Difference between revisions of "First-Hand:The Development of Pong: Early Days of Atari and the Video Game Industry"
m (First-Hand:The Early Days of Atari and the Video Game Industry moved to First-Hand:The Development of Pong: Early Days of Atari and the Video Game Industry)
Revision as of 16:24, 13 November 2009
The Start of Syzygy, the predecessor to Atari
Nolan Bushnell impressed me. He had a company car. He had a Buick station wagon. “What's a company car?”, I asked. And he told me “it's a car you can drive and you don't have to pay for it.” “Wow! That's a concept I could go for”, I thought to myself. And he offered me a thousand dollars a month salary, and ten percent of the stock in the company. You know, I was already making $1,200 a month at Ampex as an associate engineer. And ten percent of the stock seemed worthless to me. At the time, I could care less. It was a nice token, but not that important to me. I accepted the offer because I thought that it would be fun. I had seen advancement inside the company ranks as kind of a slow process, through your whole career. But I had seen people who had left and come back would get bumped up higher.
Being at Ampex, I worked in engineering but I liked talking to the guys in manufacturing, in purchasing, and with the guys in personnel. But at big company like Ampex, you only had your little purview and you'd hand off your stuff to a draftsman or something. Working in a small company on the other hand offered a broad view of everything. I figured it would probably fail in a year or two, and then I'd go back to Ampex. I knew they'd hire me back. I was a young guy.
We came from the 60s, and that's important. We learned from the Vietnam War not to trust the government, not to trust authority, so there was this automatic kind of distrust and a skepticism and questioning of authority. Questioning the status quo was a fairly common thing to do for our generation. So I figured, “What the heck. I'll take this risk.” After all at the time we lived with the threat of nuclear Armageddon. The Cold War was going on. They were building bomb shelters. So all of us who started Atari was very young. We were in our 20s. Looking at the bigger picture, why not risk. We also contracted with Bob Herbert, one of the great hardware digital engineers at Ampex.
Nolan was a very convincing person and crazy. But it fit within my style of “What the hell, let's try it.” Taking risk was something, to have the courage to buck the establishment, to argue against the war in Vietnam was extremely unpopular, so taking unpopular positions became second nature. You know, I think a lot of it came from that. I think that was important. Why was it done on the West Coast? Why was it done by young people? You have a lot of insight and intelligence and not so much practical experience that you don't know it can't be done, you know.
Nolan had agreed to make a pinball machine for Bally. This used the old relays and coils, and none of us knew relay logic. I mean, that was really primitive stuff. But it's a subclass of logic just like TTL, and this older engineer at Ampex knew it and so he loved doing it. Bob would say, “For the first time my son knows what I do; I'm building a pinball machine, and did that logic.” But he would never quit a good job. In fact, when Nolan left Ampex, Charlie Steinberg, who was the head of that division, had a talk with him. It was unusual. But he felt, “Son, you're making a big mistake here quitting Ampex doing this thing, you know, you really ought to reconsider; you're giving up a career here.”
Compared to the lab that we had at Ampex, at Atari, it was a very humble, one old oscilloscope. The big thing that got me was Nolan’s use of the TV set. He had used a General Electric, one of the last of the vacuum tube television sets. I didn't like that set. And Nolan use this weird fiberglass cabinet. It was an eye-opener to see that you could start a company with such simple stuff, and that you could make something that was engaging, which was really a copy of the old Space War game--the original Computer Space. It was fascinating. To see how you could make this with no computer was intriguing. In fact this was Nolan’s important breakthrough. I had never seen a video game on computer while I was at Cal Berkeley. Bu Nolan went to school at the University of Utah, which is one of the few places that actually had a PDP1 and Space War running. He had seen it. I had never seen a video game on a computer at Cal. But Nolan decided to do it without using a computer.
