Chapter 2

Why IBM Sucks


We all know IBM sucks. Here’s why.

Their Brains Were Small and They Died

We knew a guy once who was a song writer. He used to sing a song about dinosaurs-all the kids in the audiences used to love it when he sang this song. We don’t know-dinosaurs-it must be a kid thing. Neither of us remembers much of the song except one line of the chorus that we used to sing along with: And their brains were small and they died. Well, that’s IBM. Small brains; dead dinosaur. IBM probably died several years ago, but something called institutional inertia keeps it looking like it’s still living. Its sort of like an animated dinosaur exhibit in a museum.

Dinosaurs Also Ate Their Young

Before we tell you why IBM sucks, let us tell you why, until a few years ago, any Macintosh fanatic should have loved IBM. If it weren’t for IBM, Apple Computer-and the Macintosh-probably wouldn’t exist. We would all be using Windows (shudder), and everyone would think the IBM PCjr was a Pretty Good Home Computer.

So love IBM for making the personal computer ubiquitous and for making truly awful personal computers a de facto standard. Apple couldn’t have done it without them.

Back when IBM designed its System 650 mainframe (in the early 1950s), it figured it could sell about 50 of them. So IBM, being full of astute bean counters, built the cost structure assuming they would sell only 50 of the beasts. This would have allowed the company to make a profit on these paltry few units. IBM actually sold about 1,500 System 650s, pushing profit margins up to around 70 percent or so. IBM (as well as Digital Equipment Corporation and Control Data Corporation) made a lot of money by convincing people that computers were very big, very expensive, and very complex. That’s not why IBM sucks.

When Bob Noyce and seven other engineers quit working for William Shockley and formed Fairchild Semiconductor, IBM didn’t take notice. Semiconductors were and are terribly expensive to develop, but incredibly cheap to manufacture. Success in semiconductors comes from selling huge quantities of them at low prices rather than just a few at a Very High Price. IBM, remember, was out convincing people that computers were big, expensive, and complex; the young dinosaur was certain that the computer market wasn’t big enough to support the economies of scale required by semiconductors. Meanwhile, Noyce was selling everything Fairchild made for $1 each and their sales volume was growing rapidly. Each time Fairchild would increase production, it would lower its cost-of-goods-and selling prices-at the same time.


William Shockley (1910-1989) was a physicist who shared the 1956 Nobel Prize for physics for the invention of the transistor. Shockley had been working at Bell Telephone Laboratories when his team developed (in 1948) a way to alter semiconductor crystals so that they could both detect and amplify radio waves. In 1963, Shockley became a professor of engineering science at Stanford University. During the late 1960s and in the 1970s, Shockley caused controversy by his active support of the view-explored in the work of men such as Arthur Jensen-that intelligence capacity is a genetic trait of races. (Courtesy The New Grolier Multimedia Encyclopedia, © 1992 Grolier Incorporated)


Noyce theorized that the cost of labor would be reduced if several individual components could be combined on the same piece of silicon. He called this idea-and it was only an idea in 1959-an "integrated circuit." Today integrated circuits hold more than a million components.

In 1964, Bob Noyce formed Intel with Gordon Moore to create complex integrated circuits.

An added benefit of integrated circuits turned out to be that higher prices could be charged for them. In 1971, Intel’s Ted Hoff invented the microprocessor-a single chip that held most of the elements of an entire computer. IBM still wasn’t paying attention. Microprocessors were expensive relative to semiconductors, but too cheap for IBM to use. If IBM was to build a computer around a microprocessor, its tenuous illusion of size, expense, and complexity would dissolve. It would be the equivalent of Toto exposing the Wizard of Oz.

IBM’s central computing concept was one of data processing. Large, expensive, and complex data processing. Paper cards were punched with hundreds of holes and run through the computer by technicians. Real people didn’t need computers, according to the IBM vision. But that’s not why IBM sucks, either.

The Altair 8800 was the first practical microcomputer, but the Apple II was the first successful microcomputer and IBM began to take notice. IBM noticed that even though the Apple II was designed to be a home computer for hobbyists, the real market turned out to be business. Trip Hawkins, then in charge of Apple’s small business strategy group, was the first to recognize that a new software program-Dan Bricklin and Bob Frankston’s VisiCalc spreadsheet-would allow Apple to set the standard for microcomputers in large corporations. If IBM flinched at all, it wasn’t noticeable. Sort of like how an elephant reacts to a flea.

