chips.

[ILLUSTRATION OMITTED]

Where did the RAMAC go after those initial shipments?

The first one we shipped was to United Airlines in Denver. And it was the first time you could walk into a terminal in Chicago and immediately get an answer to whether you could get a particular seat or not. So we built 1600 RAMACs, which were 5MB apiece. In the personal computer that I'm sitting on right now, I have five times as much capacity as we had built for the entire world at the time. I have a 40-GB drive. We built 8GB for the entire world. I was also one of the historians of disk, because I was there from the very beginning until I retired in '92 and was a consultant until '98. We built 8GBs total for the whole world. Then, the RAMAC could do 30 instructions per second. Today, they measure computer power in millions of instructions per second. The RAMAC--just the disk drive, not the entire system--was $10,000 worth of megabytes. Today it's less than half a cent worth of megabyte. For all of the RAMACs we built for the world in 1956, you would have paid 80 million dollars. Today, you can buy that same disk capacity for four bucks. I base that on a recent ad for Fry's [laughs]. But you can buy a 250-GB drive for $100-$200.

How long did it take for customers to start asking for more capacity?

Immediately. The first thing we did was to build a double-density version and then a quadruple-density version ... and then we got really aggressive. A couple of guys started talking about building bigger, faster machines. One of them wanted to build arrays of these machines. The first RAMAC weighed 21,040 pounds. Today, a hundred times that capacity weighs less than a couple pounds. And we printed on this machine at two lines in a minute. Today, two lines on a production machine--not PCs but big machines--you can print three hundred pages in a minute. We knew we were onto something big, but it got bigger and bigger. And to me it's incredible, the changes that men make.

Today, if you took a dot at the end of a sentence, you could store as much information as we had on the RAMAC--which is fifty 24-inch disks. We started at 24-inch diameter disks and the last disk that I worked on that had anything to do with IBM was 24mm. That's about the size of a nickel. In that one nickel, on one disk, was 200 times the same capacity as the RAMAC. And on the RAMAC we recorded on both sides. On the 24mm, we only recorded on one side. So the change has been just enormous.

Also, I was the test engineer on a product called the 380 Console Inquiry Station, the first terminal where you could ask questions of the machine about any of those 50,000 items. For example, there was an inventory where you could ask: "How many do we have on hand?" or "How many are being ordered?" or "Where do we get them from?" We had built a kind of typewriter device that would enable you to interrogate what, at the time, seemed to be a huge database. Today it's nothing. But to me, it's the thing that makes everything go. Today, I can get on the computer and talk to a couple of people and make a reservation from San Francisco to Paris to London in September. It's a disk that makes all of this possible. Because somewhere, someplace a disk is moving which goes in and finds out the status of that airplane. How many seats are already sold? Are there any seats available? What is the price? All of this stuff is the result of the work that was already done with the original RAMAC.

About 10 years ago, George Gilder came out with a publication called "Microcosm." In it, you suggested that rotating memory would be replaced by solid state. It hasn't happened.

It's inevitable that it's going to happen. This is probably the most important statement that I will be making. I was there at the very beginning. We started out at 2,000 bits per square inch in areal density--28 tracks and 100 bits. And I can remember very gifted people saying that 10,000 was the upper limit. And then dumb guys who didn't know that that was the upper limit came up with ideas that would take us farther. One of my very best friends was criticized in IBM because he told The San Francisco Chronicle that one day we would be at a billion bits per square inch. And key people said that we would never be able to reach that. They said: "Don't tell people that because we're reaching the physical limits of this technology." But every time we hit the physical limits, we'd raise them another inch. Even since I left in 1998, they've taken the bar much higher. When I was there, a real big drive on 3.5-inch technology was 10 megabytes, now it's 250. So, it was a great ride for me and I'm still amazed at what's been accomplished. And even knowing a lot about how it's done, I'm still amazed at what people have been able to do with this technology.

If RAMAC hadn't come along we would be back in the 1940s in terms of technology--the number of bank tellers we would have, the number of airline people we would have, the number of hospitals we would have, the number of educators we would have, the number of doctors we would have ... It affects every facet of your life, whether you realize it or not. Even when you buy gas at the gas station, there's a disk somewhere that's moving around and keeping track of everything that's going on. Every time you use a credit card. Every time you write a check. Every time you make a telephone call. All that stuff is recorded on disk somewhere. And it doesn't have to be close to you. I don't know how United Airlines works anymore, but it used to be no matter where you went everything happened in Denver. For American Airlines, it was in Miami. You're in a little town in Omaha or somewhere, making a reservation, and the disk is moving down in Miami. So it's really been a tremendous accomplishment. In fact, about 10 years ago, I argued about what was the most important invention of the 20th century. And I argued for disk.