In his original concept Nolan assumed that he needed a computer because all video games at that time--the one--was on a computer. But the more he designed he realized that all that the computer was doing was keeping track of “is it going this way or that way”, and with a little bit of headwork you could do that with some gates and get rid of the whole computer, and save a whole lot of money. So that was his breakthrough. There was also another fundamental circuit that he created - the motion circuit. The idea was to make this spot, this ball, address all the different locations on the screen, when there is no memory. It was a very clever trick using a variation of a synch generator circuit.
Ted Dabney, Nolan, and I were all engineers. But Nolan did more of the business stuff, Ted handled the hardware and the facilities stuff and the route. We had a route. And I was the engineer.
Our start-up had only something like $500 paid-in capital. We did not have investors neither did we have a bank line.. But Nolan had some royalties from Computer Space. He used that money to buy pinball machines and electro-mechanical driving games. We'd buy them through a distributor in San Francisco, Advanced Automatic. We had locations in the area, bars and restaurants, in which we put these machines. The money that collected from them provided are cash flow. Literally quarters were coming in to help pay our way. We had to go and collect the quarters from those machines. We were walking around with bags of coins. In fact, we would go out and collect, sometimes at a restaurant, and pay for lunch with quarters that we had just collected for the machine, and so it was quite a learning experience. That experience also taught me about designing things work in a public environment. If that machine steals your dime or your quarter you feel you have a right to destroy the machine. So it has to be really well built, but yet cheap. And so I learned a lot about that.
Originally our mission, our charter at Syzygy, which then became Atari because we couldn’t use the name anymore, was to be an engineering consulting firm and receive royalty checks. We had a contract from Bally, the dominate company building arcade games in those days, to build one video game for them, a pinball machine, and some other arcade device. So we were getting a few thousand dollars a month from them to create these products.
We didn't really interact with the Bally people. They just sent us checks and Nolan kept bullshitting them and talking to them and made them happy. Nolan was good at that. And he was also designing a 2-player Computer Space for Nutting, which they never built, while I was doing this Pong thing. So interacting with them really wasn't the issue. A more interesting interaction was with the people that actually ran pinball machines and routes for a living, because we were on their territory. These were “Godfather” types. They did show much interest in us. But that changed when Pong started to go.
And in the early days of Atari, Nolan asked me to go to a dinner at Advanced Automatic and represent Atari. So my wife and I went. We were sitting at a table with another couple, but older. We introduced ourselves. He said, “Oh, you're from Atari.” And he then reached into his pocket and pulled out a pistol, put it on the table and said, “You know, you're operating in my territory.” And I replied: “Well, we'll stop that, you know.” I got back to Nolan the next day, I said, “He pulled a gun on me, Nolan.” Nolan replied, “Oh, he's just joking; don't worry; he's not going to kill you.” I said, “I'm not going to anymore of those things; I'm an engineer. You go to those things.” The banks wouldn't loan us any money, “Oh, you're in the arcade game business.” Arcades, that's the mob. And so they wouldn't talk to us. They accused us of being part of the Mafia. “Hey, make me a bribe; make me an offer”. I would have joined up, but no one ever offered me anything. So I had to fight the stigma but never got any of the benefits.
Creation of Pong
Nolan gave me this challenge. He tasked me to make a ping-pong game on a TV screen and told me he had a contract from General Electric to do a home video game. And so it had to be very, very cheap, very inexpensive. It had to have a $15 cost of goods. My initial reaction was, “Whoa, it's going to be tough.” On the other hand the game was very simple, just one moving spot. Later on I found out he had no such contract. He actually thought that the game we were going to make was to be more complex than Computer Space, perhaps a driving game or something like that.
The first thing Nolan did was gave me a copy of the schematic of Computer Space so I could reuse some of his designs. But he drew schematics in such a bizarre way that I couldn't really ever understand them. But with a little chalk talk on the board, he explained the fundamental trick of how you do the motion circuit. Basically, we used a synch generator generating the background synch for a vertical and horizontal frame If you then had another digital synch generator going right along with the same clock, they'd be in different spots, and if you looked at where the horizontal and vertical started on the second generator and made a spot on a video display driven by the first synch generator, that would be a fixed dot. It you turned the power off and powered it back up, the spot would appear again, but it would be in some different random spot.
