Embedded Files
Why John's Story is Important in Understanding the Kenbak-1
Why John's Story is Important in Understanding the Kenbak-1
To understand the significance of the Kenbak-1, it helps to understand both the state of technology in 1971, and its designer John V. Blankenbaker.
When researching the Kenbak-1 in 2004, it was tempting to ignore the creator and focus just on technical details. The goal was to understand how the Kenbak-1 worked, how it fit into the historical evolution of computers, and to determine if it deserved the "First Personal Computer" title.
The 1970's was a time when many people pursued electronics as a hobby, reading magazines like Popular Electronics and getting books and kits at the ubiquitous "Radio Shack" in every town. As such, it was easy to dismiss Blankenbaker as just a physics major, probably not a true engineer, just a self-taught "hobbyist." But those assumptions were not true. Blankenbaker had a long history of working on early important computer projects, and was awarded 8 patents for novel inventions in computer design. He earned a graduate degree in electrical engineering at MIT, and published well received articles on computer arithmetic, and theoretical works on computer science.
While many biographies of John Blankenbaker exist (see our "Links" for several), this biography is meant to be a "technical biography." This focuses on his technical background leading to his design of the Kenbak-1. Some of this is covered in the 2007 interview by Lee Felsenstein for the Computer History Museum, but includes material from many sources, including emails from John. I especially thank his daughter, Anne Killheffer, who sat down with John to review and fact check an early draft in 2022.
John's Early Life, Navy, and College:
John's Early Life, Navy, and College:
John Virgil Blankenbaker was born on December 24, 1929 to a farming family in Oklahoma. They later moved to Oregon when he was 7. This was far from a high-tech environment - the family didn't have electricity or even a telephone for much of his childhood. His father was a farmer, but considered becoming a minister, and even went to seminary for a time. His mother was a teacher, so she encouraged his reading and education. In spite of simple surroundings, John developed a real curiosity of sciences and how things worked. He recalled taking apart household items and an old automobile the family didn't need. He wondered why water would evaporated from a pitcher, and he watched the stars while sleeping outside on hot summer nights. Radios were cutting-edge technology in the 1930's, so he built a crystal radio, then a one vacuum tube radio. He tried to increase the output by adding a second tube, but it never worked. He wasn't only interested in science and mechanics. He recalled when he was only 10 or 11 years old, one of his favorite books was a college "business law" book belonging to his older brother.
about 8 years old, Albany, Oregon Farm. Photo courtesy of Anne Killheffer)
Albany Union High School Graduation photo.
In high school he took a college preparatory path with a science emphasis.1* By graduation, he wondered how he was going to afford college, so he was enticed into a US Navy program that would pay for four years of college in exchange for two years' service. What's more, they would send him to 11 months of electronic technician school. That was a deal he couldn't pass up.
His Navy training focused mostly on tube radio transceivers, but he also learned about radar, sonar, LORAN, and other technologies. He enjoyed this time in the Navy. For the first time he got to travel to many large cities like Chicago, Washington DC, and London, visiting museums and famous landmarks. He said this was a real "eye opener" "for this farm boy that had never experienced the big city life."
After his stint in the Navy, he enrolled in Oregon State University, and had to decide between majoring in electrical engineering or physics. He decided on physics. He was given credit for his Navy education, allowing him more time to sample a wide range of diverse courses. He took philosophy, psychology, sociology, religion, German, and even appeared in a campus play.2* It was during his freshman year that he first became interested in computers. He read a magazine article about a huge computer with thousands of vacuum tubes.3* Even more interesting, the article said all information was represented by only ones and zeros. That inspired countless hours of thinking and imagination. He figured out for himself how numbers could be represented in binary, and how to do arithmetic on those numbers. He designed logic systems built with relays and electro-mechanical levers. He could only design on paper and pencil because he had no money to actually build anything. He was living off a small Navy stipend and modest savings. He thought about making his own computer back then - mostly to help with the calculations needed in his physics labs.
John had a real stroke of luck his junior year: he won an internship working at the National Bureau of Standards in Washington DC. While hundreds of students got internships at NBS, he was one of only four lucky students assigned to the SEAC computer project in the summer of 1951.
This Popular Science, article from May 1949 was thought to be the inspiration John talked about, but it's not. Still, provides a description of computer technology at the time. Click on the image for the 5-page article on Google Books.
The SEAC computer was a huge computer based on the EDSAC in Britain. It was a serial-architecture machine, and probably the most powerful computer in the USA at that time. Completed just one year earlier, teams were working on improvements. John said his "work was trivial" building a device to test diode circuits, but he learned a lot. The interns were taught Boolean logic and equations, and at the end of the summer, John even got to write a computer program on the SEAC. Unfortunately, his program (which was to find the roots of a quadratic equation) never did work. He was only allowed to use the computer during thunderstorms, since the lightning made the computer results unreliable for any serious work. This constraint, and the short summer, limited his chances to debug the program. This probably made him realize how important close integral access to his own computer would be for programming.
But the rest of the summer was a success. He even was introduced to Ida Rhodes, a pioneering female computer programmer and friend of Albert Einstein. So while John graduated with degrees in physics and mathematics, he received extraordinary experience with digital computers, much more than most electrical engineers at that time. He was already thinking about how he could make his own computer.
