IT History Society Blog

Introduction

The World IP Day program was to promote and celebrate the many benefits of intellectual property in San Jose and the SF Bay Area. San Jose and Silicon Valley lead the nation in patent generation and the city co-hosted this West Coast event to celebrate the contributions of innovators and creators worldwide.

Marcian E. “Ted” Hoff’s keynote speech is summarized in this article. Other speakers included: James Pooley, Deputy Director General of Innovation & Technology at the World Intellectual Property Organization (a United Nations organization based in Geneva), Michelle Lee, the Director of the Silicon Valley Office of the U.S. Patent and Trademark Office, San Jose Council Member Rose Herrera and San Jose Mayor Chuck Reed.

Early Inventions and a Surprise Phone Call

At age 12, Ted Hoff became interested in electronics. He started reading electronic design magazines, built a short wave radio from a kit, and then built an oscilloscope and repaired TVs. After graduating from high school in 1954, Ted got a summer job doing electronic assembly work at General Railway Signal Company in Rochester, NY. This work involved using transistors, which were very new at the time.

During the following summers away from college he continued to work for General Railway Signal Company. That work led to two patents, one covering a circuit which detected trains using audio frequencies transmitted along the railway track and the second covering lightning protection for that circuit. The filing date for the first patent was in 1955 when he was only 17 years old!

After completing his university education (BEE 1958 from RPI, MS 1959 & PhD EE 1962 from Stanford), Dr. Hoff worked for Stanford Professor Bernard Widrow- PhD, as a Research Associate until he got a surprise phone call from Bob Noyce in 1968. Noyce had just left Fairchild Semiconductor and asked Ted to join a new semiconductor memory start-up company named Intel.  Ted jumped at that offer and became employee #12 at Intel where he worked till 1983.

Semiconductor Memory Challenges and Opportunity to Replace Core Memory

In his early years at Intel, Ted designed and developed semiconductor memories and generated application information for them. Intel’s goal (which few at the time believed was realizable) was to replace magnetic core memory in computers. In 1970, Intel predicted the IC memory price would drop to 1 cent per bit by 1972 and that would be the crossover point. In contrast, today you can get over 100M bits of memory for 1 cent!

But cost was not the only issue. The most dense semiconductor memories at that time were DRAMs (dynamic random access memories), which were characterized by the structural simplicity of its storage cell: three transistors per bit compared to six used in static RAMs (SRAMs). Later DRAMs would be made with only one transistor and a capacitor to store each bit. Both static RAMs and DRAMs lose data if power is removed, while core memories do not.

In addition, DRAMs are volatile, e.g. storage is lost if the data are not accessed frequently. To ensure data retention, refresh circuitry was added to refresh each row of memory cells every few milliseconds. This made the DRAM controller’s logic circuit more complicated than SRAM, but that was outweighed by the fact that DRAM is much cheaper per storage cell and because each storage cell is very simple thus providing higher densities.

Ted wrote a mini-textbook on Semiconductor Memories (he calls it a pamphlet) that this author used in a SCU Graduate EE course on that subject in Fall of 1971!

As the time to design, package, test and sell semiconductor memories was quite a long process, Intel decided to develop custom circuits for various customers in order to generate revenue. One of those customers was a Japanese company named Busicom who (in April of 1969) asked Intel to design MOS LSIs for a family of calculators.  Hoff was assigned the duty of helping Busicom engineers transfer their chip specifications to appropriate engineering teams within Intel.

Many complex chips were involved in that initial design effort, e.g. BCD arithmetic, shift register control, keyboard scanner (and de-bouncer), printer control, display control, etc.  As he learned more about the Busicom design, Ted became concerned that it might be a problem for Intel and with Bob Noyce’s encouragement.  As a result, Dr. Hoff proposed to greatly simplify the design by eliminating many proposed chips for the calculator. The simplification was accomplished by implementing the functions as stored programs running on a very simple general purpose central processor.  Hoff was later joined by co-designers Stan Mazor and Fredrico Faggin. The architecture he had defined was sufficiently simple that it allowed the central processor to be implemented as a single MOS LSI chip -the Intel 4004 (U.S. Patent #3,821,715)- which was released as a general purpose product in November 1971.  