So what if you had the second synch generator, and let's say, you changed the vertical counter by one less count, it would move, it would move up one line every frame, or if you made it one more count, it would move down. So there was vertical speed. So it was just two sets of counters. With eleven or twelve ICs, we had a synch generator with a moving spot. It could move anywhere on the screen and if you did it using a frame buffer you would need many thousands of bytes of memory, which was unacceptable at that time. So that was the main trick I learned.
Nolan had told me that he had this contract with General Electric for a consumer product. The fact that nobody from GE ever wrote us a letter, called or came by, should have set off alarm bells. But it never occurred to me that this might not have been true. I was too busy doing the project. I felt that I was failing because, just in making the spot, I had already used something like fifteen, twenty parts. Eventually when it was all finished there were something like seventy-five chips in it. I used the 7400 series TTL ICs. They were running about an average of fifty cents to twenty-five cents a chip. Some were cheaper…some were more expensive. The other thing about them that's important was the failure rate which was about 1% or 2%,. So every time you'd build a board with seventy-five ICs, chances are you'd have a bad chip. Troubleshooting them and make them work in production was important.
The game had one spot, one little object, score digits, fixed things like the net, and then there was sound. Those were all the objects. For the score, Fairchild had a really interesting MOS chip that would do digital scores. It was a complex and expensive chip. Instead of the decoding all the segments of a seven segment digit, you could use this chip and get all the digits up there easily, and so I put that part in the prototype. But that was not acceptable because the part cost $5 and we couldn't afford that, so I had to go back to the drawing board and de-code every segment of a 7-segment display using brute force TTL and all the tools I learned at Cal.
Nolan said that it had to have sound. But I was already over budget. I was finished with it except for the sound. So Nolan said, “I want to have the sound cheering people when you make a score.” And Ted said, “I want it to have boos and hisses.” I didn't know how to either one with digital components. I was already over budget. Then one afternoon I poked around the synch generator and found tones that were already there, and that's where the sounds came from, and that became the sounds. Articles have been written about how wonderful the sounds were. And I told Nolan, “if you don't like the sounds, you go do something better, but I'm only doing that. And so that's how the sounds got out. But it was because of cost…I was over budget.
In my efforts to design a real product, I believe that I did some innovative work. The motion circuit was very creative, very clever. So was the power supply. The first Pongs that we built in production had an open frame, 5-volt, external power supply, which you could buy for like twenty, thirty bucks. The power supply had a voltage adjust potentiometer . When this digital video game got into the hands of route operators, who were used to fixing pinball machines with a big screwdriver, a big file and a hammer, it caused problems. If the game wasn't loud enough, if it wasn't making enough money, they just turn all the knobs until it got better. Well, they'd turn the knob on the power supply and burn the whole board out. We couldn't have that.
The LM309 had just come out--I believe it's the 309; it was a 5-volt power supply on a chip. And it had no adjustment. But, the problem was that my board drew one and a half amps at five volts, and the 309 was only good for one amp. This led to one of the innovations you don't really know about. Rectifiers and a capacitor were on the board with the 309. I put a five ohm resistor around the 309 to bleed current around it. I ran about an amp through the resistor at average, so the 309 only had to regulate about half an amp. Now normally you can't do this because if the load goes away you'd now be running current through the regulator backwards and blow it out. But because it was all on a circuit board and you could never disconnect the load, it would work. I heard people say, “You can't do that.” But I did And so I saved a lot of money with this innovation. It was a very simple circuit; just one resistor. But it saved about a dollar worth of other parts.
