IBM's Early Computers

by Charles J. Bashe, Emerson W. Pugh

In describing the technical experiences of one company from the beginning of the computer era, this book unfolds the challenges that IBM's research and development laboratories faced, the technological paths they chose, and how these choices affected the company and the computer industry. It chronicles the transformation of IBM into a computer company in a remarkably few years, discussing projects that ended in frustration as well as the more successful ones, and providing a sense of the atmosphere, the people, and the decision-making processes involved during the company's rapid technological transformation.

IBM's Early Computers is a unique contribution to the modern history of computers. It focuses on engineering alternatives rather than business and general management considerations and reveals the significance of imaginative solutions to problems in design and technology, from initial experiments with electronics in digital machines to the threshold of the System 360 era. This fair and balanced account of IBM's role in shaping today's electronic revolution identifies the individuals (both inside and outside the company) whose pioneering work influenced developments at IBM.

The book's fourteen chapters briefly survey the card machine era and then cover electronic calculation, the magnetic drum calculator, the Defense Calculator and other first-generation products, ferrite core memories, magnetic tape, and disk storage development, programming, transistors, "Project Stretch" (which involved disappointments but led to one of IBM's greatest successes) high-speed printers, research, and new-product-line considerations.

Genres: History

744 pages, Hardcover

First published December 3, 1985

Reviews

[Jerry · Rating 5 out of 5]

Fundamental to the concept of the general-purpose digital computer was its adaptability to a wide variety of computing tasks. The instrument of that flexibility was the program, whereby the capabilities of the machine were marshaled and orchestrated to accomplish a specific purpose: the processing of a specific set of data to produce a new and more useful set.

Last fall I saw the Aiken-IBM Automatic Sequence Control Calculator, the ASCC as it’s designated in this book. It’s an impressive piece of machinery. It’s an electromechanical calculator, “endowed with seventy-two accumulators, each consisting of twenty-four electromechanical counter wheels… Sixty nonadding registers, similar in format, provided for the storage of constants; these registers were built of dial switches.”

At the start of the book, IBM is making electromechanical punched-card machines; twenty years later they’ve advanced past vacuum tube calculators and computers to solid-state computers. They continued making their electromechanical machines well into the fifties, although that’s not the focus of this book.

Part of the reason they needed to continue making electromechanicals is that the new computers were too expensive for the average customer. This is partly because they ignored their average customer in favor of what governments were willing to fund. It got so bad that some of IBM’s European subsidiaries began designing their own machines to meet the bulk of their (lower-end) markets.

At one point, IBM’s president directed them “to meet the requirements of the ordinary businesses we serve” and was basically ignored, because even at the low levels of the time government grants and contracts were more lucrative. Even large companies with long-term, very lucrative needs tended to be ignored. It took a chance encounter with American Airlines’s president for IBM to start taking the booking needs of airlines seriously. The result was the SABRE booking system which meant IBM terminals in every major (and many minor) airports and airline offices across the nation as well as gigantic hubs to store booking information.

This resulted in both internal and inertial problems switching to new technologies; they were continually adding new features as competitors came out with new products.

The book is designed around topics rather than time periods. Each section is very readable on its own, and each chapter as well. Because topics tend to overlap, the chapters overlap as well, but there is still a surprising lack of duplication. It’s hard to beat the transition from mechanical calculators to solid state computers as a study in an industry’s revolution; despite their occasional short-sightedness IBM was smart enough to recognize that they needed to evolve their products.

The rush to computerize their product line resulted in some amazing kit-bashing, such as using cathode-ray tubes as memory. The delay between setting a phosphor and phosphor decay could be used as an element of memory albeit one requiring constant refresh (something RAM still requires today, although not as spectacularly). And then there were the studies on using hydraulic memory components.

Much of the components they were using, they used before anyone had any real understanding of them. When using ferrite cores for memory, they ran into a problem with mechanical oscillation, just before the computer had to go out.

There was no time for basic studies or analyses. The engineers placed a large tub out on the lawn, held the wired core arrays over it, and poured over them a polyurethane material, hoping the liquid would dry into a soft coating and and damp out the mechanical vibration. “It was kind of a gamble,” one of the engineers recalled. “We were afraid it might change their switching characteristics, and we had no solvent to remove the material from the cores.”

The book covers many firsts, such as the development of the first full compiler (FORTRAN), the origin of registers, decimal vs. binary calculations, and the origin of the term “assembler”, which was literally about assembling blocks of code into a whole, partly because of the lack of enough memory to both compile code and keep it in memory.

Other terms show up either with different meanings or because the modern term hadn’t been invented yet. The IBM 610, introduced in 1957 as one of the last vacuum tube computers, is described as a “personal computer” by IBM. It was meant for use more as a programmable calculator (as we would understand the term) than a computer, cost $55,000, and was larger than the desk it sat next to in photos.

Engineers at the time also used the term millimicrosecond instead of nanosecond. The latter term hadn’t yet come into use.

This is a fascinating case study of a company successfully shifting to a completely new technology in order to maintain dominance in an industry. That’s not something that happens very often.

Generally you find that you need one good idea, not an assortment of mediocre ones or even an assortment of good ones.—Edward J. Garvey

[Ushan · Rating 3 out of 5]

This is a technical history of IBM computers from Herman Hollerith's 19th century census tabulators up to but not including the System/360 project. It can be considered the background of Fred Brooks's books on the System/360.