The Dinosaur Wakes

Dinosaurs have very small brains relative to their mass. In 1980, IBM decided it would produce a personal computer. Breaking from its normal, glacially-slow dinosaur mode, IBM’s Entry Systems Division committed to producing a computer in exactly one year. (You have to understand that this was a fraction of the normal development cycle for IBM. It ordinarily took IBM four or five years to add functions to an existing product, let alone develop a new one from scratch.) Bill Lowe and his Entry Systems Division team knew that it was impossible to create a new computer in a year. Their only option was to collect existing hardware and software from other companies and paste an IBM label on the box.

IBM realized it needed both an operating system and a version of the BASIC language. After all, each of the other microcomputers available at that time had BASIC. Lowe took the unprecedented step of approaching a five-year-old software company about supplying the operating system and BASIC for the IBM computer. That young software company was Microsoft. (Actually, Lowe wanted to use another operating system, CP/M, developed by Gary Kildall of Design Research, but when the IBM suits arrived in California to meet with Kildall, he was out flying his plane. We cover that in more detail later.)

IBM has become famous (or infamous) for their nondisclosure agreements. Robert X. Cringely in Accidental Empires: How the Boys of Silicon Valley Make Their Millions, Battle Foreign Competition, and Still Can’t Get a Date, describes IBM’s nondisclosure agreement as "the legal equivalent of a neutron bomb, destroying only the people but leaving their technology intact." IBM’s potential suppliers were required to sign this agreement-agreeing that whatever they tell IBM was not confidential, but that whatever IBM tells them was confidential-before any substantive meetings took place.

Microsoft’s founders, Bill Gates and Paul Allen, signed IBM’s nondisclosure agreement without giving it a second thought. The first Microsoft-IBM meeting consisted of IBM suits touring Microsoft’s office. A few days later, IBM returned and asked Microsoft to supply a version of BASIC for IBM’s personal computer that was under development. Gates was concerned because IBM wanted to build its computer around an 8-bit processor, and he told the IBM contingent that he could supply an operating system for a 16-bit processor as well as a version of BASIC to support the better, far more powerful processor. IBM went for it.

At the time there were three 16-bit processor options: The Motorola 68000, National Semiconductor’s 16032, and Intel’s 8086/8088. The Motorola and National Semiconductor chips were powerful and elegant. The Intel 8086 was somewhat less powerful and had a bizarre memory architecture. The Intel 8088 was a neutered 8086, with all of the problems and even less power. IBM, in true dinosaur fashion, chose the Intel 8088 for its personal computer. Dinosaurs have small brains, remember? The 8088 was the only processor then available with finished support chips, and besides IBM didn’t want its personal computer to compete with its real business: Big Computers for Big Organizations.

Bill Gates knew he could probably cobble together a version of BASIC in time to suit IBM, but the 16-bit operating system was going to be a real problem. He knew about QDOS-a 16-bit CP/M clone-developed by Seattle Computer Products, and he knew all he had to do was get it. Without telling Seattle Computer Products why he wanted it, Gates purchased the rights to QDOS for $50,000. To this day, Gary Kildall maintains that most of the QDOS code was directly copied from his CP/M. This claim has not been tested in the courts, though.

So, if you’re keeping score, here’s what that last inning looks like. IBM arouses from its sleep long enough to decide-Argh! Argh!-that it wants to get into the personal computer business. But not so much as to compete with its core business and thus decides to cobble together a machine from existing bits and pieces. IBM approaches Microsoft about writing a version of BASIC for the new computer. Bill Gates smooth talks IBM into letting Microsoft provide not only BASIC but an operating system-which Gates has bought for a song-as well. Never mind that the pedigree of the operating system that Gates wants to pass off to IBM may be somewhat questionable. IBM took the bait and Bill Gates even retained the r ight to sell the operating system to other companies!


Previous Page Next Page

Copyright © 1994-1996 Quadralay Corporation. All rights reserved.