Revised and extended 2018, October 7
K. An epistemological challenge
K1. Philosophy. Philosophy is the most fundamental and most troubled knowledge area. While I was a doctoral student in philosophy at the Graduate Center of the City University of New York, I found philosophy to be in a unique position to:
- Identify aspects that are common to the different sciences
- Resolve issues involving the interrelations among sciences
- Provide unique top-down conclusions for particular fields
- Derive conclusions of what ought to be done
There does not exist any single scientific discipline that can do any of the above. My view was outside the mainstream. Currently, analytic philosophy seeks merely to clarify rather than solve problems. I, therefore, felt pressed to find a down to earth example to demonstrate the problem-solving power and plain utility of such top-down inferences. I chose information technology. The account below is mostly chronological and alludes only briefly to epistemological considerations. These considerations are addressed in the forthcoming revision of The New Foundation of Knowledge.
K2. Semiconductor technology. Following the invention of the transistor (1947), the integrated circuit was invented (1958). During the decade of the 1960s, there were five stages of halving the linear distance between transistors, which quadruples the number of transistors per unit area. Thus, the five stages resulted in more than a thousand-fold increase in the number of transistors per unit area.
Transistors: 1 → 4 → 16 → 64 → 256 → 1,024
To the extent that the cost per unit area remained the same, the production cost-per-transistor dropped by the same thousand–fold factor. In 1968, a California company named Advanced Memory Systems was the first to develop a 1024 transistor (1 kilobit) memory chip.
K3. Time-sharing computer systems. During the 1960’s computers were used as a shared facility. The accepted wisdom was that data processing, like the generation of electricity, ought to be centralized and then distributed to end-users. In addition, electronic engineers shared the view that the solution to a given problem must be customized, that there is no panacea. This view led to the development of application-specific products such as word processors and scientific calculators.
K4. The advantage of generalizing problems and solutions. My view, to the contrary, that a problem must be generalized before a solution is sought, and then the solution ought to be generalized prior to any customization. Word processors, scientific calculators, fax machines and printers were, to me, conceptually outdated. By analogy, I believed that time telling should be based on user-dedicated wristwatch rather than a clock in the center of town, and ditto for computing.
K5. IBM. At that time, IBM was the dominant company in the computer field. Therefore, in 1967 I met with Jacques Maisonrouge, who was at the time the president of IBM World Trade Corporation. I conveyed to Maisonrouge my conviction that there will be a transition from shared computing facilities to user-dedicated computers. I then proposed that IBM explore doing so. Maisonrouge arranged for me to meet J. C. R. Licklider. I did not know it at the time, but Licklider was a key promoter of the time-sharing approach to computing. The meeting proved pointless. IBM eventually entered the personal computer market in 1981. It was too late. By then, they had lost their leadership in the computing field. Subsequently, IBM sold the personal computer business to the Chinese computer company, Lenovo. In his book Inside IBM – a personal story (1985) Maisonrouge acknowledged the importance of the personal computer, writing; “The most important change in the last few years has been the introduction of personal computers in homes, offices, and schools.” (p. 284). He did not, however, explain IBM’s failure to enter the computer field during the 1970’s and its loss of leadership in the computer field
K6.1. The Datapoint 3300 computer terminal. In 1969, Computer Terminal Corporation (CTC) of San Antonio, Texas, (later renamed Datapoint) developed a computer terminal, the Datapoint 3300. Seeking to raise $4 million by an initial public offering, they contacted the Wall Street firm Philips, Appel & Walden, which was at that time specializing in funding high technology companies. James (Jim) Walden, the managing partner, asked me to go to San Antonio to evaluate the technology of that company.
K6.2. My visit. In San Antonio, I had an extended discussion with Austin (Gus) Roche, who was the vice president for research and development at Datapoint. I suggested that the revolutionary advance in semiconductor technology would drive changes in how computing is done. I urged that CTC consider developing a personal computer.
K6.3. My recommendations. Specifically, I recommended that Datapoint:
- Develop a computer to be located where the user is.
- Check if the processor can be implemented on a single chip.
- Make the computer user-dedicated.
K6.4. The Datapoint 2200 intelligent terminal. Roche responded saying that their next product would contain a computer. He thus accepted my first recommendation. Datapoint acted on my second recommendation and asked Texas Instruments (TI) and Intel for proposals of implementing the central processing unit (CPU) of the next product, the Datapoint 2200, on a single silicon chip. But the Datapoint 2200 was designed to be an intelligent terminal for processing data on a remote computer. Thus, my third recommendation that the next product be a personal computer, was not accepted at that time.
K7. Philips, Appel & Walden. On my return to New York, I told Jim Walden that I liked the company, their technical competence, and “can do” attitude but found that their product philosophy was conceptually obsolete. To my surprise, Jim Walden challenged me by saying that if I could do it the right way, then Philips, Appel & Walden would fund me. Despite misgivings, I accepted the offer. I formed a company naming it Q1 Corporation and recruited a core team.
K8.1. Putting off the development of the Datapoint 2200 CPU chip. As Gus Roche told me when we met, Datapoint did develop a CPU of a computer and did ask TI and Intel for proposals to implement it on a single chip. Initially, Intel was uncertain whether it would be in its interest to design a chip implementing the Datapoint CPU. Intel was at that time in the memory chip business. There was a concern that producing and marketing a CPU chip would make Intel viewed as a competitor by its memory chip customers. Intel then shifted priorities from the CPU for the Datapoint 2200 to the development of an electronic calculator chip for Busicom of Japan.