A few months later (April 1972), Intel introduced the 8008 8-bit microprocessor.  In April 1974 the very popular Intel 8080 microprocessor was announced along with a set of peripheral LSI chips and a development system.  By 1975 the microprocessor was established as a replacement for many MSI random logic designs and as an embedded controller for many different types of devices and equipment.

Monolithic CODEC for Digital Telephony

In 1975, Bob Noyce asked Hoff to study the telephone industry to see if there might be a role for Intel to play. Hoff put together a team that developed a single chip CODEC (COder-DECoder), which converted between the analog signals associated with a conventional telephone call (POTS) and a standard digital telephony format known as Pulse Code Modulation (PCM).

At the time, the principal use of CODECs was for T1 carrier digital transmission between telephone company (telco) central offices. However, the Intel CODEC allowed extending the use of that PCM digital format into new application areas, such as private branch exchanges (PBXs).

The CODEC does conversions in both directions between analog and digital representations. Analog signals are converted to a 64 kilobit-per-second PCM digital signal and vice versa. Voice coding and signaling are included.

With the signals in digital format, Hoff recognized that some functions usually performed by analog circuits, such as touch tone processing, would need to be handled digitally. His group developed an early Digital Signal Processor (DSP) to provide such functionality. When a CODEC is used, some filtering of the analog signals is needed, and Ted was fortunate to have consulting assistance from Professor Paul Gray of UC Berkeley. The result was that Intel could offer its CODEC with a switched-capacitor filter. The project succeeded and by 1980 Intel was offering CODECs, switched-capacitor filters and a DSP chip. These telephony LSIs appear to have been the first commercially available.

Career after Intel

In early 1983, Ted joined Atari which had extensive R&D programs underway for PC applications and video games. Unfortunately, Atari didn’t have good financial controls nor good visibility into its market. While initially very profitable, the video game market really started to saturate. Without market visibility, Atari overinvested. They went from annual revenue of well over $2 billion down to below $1 billion in the space of a year. The unanticipated market changes and bad financial controls led Warner Communications to sell the company in mid 1984.

In 1986, a former close co-worker of this author (Gary Summers at Signetics from 1976-79)) started a company called Teklicon to do consulting to attorneys who were looking for either testimonial experts or advice about where to find particular art (i.e. intellectual property or patents). Ted worked with Teklicon from 1986 doing IP research and expert witness work till his retirement in 2007.

In this, his second career, Ted learned that patents were critically important to the semiconductor and high tech industries. He thinks that IP awareness should be part of every engineer’s college education.

Awards and Accolades

Ted received the National Medal of Technology and Innovation from President Obama in November of 2010. He was awarded the Stuart Ballantine Medal in 1979 and The Franklin Institute’s Certificate of Merit in 1996. Hoff was inducted into the Inventors Hall of Fame in 1996 and goes there each year for a reunion with fellow inventors. He was a recipient of the Kyoto Prize in 1997, and the James Clerk Maxwell Award of the IEEE in 2011. He was made an Eminent Member of the IEEEs Eta Kappa Nu in 2013. He is a Life Fellow member of the IEEE.

…………………………………………………………………………………………….

The mathematician Andrew Hodges, author of the definitive 600-page biography of Alan Turing, also wrote a short book about this British mathematician which was published in the series “The great philosophers”. And I asked myself: what is Alan Turing doing in a series about philosophers, alongside Socrates, Spinoza, and Kant? If you ask anyone what Turing had done, they’ll tell you he cracked the Nazi Enigma cipher using computing machinery. Now, that is the work of a hacker, and indeed Turing was arguably the world’s first hacker; but a hacker is not a philosopher. Or they might tell you he designed the pioneering ACE computer, but that is what Engineers and Computer Scientists do; I’ve been an Intel engineer for many years, and nobody called me a philosopher!