Since Nolan told me that it had to be a consumer product, I reasoned that the game would be uninteresting if the ball just had one horizontal speed. Two guys could sit there and play it forever. And that would not be much fun. So I added the speed-up. There was a counter in there. After a certain number of volleys it would go faster and then faster yet. And that was one feature that I added.
Another good feature was actually a defect in the design. I used the 555 timer as the control for the paddles; to make the paddles go up and down. It fired at vertical synch and its duration was controlled by the player control. But because limitation in the chip, the paddles would not go all the way up to the top of the screen. We discovered that that was a benefit. By not being able to go all the way to the top, a good player could make the ball go up there and get it past his opponent. Without this “feature” two good players could play forever. This would have ruined the game. Hence my motto: If you can't fix it, call it a feature. So that became the feature.
There was the effect of the paddles imparting “english” on the ball. Since the paddles were 16 scan lines high. I figured if it hit the top of the paddle it would have the highest vertical velocity; I set the vertical velocity by where on where the ball hit the paddle. All right. But if you do that, you want to have the velocities symmetrical and I wanted to have a flat shot straight across, and I wanted to have three speeds up and three speeds down. This made seven vertical speeds and this did not map easily into the 16 scan lines I had for the paddle.
So, I thought and thought about this, and how to divvy it up. Basically I had two horizontal that went across, and two in the middle would give no vertical, and then so it became symmetrical. But to do all that math I wound up having to put in a 4-bit adder chip, which is one of the more expensive chips I had in the design. A lot of logic design was needed to get these angles just right and translate them into the vertical speeds.
I wasn't really doing angles. I had a horizontal speed which was controlled by how many times it hit the paddle, and then vertical speeds controlled by where it hit on the paddle. So the angles would change but I didn't care. So it was really about horizontal and vertical speeds.
The project took, from a standing start, about three months to have this done. I used parts I had left over from Ampex and a few other parts that we could buy. From the original goal’s perspective, I saw it as a failure. But Nolan never told me it was just a throwaway. He just wanted to start me on a simple project that I could finish in order to get the hang of the process. In his mind, a real product was never the intended outcome. I on the other hand thought that we were designing a real product. A fate would have it, circumstances transformed my project into the commercial product called Pong.
Nolan was going to make this great video game, and the original goal was to give something to Bally. We owed Bally a video game, and the next generation Computer Space to Nutting. But Computer Space was not a big hit in the marketplace, and so it wasn't a big, important deal. But all of a sudden I had made Pong playable, but Nolan still didn't tell me there was no GE contract or anything. Then one day he told me, “You know, this is kind of a playable game. Let's put it at one of our locations and see if anybody plays it. It might be fun; it might be a real game.” So I said, “Okay. Just as a trial; just to see how the public accepts it.”
Over the weekend somebody built this cabinet--a tabletop cabinet, very simple. And early on I had gotten a Hitachi transistor television from Wal-Mart or Walgreens, for $75 I hacked it, because of my experience fixing TVs, and turned it into a monitor. Back in those days you could buy a monitor from Motorola. But those monitors had fuzzy green screens that were in airports and they were never working, and they cost $500 apiece. I could do a TV set for $75, and it had an audio amplifier and speaker so I could do sound off of it too. I turned the antenna terminal strip--the antennae terminal on the back that had four screws on it--to ground, video and sound input. You had to reach around to adjust the volume control. We put this thing in a cabinet, and placed it in our favorite location. Our best location was a place called Andy Caps Tavern, not far from our offices where we already had about five or six pinball machines and a Computer Space running there. And we set it on the top of a barrel near Computer Space.
Pong was an unusual video game. There were two knobs and a coin box. Each knob turned a potentiometer. When you think back on it, nothing like this had ever been done before. There were no instructions, and it required two people. It was the only video game that ever required two people--no one-player mode; it had to be two people. But it was also a quarter play. What were the plays on the other pinball machines then? A quarter plays? No, they were ten cents. A quarter would get you three plays. So one of the things--one of the reasons Pong was so successful was the fact that it earned quarters, and the operators loved it because they had revenue, they had big revenue all of a sudden. It was a popular game. We broke the ten cent barrier with that machine.