IBM was a 1924 rebranding of a company formed in a 1911 merger of three companies: a manufacturer of scales and meat and cheese slicers, a manufacturer of industrial time recording equipment, and a manufacturer of punched cards and tabulators. The tabulator company made machines to count the number of punched cards with holes in specific locations; it was used in the 1890 United States census, and censuses in Austria, Canada, Norway and Russia. Gradually the company improved the tabulator machines and made a sorting machine that used the Radix Sort algorithm. To subtract instead of adding, some machines used ten's complement arithmetic, and some a special hole with the sign bit. In 1944, IBM made a fully automatic electromechanical calculator for Harvard University, which called it the Mark I, with relays and pinwheels; its type of computer architecture is still called the Harvard architecture in its honor. Later there were machines using both relays and the much faster switching vacuum tubes. Vacuum tubes were used in radioelectronics as amplifiers in the linear response range, and in digital computers they were either completely cut off, or saturated; this caused them to degrade over time until the cause of the degradation was found and new tubes were designed. One thing that surprised me was how many of IBM's early computers were used for Cold War military calculations: trajectories of guided missiles, intercepting potential Soviet bombers, cryptanalysis for the NSA; I wonder, what percentage of IBM's revenue came from these customers, as opposed to insurance companies and accounting departments. Later chapters talk in great detail about ferrite core memories (there was a special machine to thread them; it wasn't just laid-off seamstresses), magnetic tape, magnetic disks, and the switch from vacuum tubes to solid state electronics: diodes and transistors. The automatic transistor assembly line did not yet need a clean room!

Software did not play as large a role in early computers as in later ones, but there is a chapter on it, too. FORTRAN was developed at IBM, and so was COMTRAN, an early version of COBOL. A sample statement from a COMTRAN program is given:

IF AGE IS MORE THAN 21 AND (SINGLE OR MARRIED AND NO DEPENDENTS) THEN SET DRAFT STATUS EQUAL TO 1.

There was also a great deal of basic applied and research done at IBM. Nikita Khrushchev's May 1960 speech to the Supreme Soviet about the downed U-2 flight was translated into English by a machine translation system developed by IBM and Georgetown University; the result was given to the Congressional Committee on Science and Technology. Arthur Samuel the pioneer of artificial intelligence wrote his checkers-playing program at IBM.

Great things came out of IBM later, too: relational databases, RISC processors, the chess computer that defeated the world champion. Still: what a company; what a time!

[Nathan Davis · Rating 5 out of 5]

This has got to be the most enjoyable book I've read in recent times.
Most computer books are a lot of fun, but there's always that sense of "I wish there were more technical details." Yes they solved a complex issue, but how? This book finally gave enough solid technical details to really suck me in. The only other book to even come close is Soul of a New Machine.
Computers have always had a bit of a chicken-and-the-egg element to me. If you get a new computer, and want to write programs on it you need to get a compiler. Well, where did that compiler come from? Another computer. You can't write a compiler without a computer, and the computer can't make a compiler without something to write it with. So, how did you we get here?
This book goes from the earliest IBM machines which did nothing more than tabulate a stack of punch cards, to the first punch card systems with limited memory, to punch cards that could store code, to punch cards that could modify internal code, to tape machines, to disk drives.
One thing that blew me away was how difficult a challenge memory was and the solutions involved there. Memory was a unique problem because you needed a way to reliably record information, then quickly read it back or modify it. Radio tubes could be used, but they were unreliable and created enormous amounts of heat. One solution for memory came by using television tubes as a form of memory. The TV would display a grid of 16x14 blocks. They would blast a section with a large amount of voltage, and this would cause it to glow brighter than neighboring sections as long as a constant voltage was applied. Then, on top of the TV screens was an array of 16x14 sensors that detected the light level to "read" the bit as on or off.
So the computer had this giant cabinet that was filled with a half dozen tv screens displaying the bits. For the time, it was interesting solution that was wonderfully reliable and afforded the engineers a way to literally read the computers current memory by hand.
Another thing that I found interesting is why the tubes were so unreliable. Radio tubes are used at about (I forgot the exact amount) around 5~10volts to amplify a radio signal, whereas Tubes used for radios they had to run it at about ten times the voltage to get it to switch fast enough. Furthermore, the tubes weren't amplifying a signal they were switching to the maximum current the medium could handle, thus making it the equivalent to turning your radio up to full blast and down again thousands of times a second. They were never intended to before at that level, and as such burned out quickly.

But what makes me really sad is, I doubt I will ever meet another person who has read this book. It's a pretty long slog :(

[Kyle · Rating 4 out of 5]

While this requires some knowledge of computer programming and maths, it is a fairly accessible history of computing, from IBM's perspective.
I read the book more for the general technical history than for the corporate history, and it was suited to this. There is quite a bit of biography, especially at the beginning, it focuses quite a bit on some dead-end technologies (cathrode ray tube memory) and it generally takes a pro-IBM view of things, but it's worth working through.

[Tobias Eriksson · Rating 3 out of 5]

A very slow read, nowhere as good as The Soul of A New Machine. But it was still interesting to read about the earlier days of computing.