Who invented something depends on your definition of 'something'

-February 17, 2011

It’s simplistic to say Bardeen, Shockley, and Brattain invented the transistor. Equally shallow is saying Kilby was the sole inventor the integrated circuit. There is no argument; they were smart hard-working fellows. But the process of invention is a lot messier than we like to think it is. My boss, EDN editorial director Ron Wilson has pointed out that one of problems with the patent law is that it forces the government to select one and only one winner. That is why some of my cleverest friends do not file patents. They say it is just an invitation to spend $450k in federal court. They prefer trade secrets and just moving faster than the competition.

Look at the invention of the transistor. We like the story that three Americans, working for that venerable government-sanctioned monopoly, Bell Labs, invented it. But in 1948 Herbert Mataré and Heinrich Welker working for Westinghouse in Paris also invented a transistor (pdf). They called it the transistron and you could argue it was a more polished and professional product. They came out with their transistron about the same time that Bell labs did. The really disturbing thing is that Mataré invented and observed transistor action in a point contact transistor in 1943, 4 years before Bell Labs announced it’s transistor.


matare_transistron.jpg

The transistron was invented in 1943 and brought to practice in 1948, a few months after Bell Labs announced their transistor.

bell_labs_transistor.jpg 

The transistron was a bit more finished than the Bell Labs transistor pictured above.

You can understand why we were never taught Mataré’s story when we were in grade school. See, he was in Germany. Yeah, in 1943. Uh-huh. Working for the Nazis. So that headline does not read too well-”Adolph Hitler funds invention of the transistor.” So, out Mataré goes. His physicist pal Welker worked for the Nazis too. Yuk. So because we don’t like the politics of their funding source, we sweep their considerable accomplishments under the rug. History is indeed written by the victors. One can marvel at the irony of the Nobel Prize being deprived from a nice Belgian guy who worked for a despicable racist, and given to William Shockley, who was a despicable racist.

The case of the integrated circuit is even more complex. Jack Kilby was a great guy, humble and decent and made great strides while working at Texas Instruments. But a lot disagreements and arguments and confusion come down to definitions. It depends on how you define “integrated circuit” as to who invented it. All my Silicon Valley pals, who are probably prejudiced against a Texas company, will tell you that what Kilby made was a lab experiment and you could never sell it. What made an integrated circuit was the planar process invented by Jean Hoerni while at Fairchild. It was the planar process that allowed you to connect all the transistors with metal, instead of using bond wires like Kilby did. TI and Fairchild sued each other for years until the management realized that either set of patents were useless without the other, so they agreed to share them and not claim first rights. TI forgot that deal a little when its 50-million dollar marketing department came up on the anniversary of Kilby’s work. Kilby was a class act and real gentleman, and always gave credit to Jean Hoerni for his accomplishments. It is widely believed that Hoerni would have shared the Nobel Prize with Kilby if he were still living at the time Kilby won.

And then there is the problem of vision and intent. If you read the great History of Semiconductor Engineering by Atmel process expert Bo Lojek, you learn that even Hoerni himself did not fully appreciate the significance of the planar process. It was fellow Fairchild employee Bob Noyce who jotted a note that demonstrated that he saw where this could go. Noyce was later a founder of Intel.

And then you come right back to definitions. If Kilby invented ICs in 1958 when he used bond wires to hook separate devices together, what did Oscar Darlington invent in 1952, when he used bond wires to connect transistors built on one substrate? Oscar swore that he wanted to say you could have more than two transistors in the patent, but the lawyer talked him out of it, saying to keep the scope specific to insure that they would be awarded the patent.


kilby_ic.gif

Jack Kilby invented the integrated circuit in 1958. It is what we would call a hybrid circuit today.


darlington_1952.gif

So what did Oscar Darlington invent in 1952? He too had bond wires connecting devices, but his were on a common substrate, something much closer to a modern definition of an integrated circuit.

Now, this is not to minimize the contributions of Kilby and all the other pioneers of the semiconductor industry. But invention is a convoluted messy business, and out on the edge, everybody is adapting and borrowing and vamping off others. Its more like a Delta blues player taking a song from someone and adapting it. Mississippi Fred McDowell talked about copying music but then “Playing it my way”. I am sure the ASCAP and BMI would see things differently. The copyright lawyers would be all over those old blues players and the music would have suffered greatly for it.

  • “I’d get the sound of it in my head, then I’d do it my way from what I remembered…” “I made up a lot of the songs I sing. It’s like you hear a record or something or other. Well, you pick out some words out of that record that you like. You sing that and add something else onto it.” — “Mississippi” Fred McDowell

And while you are pondering how murky and indistinct the process of invention is, here is a nice article about why Bell Labs missed the IC. Turns out there was a manager with a strong personality, maybe even a bully, and he didn’t think the IC was very important. So you have the lab that is credited with inventing the transistor completely missing the integrated circuit.

My mentor, Big John Massa, pointed out a similar story with the personal computer. Why would IBM, the mainframe people, be successful with a PC? It was DEC, the people that made mini-computers that, by rights, should have been the ones to see the need and fill that need first. My mentor pointed out that DEC did try to make a PC with something he remembers as the Rainbow, and he told me DEC really didn’t get what the PC was all about.

In this day and age people marvel about Microsoft missing out on the handheld OS market. But it does not surprise me. It is just as described in the great book, the Innovators Dilemma (paperback for you cheapskates). Harvard professor Clayton Christensen explains that big companies get set in their ways. The cable-operated steam shovel people look down at hydraulic backhoes. The big giant steel mills look down on mini-mills. The big 5-inch disk drive makers ignore the 3-inch drive proposed by the engineers. It was a great book and you must read it. He learned that the technical people always push to work on the new thing, and marketing and management discourages them. In the disk drive case, the marketing people at the companies that made 5-inch drives went to their PC customers who certainly did not want to pay more for a drive that held less. The marketing types never thought it was a new market, and they should ask Compaq and the laptop companies, who loved 3-inch drives. Conner Peripheral was born from the technical guys that had tried to convince the 5-inch drive company to introduce a 3-inch drive.

Loading comments...

Write a Comment

To comment please Log In

FEATURED RESOURCES