In fact, the first Pong stopped working in a few weeks because the pots wore out. I was surprised but I did the math. I figured how many coins were in there. Everybody made at least 15 or 20 turns every time they put a quarter in. When multiplied by the number of quarters in the box, the result astounded me. “My goodness”, I told myself, “this is getting to a hundred thousand turns in a month”. And we're exceeding the lifetime of cheap pots. So I had to go to Allen Bradley and get a long-life, expensive potentiometer that would last six months on location. That was important. It's funny because looking on it philosophically, that was one of the first digital devices ever to be in the hands of people…certainly the first digital video device ever. And the interface was a knob and no instructions.
The game had no instructions, nothing, and yet it was a big success. There had been nothing like it. All of a sudden the quarters started coming in, and it made big money. Within the first two weeks of it being tested in Andy Caps I get this call from the owner who says, “Hey, it crapped out; it stopped working. Could you come down and fix it? I got people who want to play this thing.” People were showing up just to play this machine. So I went down there. One of the tricks I had learned was to give a free game, when you were behind the machine working on it. That kept all the customers distracted because very few of them had any idea how many quarters were in that machine. There might be a hundred bucks of quarters in that machine. So I'd give a free play.
Well, I opened up the coin box to do this, and the coin box gushed out quarters. I figured, Oooo, when I started putting them in a bag, I said, “I can fix this”. I remember the next day, told Nolan what happened. He said, “Hmm, that's interesting.” So then Nolan got the entrepreneurial idea to go and change the whole mission of Atari: We're going to build machines; we're going to be a manufacturer.
Atari changes its mission
At this point the company had about four or five people and we're going to do manufacturing. Ted said, “No, I don't want to do that; we're engineers, and we want to just do brilliant designs and receive checks from people building them. There's more money in doing this.” Nolan kind of twisted our arm, and instead of saying yes, he just went ahead with it anyway.
Then he went back to Bally, realizing that we had a hit on our hands, and we could make more money selling this machine than we could giving a license to Bally. So he had to go back to Bally and convince them that they didn't want this machine. By now we'd built a few machines. We had some minimal production; we could build maybe two or three machines a day. The word got out fast in that industry, and Nolan succeeded in convincing Bally Midway that they didn't want and convinced us to get in the manufacturing business. With the experience of the pistol and a few other things, we realized that we wanted to be the manufacturers; that's where the money was going to be. We didn't want to do the route operations. Oh, we had our own route and that was fun.
We then sold Pong to distributors, which is another interesting point that Nolan had, from a business standpoint. In so many industries there were three distributors in every major market, and they competed along different product lines. You know, GE, Hotpoint, or whatever. And here it was historically based on jukeboxes: Seeburg, Rockola and Williams, or whoever made jukeboxes. And so we could only sell to one-third of the distributor base out there. There was a real incentive for copiers--people to copy, and fit the other channel. And Nolan solved this eventually by creating our own competitor called Key Games. It was Nolan's brilliant business sense. His next door neighbor, Joe Keenan, who was a very intelligent, sound professional businessman, which was a breath of fresh air compared to our crazy company, started this other company called Key Games and we gave him one of our engineers, our circuits, our designs.
In fact, if you looked inside the machines, it said Atari on the board in some of their machines. But the distributors and the operators loved it. I mean, they love. And we would go out and say, “Those bastards stole our machine and they stole our design.” And they distributors would say, “Yeah, great; we're going to go for them.” We knew they'd run over there and buy from them because they thought they were getting around us. But we owned Key Games. Any buyer could find that out in the California records. Nolan and I were on the board of Key Games. Nolan was the chairman of our competitor, but the operators loved it. They didn't want to look. They wanted it to be this way because they figured they could play us off against each other. In fact, in a few trade shows there would be a Key Games booth and an Atari booth, and we'd yell, “You bastards,” at the Key Games people.