It's interesting to note that the SEAC computer was a serial computer, much like the Kenbak-1. Likewise the CPU speed was also 1 MHz, just like the Kenbak-1. It had 512 words, not far from the Kenbak-1 with 256 bytes. But the similarities ended there. Each word was almost 6 times as wide, and it weighed more than 200 Kenbak-1's.
Hughes Aircraft Computer Work:
Hughes Aircraft Computer Work:
After graduating in 1952, John could have gone back to the NBS/SEAC project, but he applied to Hughes Aircraft because they had a program where you could work half time and go to school for a master's degree half time. He was assigned to their digital computer department, where they were working with a suitcase-sized computer for aircraft. Hughes had a strong computer program working to automate many aspects of military aircraft which eventually became the MA-1 System. While the computer still utilized vacuum tubes, it was compact because each "peanut tube" was only the size of a finger.
After only a few months, Hughes Aircraft decided to build a large general purpose business computer. John was assigned to this project, specifically to design the Arithmetic Logic Unit (ALU) and his team worked out the full computer design, and started building the hardware components. It was during this project that a manager told him that each "flip-flop" (an electrical module which can act as a register or a single bit of memory) would add about $500 to the project. That high cost was because each flip-flop was made of 3 vacuum tubes, and other components and hardware, and all the development and fabrication costs would significantly increase total project costs. This put a lot of pressure on John to reduce the number of flip-flops. He spent a lot of time thinking about simplified computers with fewer flip-flops. Eventually he envisioned a single flip-flop computer, but with a large amount of serial memory, and realized this simple computer could solve very complex problems. This simple minimal computer could "emulate" any more complex computer.4* He presented this idea to the patent office at Hughes telling them "This machine, right here, can do what any other computer in the world can do." But in spite of his efforts, they weren't interested. Hughes, however, was impressed with his other ideas, and applied for 6 other patents on his behalf.
Unfortunately, Hughes Aircraft hired a new general manager, and a new market analysis suggested that only about 20 of their business computers would ever be needed by the market.5* Hughes didn't think that would be profitable, so they canceled the project only a year or two after they started it. John was reassigned to other projects, mostly working on control systems used in manufacturing.
Patent 2,923,474 filed in September 1953.
The first of six patents Hughes filed on John's work.
John's years at Hughes were very technically productive, because within two years, Hughes filed four patents under his name. Pictured is one page from patent 2,923,474 "Multiple Input Binary-Coded Decimal Adders and Subtracters" (filed Sept. 2, 1953) but they also filed 2,892,587 "Result-From-Carry Adder-Subtracters" (Filed Sept. 3, 1953), 2,850,233 "Electronic Five's Multiple Generator" (Filed Sept. 15, 1953) and 2,888,202 "Multiple Input Binary Adder-Subtracters" (Filed Nov 25, 1953). He was, undeniably, doing some innovative work on his ALU design project, all while earning his master's degree.
John made the front page of HughesNews as part of a program where patent applicants were given $100 for each patent they were awarded. He received the biggest check. $400, in the group.6*
John remained at Hughes after the business computer was canceled, but he felt the work was less interesting. He did still rack up two more patents: 2,840,709 "Frequency to Digital Conversion" (Filed July 2, 1956) and 2,983,909 "Algebraic Scale Counter" (Filed July 30, 1956.) He also wrote a very well received article titled "How Computers do Arithmetic" which was published in Control Engineering, (Vol. 3, no. 4 April 1956) and was cited in many works and patents for decades. This article summarized ways to digitally represent numbers and methods to perform logical calculations on them.7*
He finished his master's degree at UCLA (in physics) but then decided he wanted more electronic education. While he understood vacuum tubes, the 1950's saw the transition to transistors and he wanted to learn more about them and other electronics. He found that MIT had a program which was just what he wanted.
Photo and bio from Control Engineering 1956.
MIT and Consulting with Ramo Wooldrige:
MIT and Consulting with Ramo Wooldrige:
Massachusetts Institute of Technology's "Professional Degree" was a graduate degree - something between a master's and a doctorate. It allowed people with significant work experience in engineering, but an undergraduate degree in a different discipline, to earn an advanced engineering degree. John enrolled during the fall of 1956 and spent three years there.
During his first summer at MIT in 1957, he got a job consulting for the Ramo Wooldridge Corporation. Simon Ramo and Dean Wooldridge had also left Hughes Aircraft in 1953 to form their own company. This was one year before the merger to form Thompson Ramo Wooldridge (TRW). During this summer job, he worked with Gene M. Amdahl and Lowell D. Amdahl, two "soon to be legends" in the computer industry. This was during a short stint when Gene quit IBM due to frustration with the bureaucracy, only to return in 1960 as the chief architect of the extremely successful IBM System 360 computer. He founded his own Amdahl Corporation soon thereafter.
During this summer John told of his "single flip-flop computer" idea which he had been thinking about for almost five years. This was the minimal computer with only a single flip-flop, but with a lot of memory which could emulate any more complex computer. During this summer at Ramo Wooldridge he finally wrote a paper describing this machine, and it was accepted for publication in IRE Transactions on Electronic Computers, (J. V. Blankenbaker, "Logically Micro-Programmed Computers," in IRE Transactions on Electronic Computers, vol. EC-7, no. 2, pp. 103-109, June 1958)8*
Figure one from Blankenbaker's 1958 paper. Published when he was at MIT and Ramo Wooldridge,
John's paper, written the summer of 1957, and published the spring of 1958, stated at the very end that he "acknowledges that this machine was independently discovered by Gene M. Amdahl and Lowell D. Amdahl." That that statement is a little mysterious.9* None-the-less, his paper was widely read, and earned a lot of notoriety for him.