K8.2. Shelving the development of the Datapoint 2200 CPU chip. TI, for its part, filed a patent application for the single-chip implementation of the 2200 CPU, which was eventually granted. However, it was unable to deliver usable chips to Datapoint in time. As a result, the Datapoint 2200 intelligent terminal was produced with its CPU implemented using discrete components. This, in turn, led Intel to shelve the project of developing the 8-bit CPU chip for Datapoint.
K8.3. My meeting with the Robert (Bob) Noyce, the president of Intel. On hearing this, I flew to California to meet with Bob Noyce who was the President of Intel at that time. I conveyed to Noyce my view that since a 4-bit chip is insufficient for representing alphabetic characters, the chip Intel was developing for Busicom would have a limited market. In contrast, I said that the 8-bit CPU that Datapoint developed at my urging would revolutionize information technology, and Q1 would be Intel’s first customer for that chip.
K8.4. Getting Intel the 2200 CPU rights for single chip processor. Noyce said that Intel would have resumed the development of the single chip Datapoint 2200 CPU 8-bit except that it would first need to obtain the consent of Datapoint to do so. I told Noyce that I would provide Intel with the required Datapoint consent. I flew to San Antonio, met with Phil Ray, who was President of Datapoint at the time, obtained the consent that Intel develop and sell the single chip processor based on the Datapoint 2200 CPU, and so informed Noyce. I expected that in the subsequent formalization of such agreement Datapoint would receive some percentage royalty on the sale of the future 8-bit chip. This did not happen. Intel then developed the 8-bit single chip processor, calling it Intel 8008. As I promised Noyce, Q1 became first customer for the Intel 8008 single chip 8-bit microprocessor.
K9.1. The world’s first 8008 and 8080 computer installations. The Q1 computer was designed to be a user-dedicated general-purpose computer system. It was the first personal computer. It was also the first computer system to incorporate a single chip 8-bit microprocessor. In December 1972, the first Q1 personal computer was installed at Litcom, a division of Litton Industries in Long Island.
K9.2. Nixdorf Computer, Paderborn, Germany. Early in 1973, Heinz Nixdorf, the president of Nixdorf Computer, invited me to visit his facility in Paderborn, Germany. I went there with Dr. Ron Sommer, who was a vice president of Q1 and fluent in German. It resulted in a $40k/month software development agreement for the Intel 8008 and anticipated 8080 microprocessors.
K9.3. The New York City Israel Supply Mission. Later in 1973 Q1 received an order, subject to acceptance tests, from the Israel Supply Mission in New York City for four Q1/Lite systems, to be based on the expected second generation of the Intel 8008, later named the Intel 8080. A Q1/Lite computer with a pre-production 8080 microprocessor was delivered late in
1973 and later replaced by a unit with a production-level 8080 microprocessor.
K9.4. NASA. In 1975, the National Aeronautic and Space Administration (NASA) ordered Q1/Lite systems for all its eleven worldwide bases.
K9.5. IEEE. Also in 1975, the Institute of Electrical and Electronic Engineers (IEEE) organized their first international conference about the microcomputer revolution, which took place in New York City. I understand that on the recommendation of Bob Noyce, IEEE invited me to organize and chair the opening session. It felt strange: I am neither an electronic engineer, nor a computer scientist, and had no prior association with the IEEE. Despite being an outsider, or perhaps because of that fact, I was able to see where the field was going and could go in way that people inside the industry could not.
K9.6. The UK National Enterprise Board. In 1979, the National Enterprise Board of the British government invested over $11 million for the right to represent Q1 in Europe.
K9.7. Exit. I then recruited a President to replace me and returned to my main interest in making explicit the epistemological implications that sensations are innate – the direct opposite of the tabula rasa assumption.
K10. World dominance
K10.1. Microsoft. The introduction of the Intel 8080 prompted Bill Gates and Paul Allen to quit Harvard and form Microsoft.
Microsoft developed software, including the Windows operating system, for the Intel 8080 and subsequent member of that family, the x86. By the end of 1970s, Wintel (the Intel x86/Microsoft Windows) became the dominant personal computer engine in the world.
K10.2. $1 billion? Until then, Intel revenues were from selling semiconductor memory chips. By the end of the 1970s, Intel discontinued the memory business, and the x86 became its main source of revenue. Lamont Wood, in his book Datapoint (2012) wrote about the 8008 microprocessor that “In hindsight, it’s clear that this chip, through its direct descendants, was the foundation of the digital world”. In chapter 9, named The Worst Business Decision in History, Lamont writes that relinquishing the intellectual property rights for the Datapoint 2200 CPU to Intel was a $1 billion giveaway.
K11. A personal perspective
K11.1. Confirmation. To me, the sequence of events confirmed the power of abstract top-down reasoning. Some would argue that it was just improbable happenstance, that I just happened to be in the right place at the right time. This can be put to a test. Currently, there is considerable functional overlap among user-dedicated computing devices, including desktop computers, laptops, tablets, cell phones and smart wristwatches. This is a situation where the optimal next step cannot be reached in bottom-up research of computer science or electronics. It requires conceptual top-down reasoning. If any computer company is interested, I would be willing to make explicit how my top-down considerations outline the next phase in information technology.
K11.2. Next. The more important challenge is to apply top-down thinking to more fundamental issues. I try to do this in the forthcoming revision of my book about the foundation of knowledge (2017).
Alroy, Daniel. The New Foundation of Knowledge. 2017
Maisonrouge, Jacques. Inside IBM. 1985.
Wood, Lamont. Datapoint. 2012.