Pong went into a lot of locations, places where pinball games could not go. Pong had very subdued graphics. There were no naked ladies, no explosions, etc. So it went into locations where you couldn't get a pinball machine in. Which was important for our revenues. That was also Nolan's part of the design.
We also spent some time thinking of the design of the cabinet. Nolan was very creative in that regard, and the first Computer Space--the first video coin-out video game--had this bizarre fiberglass sculptured cabinet. He had done a beautiful job with Computer Space, but he learned the hard way when he used a fiberglass. He wanted something that looked different and attracting attention, and indeed Computer Space was indeed one of the most unique cabinets ever made for an arcade machine. But because it was fiberglass, the production method was very slow. We had to build one big mold, which could make one cabinet per day, maybe, per tool. But if you wanted to build fifty machines a day you had to have fifty tools. It cost a lot of money. It took a lot of time. It was very slow. So he learned…. Now, unfortunately Computer Space never got to the volume where it became a big problem, but if we had done that with Pong we would have been in deep trouble. We could never have made enough to supply the customer base. So he learned to go with the straight wood cabinet, simple to build, and fortunately there was a vendor in the Bay area, P.S. Hurlbut, that had the ability to manufacture this sort of stuff in volumes and became our cabinet supplier.
Around the beginning of 1973, we realized that, with Pong, we had a big hit on our hands. Now it was a race. The machine was popular and people copied it, stole the idea. So it was a race to the market between us and other manufacturers. We made over a million dollars in revenue our first year, which was phenomenal. Here I am a 24-years old ex-hippy, out of Berkeley, an anarchist, if anything, and I'm doing this on a lark figuring it will collapse and I'll go back and work at Ampex. We had revenue of a million dollars. Whoa! “This is weird; this is good”, I told myself. I didn't believe it was going to last very long. So my attitude was one of “Let's just see how far it can go.”
Pong moves from the arcade to the home
Nolan was convinced that Atari was going to be a billion dollar company. At the time he seemed crazy. We had no money. We had no manufacturing capacity. In mid-1973 Nolan issued a memo to the Engineering Department, which was effectively me. Nolan wrote: “In accordance with a concise business plan, Engineering will have extra teams of engineers to do projects, build a booth for the MOA trade show, build a multi-player arcade used in some kind of a county fair-type of game.” Item 6 on the list was a home version of Pong. And a few other items after that, and at the end he says, “And questions to the effect that we don't have the manufacturing capacity are not appropriate to this discussion.” I sent a memo, from Engineering, back to Nolan saying: “Is the fact that we have no money germane to this discussion?” And he wrote back on the memo: “No!”
Nolan’s strategy for large growth was to make Pong a home video game. The best you could do in coin-op was compete with Bally, who was the dominant guy. The best you could do was to capture 100% of the coin-op marketplace, which was a very finite marketplace. But if we could take this into the home, there was a lot of money to be made. Magnavox already had the Odyssey game, and they had made a home game, but they did not market it well. It was not a fun game. But the concept was there. If we could make Pong into a home game, which had been the original goal that he'd lied to me about with Pong in the first place. I knew the only way that we were going to be able to make a home game would be to get the cost down to under $20, and the only way we would be able to do that would to make a custom chip.
By the time we started to get serious about the home market, we had many games in manufacturing and production. We had a fairly good engineering team--maybe 20 people--, and I had never done a custom digital chip, not many people had, there was no such word as ASIC; it wasn't invented then. I hired an engineer, Harold Lee, who knew chip design. He knew logic design. He knew the whole field. He was a very interesting guy--very bright guy. I had hired Harold to make a custom chip that we could put in arcade video games so as to make it impossible for people to copy our products.