The summer job led to Thompson Ramo Wooldridge filing for a patent on the work done by the group. It was an incredibly complex 28 page patent, taking eight years and several revisions to be granted in 1966 (patent 3,246,303, "Stored Logic Computer" Filed Sept 11, 1961, patented Apr. 12, 1966.) John was given credit on this patent, but listed 5th out of 5 inventors. John said 56 years later that he wasn't even aware of the patent, but was gracious the Amdahls included his name.
During John's second summer he consulted for Litton in Beverly Hills, which produced small computers for general purpose and defense work. He then took one semester off to work for General Precision designing display terminal systems for an FAA application. He had to write a thesis to graduate, but it took some work to convince his advisor, the soon to become famous Bernard Widrow. John said "I had noted a novel structure and novel method of analysis for the prediction of filtering of binary sequences. When I even presented these ideas to my thesis advisor, he stormed in and said 'You can’t do that!'" With time, John convinced Widrow that this was a useful novel approach, and it was accepted as his thesis work. Years later, Widrow gave praise for John's paper.10* (Blanenbaker [sic], John Virgil. “Prediction and Filtering of Binary Sequences.” Massachusetts Institute of Technology, Department of Electrical Engineering, 1959. Click for Text)
He graduated with his "Professional Degree in Electrical Engineering" in 1959, and married his wife, Eleanor that same year.
Just one page of the 28-page patent with the Amdahls, where John is listed fifth out of 5 inventors. Still not bad for a summer job.
Post-Graduation Consulting, and Scantlin Electronics:
Post-Graduation Consulting, and Scantlin Electronics:
John and his new wife, Eleanor, moved to the Princeton NJ area because it was halfway between Washington DC and New York. He did a wide range of consulting jobs, and he says he "spent some time studying alternatives in computer design." He worked for one year at Curtis Wright making digital logic trainers for a Navy ballistics system. Then in 1962 he consulted for ITT on a telegraph message switching system for the State Department.
Around 1963 he was recruited by Montgomery Phister, to work for Scantlin Electronics. John had known Montgomery since his days at Hughes, as they worked together on the canceled business computer. Montgomery was now in New York working for Scantlin Electronics, which made systems that compiled stock and commodity prices from the exchanges, and provided that data to brokers. John was initially assigned to the New York City office because Scantlin needed people where the exchanges were. John was promised he would be moved out to California in a few weeks, but that didn't immediately work out as planned. John mostly worked on software in New York, mostly done at night since the computers were busy during the day. It wasn't easy because he commuted late hours of the evening, and had to rent apartments in the city. But at the end of 1964 he was finally transferred to the engineering department in California. He worked for Scantlin for many years, improving the speed and capabilities of their systems. He even got one more patent while at Scantlin, 3,483,553 "Keyboard Input System" (Filed June 8, 1967) (shown).
John was named a "Vice President" of Scantlin in March 1970, but he didn't enjoy the supervisory rolls. The company was losing market share and not doing well financially,11* Milton Mohr was brought in as the new president to turn around the company, and wanted John to act as a buffer between himself and the founder, Jack Scantlin, who was being pushed out. John wasn't at all interested in that role. He left in 1970, and was given a $6,000 parting gift.
Unemployed, but with a little bit of money, John wondered if this was the time to build his own computer.
Why John Blankenbaker Wanted to Make a Computer:
Why John Blankenbaker Wanted to Make a Computer:
John had multiple reasons for making his own computer, as explored a bit in the oral history by Felsenstein. First of all, he felt an affordable computer would be really useful to people learning about computers. While there was a lot of interest in computers in the 1970's, very people knew much about them. Secondly, even people who knew about computers had poor access to computers. Programming classes at that time often required students to submit a stack of punch cards with their program. Those cards were sent to a big business computer overnight, and the students would receive a printout the next day showing if the program worked. This was a terrible way to debug a complex program. John recalled his first program on the SEAC, when limited access time to the computer prevented him from getting it working. Lastly, while John knew it would be challenging to make his own computer, he was confident in his abilities. He had been thinking about it for 20 years, and telling people for 15 years that he could make a less expensive computer. This was his chance to prove it.
While the term "personal computer" had yet to be coined, John desired was to make a computer affordable enough so an average person could have their own. That close one-to-one person/computer interaction was a big part of what we think of today as a "personal computer."
Deciding How to Make His Computer:
Deciding How to Make His Computer:
When John started to design his computer, he tells us that he pulled out some electronic parts catalogs (Allied, Newark, and others) to see what parts were available. Intel had recently introduced serial "shift register" memory which were "pretty cheap, and pretty good storage." This solved a major hurdle, as memory would be very difficult to implement with older technologies. He looked over available logic IC's, lights, and switches. He thought through the whole process: design the logical circuit, translate it to a printed circuit board, and put it together in a case. These were all things that he'd done before on other projects. He picked the name "Kenbak" (a shortened mid-portion of his last name), as he felt his last name was too hard to spell or even pronounce. It was similar to the popular "Kodak" name branded by George Eastman, making it short and easy to remember.