 

Alan Turing (1912-1954)  is widely considered to be the father of computer science and artificial intelligence.  He once said that a computing machine was intelligent if it could deceive a human being into believing it was human.  Science fiction writer Arthur C. Clarke later dubbed that the “Turing Test.”  

 

The term Artificial Intelligence was first coined at a 1956 Dartmouth conference on thinking machines.  AI went through several distinct periods after that.  Grady described the following phases of AI:

• Cybernetics
• Semantic information processing (or symbol manipulation)
• Knowledge engineering and expert systems/ inference engines
• Brute force methods
• Statistical approaches
• Societies of the mind

Other references on AI History (independent of this lecture):

Mr Booch showed a short video clip on the architecture of IBM’s Watson AI computer which attempted to address how it works.  But that wasn’t really accomplished in the tongue and cheek dialog between Grady and Watson that was shown.  Watson is an artificial intelligence computer system capable of answering questions posed in natural language.  It was developed in IBM’s DeepQA project by a research team led by principal investigator David Ferrucci.  The machine was specifically developed to answer questions on the quiz show Jeopardy!  It learns from experiences which improves its subsequent memory searches.

 

But Watson is now on to bigger and better things than game playing.  It’s now being used a physician’s assistant at hospitals. In February 2013, IBM announced that Watson software system’s first commercial application w  ould be for utilization management decisions in lung cancer treatment at Memorial Sloan–Kettering Cancer Center in conjunction with health insurance company WellPoint.  IBM Watson’s business chief Manoj Saxena says that 90% of nurses in the field who use Watson now follow its guidance.

 

Source:  http://en.wikipedia.org/wiki/Watson_(computer)

 

 

A clip of Roger Penrose was quite illuminating.  Penrose said that natural understanding depends on conscious thinking, which can NOT be simulated on a computer.  Mr. Booch respectfully disagreed with that assertion. But he wonders how far we can go in building a sentient mind in a machine or device made from silicon and software.

 

We then saw a clip of a female Canadian android parsing natural language (English), feeling pain and sensing touch.  Then a “Google Car” driving itself.  Grady proposed “intelligent drones” that would avoid civilian buildings as targets and hence reduce collateral damage from drone strikes.

 

 

We are slowly surrendering our choices and intelligence to “sentient devices” like Siri – the personal assistant accessed via Apple’s iPhone.  Outsourcing human jobs to machines has contributed to the “global economic malaise,” according to Grady.  That’s because many jobs are being replaced by computing machines (e.g. switch board operators by automated computer attendants).  

 

 

The European Commission (administration for the European Union countries) has selected research projects to study how the human brain works.  Those projects may simulate brain functions.

http://www.nature.com/news/brain-simulation-and-graphene-projects-win-billion-euro-competition-1.12291

 

DARPA is funding projects that will build silicon replicas of neurons.  With 100B neurons and 1B x 1B bits of information that’s really BIG DATA ANALYTICS!

 

And the Obama administration has proposed a decade-long scientific effort to examine the workings of the human brain and build a comprehensive map of its activity, seeking to do for the brain what the Human Genome Project did for genetics. 

http://www.nytimes.com/2013/02/18/science/project-seeks-to-build-map-of-human-brain.html?pagewanted=all&_r=0

 

 

We are on a journey to build replicas of sentient minds using silicon and software.  Such sentient devices would give us the illusion of being human.  Can we build such devices that dream?  Maybe.  Dreams could be automated and simulated.

 

Grady believes that sentient devices are inevitable.  How then will we co-exist, evolve and live with them?

 

 Twitter:  grady@_booch

Website:  computingthehumanexperience.com