He thought about the architecture and instruction set. He decided against interrupts, but did want multiple programming registers, and all the addressing modes he could think of. He wanted a good assortment of jump conditions, and instructions to set any bit of memory to 1 or zero. He wanted shifts, and rotates too. He came up with a very extensive instruction set, which some say rivaled both the PDP-8's instruction set, and Intel's 8080 microprocessor which followed.
He realized he would have to write two documentation manuals: one to teach brand new students with no computer experience (the "Laboratory Exercise" manual) which would start from the basics. For experienced programmers, he needed a "Programming Reference Manual" to give concise details about each instruction.
To design the computer he said, "I am a believer in the state concept, and Boolean equations." The heart of his machine was a state machine, where each state did a particular function, like finding the next instruction to execute, or waiting for the correct memory address to come around to read the data. He worked out the details, not just to execute all the instructions, but also to read the front panel switches during loading and reading memory from the front panel. This required 29 states, just under the 32 allowed with a 5-bit state machine.
Many histories of the Kenbak incorrectly think that John's paper "Logically Micro-Programmed Computers" from 1958 was the basis for the Kenbak-1, and the Kenbak-1 was sort of a "proof of concept" of that paper. But that is not correct. In John's oral history with Felsenstein, John clarifies "It is probably not best to compare this too closely to what I described in logically micro-programmed computers. The connection is that logically micro-programmed computers led me to believe that computers could be much simpler than what was being built and that was my motivation. That was that it can be done, something simpler can be done, but I didn’t go all the way in simplicity because it’d take forever to solve a problem so I chose something in between." He further explained "I have in some other work used the concepts of the logically microprogrammed computer to a greater extent than I did in the Kenbak-1."6 The Kenbak-1 actually had many "flip-flops" or registers, around 75 if you counted limited function flip-flops made of individual gates.
The Bud Industries "Grand Prix" style case,
now iconic for the Kenbak-1, from 1970
Allied Electronics Catalog.
The Raytheon Keyboard Switches, used in the Kenbak-1, [Open in new window to enlarge (from 1969 Electronics)]
He started working on his computer in September of 1970. Within a month, he had the paper design, and started to lay it out on a PC board. He purchased several cabinets that might potentially fit, settling on the "Grand Prix" case from Bud Industries which seemed about the right size. His only help came from his brother who spent a little time helping him tape the traces onto Mylar sheets for the printed circuit boards.
By March 1971 he had the circuit laid out and took the Mylar sheets to the vendor to make up the sample board. This was his prototype board. He soldered in the components and ICs, but it didn't work at first: his "bit time register" circuit to make separate clock signals for the sequential 8 bits did not initialize as expected. He had to add an RC circuit and make several wiring changes fixes, but he got the prototype working.
His initial aim was to make a computer with $150 in parts, but it ended up costing about $250 when buying 50 pieces at a time. He hoped in higher quantities, he could reach the $150 price point.
It was only after he had his working prototype that he sought investors for the Kenbak Corporation. He wrote to many friends he'd known or worked with over the years, and got several to invest a total of about $20,000 in stock. His five investors were John Blattner, Christopher Camp, Jim Dougherty, Vance Holdem, and Montgomery Phister, all of whom had known him personally.12* He demonstrated the Kenbak at high school educator conferences, and advertised in several scientific, educational, and trade magazines. He even demonstrated his computer at the "Homebrew Computer Club" but was unsure if Steve Jobs and Steve Wozniak (the founders of Apple) were in attendance at that meeting. He had great hopes for his computer.
Block diagram of the Kenbak-1 computer, showing all the logical parts
Marketing and Sales Problems:
Marketing and Sales Problems:
John received positive feedback from many users, and compiled some comments from letters to the right. But selling to high schools was difficult since there was a very long lead time (years) from when a teacher decided they wanted something, to getting the funds worked into the school budget.13* Very few people would just mail a check based on an advertisement in a magazine, and he found that he spent a lot of time corresponding back and forth with potential customers. He came up with a marketing plan where he would send a Kenbak-1 to a potential buyer to try out for 2 weeks. At the end of that time, they would either mail it back, or send him the check. This worked well as the cost of shipping the computer "wasn't much" he said.
John focused most of his sales efforts on the educational market, going to teacher conferences, and advertising in educational journals like College Management and Nations Schools, but he did also get some advertisements in hobby computer newsletters like Computer World. Still, his Scientific American advertisement was his most successful (and most expensive.) He did get an article written in the November 1971 "Amateur Computer Society" newsletter which described the machine, instructions, and plans for a paper card reader and other extensions in the future. John also explained why he wouldn't offer a "kit" which some hobbyists wanted.14*
This Amateur Computer Society newsletter published several updates on Kenbak Corporation's plans in later issues. The March 1972 issue announced that plans for the punch-card interface were being "shelved" in favor of a cassette tape interface. The same issue announces that the "Theory of Operation" manual was available for $10, and suggested a hobbyist could make their own Kenbak-1 from the schematics if they wished to do so. Kenbak Corp might even sell a fully built logic circuit board without case, power supply, lights or switches for $450.15* It seems John was reaching out to the hobby computer market, and they seemed interested, but very few actually purchased one.16*
Some User Reactions to the KENBAK-1 Computer
"Thank you for the opportunity of using this excellent educational device. I have taken the KENBAK-1 computer to high schools, grade schools, and used it in my own college course. It is a great motivational tool and exhibits most all computer fundamentals. In particular, I felt that its simplicity and excellent choice of instructions were bonuses . . . May I have your permission to feature the KENBAK-1 computer in an article? I believe that such a machine is ideal for education."
Computer Science Professor, University
"I want to compliment you on both the sophistication and the reliability of the KENBAK-1 computer. For the serious student, it is a great way to learn about machine/assembly-language programming and about the logical organization of a computer."
Mathematics Department Chairman, Private School
"I was extremely pleased at the progress of my students working with the KENBAK-1 computer. Concepts that we had only talked about came alive to them. The one thing that pleased us the most was the definite excitement that the students exhibited in using the computer. They enjoyed working with it and took their own time to develop more fully their understanding of the concepts." Business Instructor, Junior College
"I spent the last two years working with a Sigma 5/7 computer. It is a little surprising to find a small beast that operates essentially like the large ones. The fact that the students deal essentially in machine and assembly language when working with the KENBAK-1 undoubtedly gives them considerable insight into how a digital computer operates."
Mathematics Teacher, Private School
"The computer is a fine instructional device for classroom purposes with students beginning the study of computers and/or data processing."
Business Instructor, Junior College
"Received my KENBAK-1 computer in excellent condition and overall I have been very pleased with it. I have found it rather easy to write programs for it, due largely to the extensive instruction set."
Private Owner
"The KENBAK-1 computer has been evaluated by qualified personnel from a number of agencies here. It has been identified as an excellent inexpensive tool for teaching at the beginning level and would serve to supplement a course in Introductory Data Processing."
Director, Career Development Center, US Army
[The above comments were compiled January 27, 1973, from letters in the files.]
Here is a newspaper photo of John Blankenbaker teaching a computer class to mathematically gifted 5th and 6th grade students in a special after school program. John received a great article printed in the News-Chronicle (Thousand Oaks, California) · Sun, Jan 16, 1972. The newspaper calls it "the world's first elementary school computer class."
The article explains that the focus was teaching students about binary, octal and hexadecimal numbers, which are an important first step to machine code programming. It's not clear how successful this was, but an article in a local newspaper was great publicity.
A Few Problems with the Kenbak-1:
A Few Problems with the Kenbak-1:
While users were generally impressed with the Kenbak-1, there were some problems. First, the computers sometimes experienced intermittent failures, apparently related to overheating. Larry Page, a teacher at Syracuse, NY high school was an early user and recalls the Kenbak-1's were used for about 5 years, until they "melted down." He described intermittent failures, where the lights would stop responding.17* Robert Nielsen, who used eight of them in his South Carolina technical school, noticed similar failures, but with his electronic background he recognized this was due to overheating. Robert drilled holes in the top cases of all eight of his machines, and the problem was solved. John had considered heat dissipation when he was designing the computer: He estimated the power dissipated by the 132 IC's would be around 60 watts, so placed a 60-watt light bulb in the case overnight to see how hot it would get.18* It got quite hot, so he added a fan. Unfortunately, the fan's location moved little air in and out of the case, so heat would still build up. One computer (Nielsen #1) was modified with a more efficient external muffin fan which was much more effective.
Another common problem was the pushbutton switches. John used Raytheon keyboard switches which were economical, low bounce, and had a very smooth action. The problem was that they were designed to be soldered bottom down on a PC board rather than glued by their top to a panel. Glues of the early 1970's didn't hold up well, and many switches broke off if pushed too hard.19* Nielsen told how they replaced the pushbuttons with momentary toggle switches in one computer (Nielsen #2). Most, if not all extant machines, had to have buttons re-glued at some point.
A third problem was harder to fix. The Kenbak-1's limited memory and I/O. With only 256 bytes of memory, 8 input switches, and 8 output lamps, and no ability to "save" a long program, the ability to do serious work was affected. John described his plans of producing a paper card reader in the Amateur Computer Society newsletter (that's why he included the slot in the front panel) but found it was difficult to make reliable. He then announced plans to make a cassette tape recorder instead, but several months later said that plan was "shelved", as it wasn't really necessary for the educational users. One user, Tom Crosley wrote in the Amateur Computer Society newsletter that he built a TTY interface, expanded the memory with a page register, and was even thinking about making a magnetic tape interface.20* But these limitations were acceptable to users because there really wasn't any similar products for under $5000.
In the end, it wasn't these problems and limitations that doomed the Kenbak-1. It was the painfully slow sales when there was no time to waste. Technology was advancing quickly in the 1970's. Intel's 4004 microprocessor was introduced in late 1971, quickly followed by the Intel 8008 and 8080. New memory chips were also coming to market quickly. When the Kenbak-1 hit the market, there was no competition with similar capability and price, but announcements of new technology showed amazing computers were coming soon. Perhaps if John would have sold 200+ computers that first year, he could have hired more help with manufacturing, marketing, and office work, so he could focus on what he knew best: designing computers. If brisk sales allowed John to modify and improve the computer, perhaps he could have designed a "Kenbak-2" with newer technology and kept ahead of the competition. It is easy to imagine the Kenbak Corporation giving some real competition to the MITS people who made the Altair 8800, or the Palo Alto dropouts who made the Apple-1.21* But that wasn't to be. Painfully slow sales meant John didn't have the resources to continue to innovate, and keep his lead against the competition.
The End of the Kenbak Corporation:
The End of the Kenbak Corporation:
With slow sales into early 1973, John realized he couldn't remain in business. In the Felsenstein interview John lamented "we didn’t choose the right market to market to, we didn’t choose the right market, we should have emphasized the private individual and we should have hit these popular magazines... ...Some people thought maybe I should have made a kit, I never agreed with that, it was too difficult but at $750 that was in the reach of private individuals and if they wanted a time payment plan, I could talk to them." John had reached out to the hobby market in the March 1972 Amateur Computer Society newsletter with offers for the schematics and a cheaper logic board only computer, but that was probably too little, too late. By February 1973, John announced in the ACS newsletter that plans for the cassette tape interface had been shelved, feeling it wasn't necessary for educational users. He also admitted that only half a dozen private individuals (non-educators) had purchased a Kenbak-1, and only half of those (about three?) were programmers or electrical engineers. John also reported the price was increasing to $850.22* It's likely, with this public disclosure of dismal sales numbers, he was already thinking about an exit strategy.
After selling just over 40 machines by mid-197323*, the Kenbak Corporation closed down. John sold all design rights, extra parts, and remaining computers to CTI Educational Products in South Carolina. CTI was well known for making electronic and logic trainers for technical schools and junior colleges, but their old products were becoming obsolete, and they needed something new to offer schools. CTI initially put foil "CTI" stickers over the Kenbak-1 name on the front panel, and then fabricated at least 4 new front panels with the CTI logo silk screened on front. These new panels omitted the "slot" in the front panel. According to a note in the Amateur Computer Society newsletter in November 1974, the price of the "CTI Model 5050 Trainer"24* was increased to $1035 by that time. By all accounts, CTI never made a single computer themselves from scratch, and only finished and sold the few remaining machines John had already produced. John thought that CTI sold some computers to a school in Canada who he had previously sold to, but Robert Nielsen, the owner of Nielsen Technical Institute, maintained that CTI never sold any computers other than the ones his school brought. Robert was likely correct, as he worked closely with the CTI marketing team in developing training videos and documentation for the computers.23* John said CTI never fulfilled all their financial promises, and eventually mailed him back a bunch of old parts in lieu of the last payment. CTI soon followed the Kenbak Corporation into bankruptcy.26*
After Kenbak Corporation:
After Kenbak Corporation:
After the Kenbak Corporation folded in 1973, John was back in the job market.
He first joined a small startup called International Communication Sciences, which was started by a person John knew from his Scantlin Electronics job. They had voice digitizers (vocoders) which could turn analog voice signals into digital data at only 2400 baud (2400 bits per second.) They were developing a system to multiplex 4 different voice signals onto a single telephone line at 9600 baud. With the high price of telephone lines at the time, putting 4 voices on a single line could provide considerable cost savings. John designed a high speed "vector" computer operating at 8 MHz cycle time, which could calculate the Fast Fourier Transforms in real time needed to digitize the speech. His computer was a good working design - the problem was in the vocoding algorithm - the 2400 baud voice coding wasn't very clear. "Sometimes it came out great, sometimes it came out like Mickey Mouse or Donald Duck," he said in his oral history. John was proud of his design, but that couldn't fix the problems with the poor quality of the voice. After seven years, the company went bankrupt, but before it did, Robert Adams, the Vice President invited him to join a new business venture.
Symbolics, Inc:
Symbolics, Inc:
Robert Adams told John about a startup that he was forming with a friend, Russell Noftsker. The business, known as Symbolics, wanted to commercially produce a new computer created at Massachusetts's Institute of Technology's Artificial Intelligence (AI) Laboratory. This new computer was a specialized "Lisp machine" based on the original CADR Lisp machine. Robert Adams became the President and Chairman of the Board, joined by Russell Noftsker, John, and several others from the AI laboratory at MIT. John was one of the founding employees, and named their "Director of Engineering." When he joined in 1980, they didn't even have an office - John worked in Robert Adams's home in an enclosed patio.27* This project became the Symbolics LM-2. He spent a year turning the complex wire-wrapped prototype into a commercially producible machine. This was a time when the FCC was enforcing strict EMI interference standards, as it was proving a challenge to keep radio frequency noise within the computer case.
The LM-2 computer (Lisp machine 2) was a powerful single user workstation with high-speed bit-mapped graphics, based on a specialized Lisp machine architecture. The terminal is shown pictured. While the terminal appears small, it was accompanied by washing-machine sized processor and a large disk drive, and drew over 40 amps of current at 120 volts in the smallest configuration. This was quite successful - Symbolics built and sold about 80 of these machines before introducing the improved Symbolics 3600 computer line.
Despite this success, there was a lot of conflict in those early years at Symbolics. Robert Adams, the first president, was pushed out by the staff at the MIT AI lab after only one year. Montgomery Phister, John's old friend from Hughes and Scantlin, was an investor, and was called in to help mediate the management shakeup.28* Robert Adams was replaced by Russell Noftsker, but the conflicts didn't abate. John was disappointed to see many years later that the Symbolics Website completely removed all mention of Robert Adams, their first president.29* John left in 1983 due to ongoing conflict, but as a founding employee, he left with significant company stock benefits, which had appreciated due to the success of the LM-2.
Moving back East: Post Technical Work:
Moving back East: Post Technical Work:
After Symbolics, John went to Quotron, the new name of Scantlin Electronics. When the founder, Jack Scantlin was pushed out in a big management shakeup, the new president, Milton Mohr, renamed the company after their most popular product.30* John became the "Manager of Communications" where he worked for 2 years managing some 13,000 phone lines during the turbulent divestiture of AT&T.
By 1985, John's in-laws were getting older, and his wife wanted to be closer to them. They moved back east to Pennsylvania. The "investment bug" had hit him while he worked for Scantlin, so he wrote a small-distribution investment newsletter named A Non-Random Walk which he edited for more than 7 years. He did a little engineering work, including working for a small company "Science Products for the Blind" founded by a blind physicist, Dr. Thomas Blendham.31* He started tutoring students at nearby Lincoln University and was soon asked to teach part time. Before long he was teaching physics full time and math part time. He taught at Lincoln University until about 2002.
It was during his retirement that the Boston Computer Museum began soliciting entries from its "Early Personal Computer Search" in 1984. John wasn't actively following the hobby computer industry but his old friend, Montgomery Phister, saw the announcements and encouraged John to enter his Kenbak-1.
John developed a new passion in his retirement. Starting in 1989, he began writing and editing a bimonthly newsletter journal entitled Beyond Germanna, which researched the history and genealogy of the ancestors of the German colony in Virginia. That became a great love of his. He wrote or edited the newsletter for 15 years, inspiring many to study and further research this topic. By some real detective work, he wrote the definitive early history of how and where these early settlers lived, synthesizing information from original documents, oral tradition, and geographical features. He made a big impression in the Germanna Studies community, and was honored by the Germanna Foundation as an inspiring researcher who taught many historians to follow.
As of early 2023, John and his wife Eleanor are living in New Hampshire, and just celebrated his 93rd birthday on December 24, 2022. While he admits he's not as active as he once was, and some names from 50 years ago are becoming harder to recall, he is still enjoying his large extended family and staying in contact with friends all over the world by email.
Footnotes:
Footnotes:
1) John's high school yearbook seems to describe him as either very thin or agile, it's not clear which (Albany Union High School yearbook, pg. 37, available on classmates.com)
2) Oregon State University yearbooks show several photos of john, including one in the school play "Room Service" (1949, page 48.) (https://oregondigital.org/sets/osu-yearbooks)
3) When John told about his early interest in computers, he usually tells of reading a magazine about a computer with "thousands of vacuum tubes" and numbers represented by only "ones and zeros." We searched for a contemporary article in several years of "Popular Science," "Popular Mechanics," and even "Smithsonian" magazine, but failed to find it. All these magazines had computer articles, but nothing matching his inspiring article. The pictured May 1949 "Popular Science" article was at the end of his freshman year, and seemed close, but John is certain this is not it. He likely had access to many magazines at the college library, or perhaps his memory could have merged together two different articles. But the exact inspirational article seems "lost to time."
4) John recently credited Floyd Steel (the founder of Computer Research Corporation, see Wikipedia entry,) as an inspiration on his single flip-flop computer idea.
5) This story of an executive forecasting a very limited market for a general-purpose computer reminds us of the legend of Thomas Watson, the president of IBM, who reportedly said in 1943 "I think there is a world market for maybe five computers." A contemporary record of either of these quotes is lacking.
6) This photograph of John shows him smiling ear to ear, really enjoying the moment. Locating this photo was worth the hours of searching. Thanks to Anne Killheffer, and Stacy at UNLV libraries for finding this!
7) This article is copyrighted so it can't be included here, but it shows a much more serious photo of John, only a year or two after the Hughes News photo.
8) This article is also copyrighted so can't be included here, but it can be found on the internet for free with a little searching.
9) John's statement that the machine was "independently discovered" by the Amdahls seemed suspect. Since John was working under these two powerful personalities, it's easy to imagine that he told his idea and the Amdahls dismissed it suggesting it's trivial, and they already had thought of it. It is also easy to imagine that John felt pressured to publish his idea quickly that summer to "stake his claim" and prove it was a non-trivial and important idea. However, when asked about this, John had nothing but good things to say about the Amdahls, and denied any competitiveness or animosity.
10) Yes, MIT mis-spelled his name in the thesis catalog. After John's description of his interactions with Bernard Widrow, I wondered if Widrow really wasn't too impressed with John's thesis, or thought it was marginal. I reached out to Dr. Widrow (sixty years later) to see what he thought in retrospect. Bernard Widrow said "I think it’s a fine piece of work. It’s well written and clear in its method and analysis. There are some things in the thesis related to a course that I was teaching at that time. So I probably had some influence on the thesis, but it is highly original and the whole approach is his and not mine. It is not simple stuff. If published today, it would be new work and publishable." That's a pretty positive review from one of the greatest pioneers of digital signal processing.
11) The story of Scantlin Electronics and Milton Mohr is interesting, and a great example how a company fell into decline but was turned around. The LA Times published the story in 2000 available here https://www.latimes.com/archives/la-xpm-2000-jul-24-me-58233-story.html
12) John took great price that he didn't ask any strangers to invest, only people who knew him and his capabilities personally. The Felsenstein interview listed the investors, but misspelled many names. They should be correct here.
13) John blames the long budget cycle for schools for the slow educational sales, but I wonder if that was just an excuse the schools gave him, and propose an alternate reason in the FAQs section suggesting his real competition were time-sharing systems which offered higher level languages, like BASIC and COBOL.
14) The November 1971 article explained why John didn't want to sell a kit Kenbak-1. But he did suggest perhaps the company could host an event where people came in on weekend to assemble one "under supervision" over several Saturdays.
15) Only 4 months after dismissing the idea of a kit Kenbak-1, John suggested ACS members could purchase the schematics for $10 and make their own computer from scratch. The article says the 131 integrated circuits would cost about $94.75 by mail order. This would be unreasonably difficult without a PC board, and the user would still have to supply lights, switches, power supply, and a case. It's doubtful anyone took this seriously. But it may have gotten people to consider buying the assembled PC board (logic board) for $450. Note that John doesn't actually offer the assembled PC board, he just says they were considering doing so, and readers could write to him if interested.
16) John confessed in many biographies that he didn't reach out to the individual, or hobby user enough. There's no denying that the hobby market was the driving force for the PC revolution to follow, but I do think he reached out quite well, with his Computer World advertisements and the letters in the Amateur Computer Society newsletter. I suggest in the FAQs section that hesitancy of the individual buyer may be more from the limited details in the advertisements.
17) Larry Page didn't realize the failures were heat related, but thought in in retrospect that it made sense. Still, the Kenbak-1's remained in use as part of their computer science program for 5 years, until they got replaced by a microprocessor simulation program. Machine or Assembly code is rarely a part of high-school curricula since "BASIC" became available.
18) John's estimate of 60-watts of heat dissipation wasn't far off, but in reality, under than 15 watts was radiated by the logic PC board, and the remaining was dissipated by the linear power supply. Regardless, the fan did little to move heat outside the case.
19) Most if not all of the original Kenbak-1's have had pushbuttons break off. Repairing wasn't easy without getting glue on the sliding plunger, making the button stick.
20) It's not clear how much of Tom Crosley's modifications in letters to the ACS were functional or planned modifications. He definitely got a TTY interface working, and made a simple program allowing text editing through the teletype. He also told me he got the memory page register working. But in spite of him reaching out to ACS members to find other Kenbak-1 hackers, he says he never got any replies. Perhaps John's February 1973 announcement suggesting very few engineers or programmers purchased a Kenbak-1 is the reason. Tom's computer still exists, in our database of remaining original machines.
21) With John's experience and determination, it's easy to imagine what could have been if John focused his talents on re-designing and improving computers, rather than struggling with sales, marketing, and manufacturing, all by himself. He hard far more digital design experience than most of the computer builders of the 1970's.
22) This February 1973 announcement signals the end was coming. John withdrew his prior promise of a card reader, and the cassette interface that was promised in its place. He also he turned his back firmly away from the hobby market, doubling down on his commitment to the educational market. He explains how only about 6 computers had been purchased by private individuals, and only half of those (3?) were purchased by programmers or electrical engineers. He justifies withdrawal from the technical/hobby users, and focusing on the educational market. I suspect he realized he couldn't stay in business by this time.
23) The exact number of Kenbak-1's built and sold is not clear. This is addressed in excruciating detail in the FAQ (frequently asked questions) section of this website.
24) Many sources say CTI renamed the Kenbak-1 the "H-5050" but this is not true. H-5050 was the part number of the Laboratory Exercises Manual, and P-5050 was the part number of the Programming Reference Manual. The official computer name was "Digital Computer Systems Trainer Model 5050."
25) Robert Nielsen was a colorful person who may have exaggerated some of his role and importance, but he undeniably had a close relationship with CTI, and assisted their marketing department by making training videos. He described his touring of the CTI factory, and their preparations for manufacturing the Kenbak-1. So his statements that CTI never made their own Kenbak-1 out of scratch, and his statement that he doesn't think CTI sold a single Kenbak-1 except to his school, must be taken seriously.
26) John felt that CTI never lived up to their end of the bargain, failing to pay some money due. Possibly they promised to pay for some parts he shipped them, or perhaps they promised to pay a final amount after they got their manufacturing process working. But after a lot of excuses and delays, John received a large box of left over computer parts in leu of payment.
27) John said in the Felsenstein interview that his initial office was in Robert Adam's basement but he told me it was definitely in an enclosed patio, in Robert's home.
28) Montgomery Phister is a recurring name, and remained a lifelong friend. John met him at Hughes Aircraft working on the business computer project, and John referenced one of his books in his thesis work at MIT. It was Montgomery who recruited John to Scantlin Electronics, and then Montgomery was one of the investors in the Kenbak Corporation. This last story explains how Montgomery was an early investor and mediator for Symbolics Corporation. A short biography of Montgomery Phister is here: https://www.amazon.com/stores/author/B001KCNKDQ/about When the Boston Computer Museum had the "First PC" contest, it was Montgomery who saw the announcement, and told John he should enter the Kenbak-1.
29) Even today, the Symbolics Wikipedia page (https://en.wikipedia.org/wiki/Symbolics) fails to acknowledge their first president, Robert Adams, except for a footnote hinting otherwise. The Wikipedia page on the Lisp machine also talks about Noftsker starting Symbolics, but never mentioned Robert Adams (https://en.wikipedia.org/wiki/Lisp_machine). This bothered John.
30) As mentioned previously, the story of Milton Mohr's turnaround of Scantlin Electronics is a great example of making some of the difficult changes to save a company. The LA Times published the story here: https://www.latimes.com/archives/la-xpm-2000-jul-24-me-58233-story.html
31) This interesting company got its start by making speaking calculators that could be used by bind people, but during John's time they worked on cash registers, glucometers, and blood pressure machines. John told amazing stories about how the blind founder, Dr. Thomas Blendham, managed to do extraordinary things in spite of his disability. This is in his oral history by Felsenstein.
Google Sites
Report abuse