Our competitors didn't know how our game worked. They would just take the circuit board, copy it, put parts in it, and compete with us. I wanted to make it harder to copy. Well, I learned something, which strikes me as almost axiomatic. Any security technology you put in that is sufficiently good to keep people from copying your game, will also make it impossible for you to build the game. I haven't done the research to see if there's an axiom there, but it seems like it is.
This was getting nowhere because the technology was evolving in the coin-op arcade so fast. Harold said, “Al, it will take about nine months to a year to do a chip, and by the time you got the chip finished you're technology would have changed; I don't know if it's going to be useful or not and it's going to cost too much money. But I think I can get your simplest game, Pong, on a chip.” And I said, “Cool, we can do that.” So I pitched it to Nolan, “Let's go do this”. He agreed and I went off with Harold Lee to design the chip. I hired Bob Brown to do the test program, and my wife helped wire-wrapped the design and we made it work. We rented time on an Applicon CAD design terminal and had a Pong chip in about six months.
Well, finding a manufacturer was not a simple matter.. Nobody was doing custom chips for other people, ASICs basics weren't involved. At the time Silicon Valley was full of semi-conductor factories. Ironically there are very few of today. AMD, Sygnetics, there were a lot of companies around. We got the design rules from about four of these companies, and we distilled a universal design rule that would work with any of them. Now, Carver Meade calls it Lambda design rules, but we did it on our own. We made one set, and we owned our own tool. People did not do that then. If you wanted a custom chip you would go to a semiconductor company and their engineers would design it, and you have to buy it from just that company. And we, for some reason, didn't do that; we designed our own chip; we had our own front end; we owned the artwork. Our first vendor, our supplier, who agreed to do it for us was AMI. Initially, we thought maybe 50,000 - 100,000 a year was a reasonable estimate of the production volume. We really didn't know. The chip ran at 3.5 megahertz. This was back in 1973, '74. CMOS was very, very slow; it was used in watches and calculators.
We used N-Channel, not P-Channel. It was faster, but it didn't have Implant so it required plus 12, minus 5 for body bias, and 5 volts for the logic. And it was fast. But would it work? No one had ever made a consumer part with N-Channel and we were running at 3.58 megahertz for the master clock on this thing. And the question was, would it work at those speeds? Up to that point consumer applications, like calculators, had only been for speeds in the one hundred kilohertz region. When we got the first chip back, the first prototype, we plugged it in and fired it up on the oscilloscope and it was working. I had on a variable oscillator that I could run it up ten megahertz. The chip ran well. “Wow!”, we told ourselves, “we did it.”
Testing was an important challenge the chip design. I knew that I didn't know what I was doing. At Cal we had discussed analog chip design and I knew the concept of silicon-integrated planar circuits and how you make them. The details I didn't know, and testing was like, Huh! Fortunately, I had hired Harold Lee who had experience on this, and he knew lot of stuff that I didn't. Testing is the big deal. You can make a chip but if you can't test it in less than three seconds it's not a production part, and in those the tester was known, it was the Fairchild Sentry tester. A key member of the team, Bob Brown, who was an experienced guy, knew how to write test programs for the Sentry. So the test program was started when we started the chip. It was important because a chip had to be designed to be testable I didn’t have any experience building a chip but my small team had that experience. I had the luck to hire the right guys that knew how to do it. We spent a lot of time building that test program, which we basically handed to AMI. So they had the test program; they had the wafers, the Calma tapes…
There were a couple of bugs in it but we knew that we had a successful design. We knew the hard part was going to work. And then we asked ourselves, “now what next? We had no plans. We didn't have a business plan, you know, I just did the engineering, to see if could make this work or not. Meanwhile, Steve Bristow was running coin-op and that was creating all our revenue. And I was off doing this. One of the guys said, “a consumer product”. There was no consumer work being done in the Valley here. So we said, Sears Roebuck; that’s a consumer company--big company and they have that big building in Chicago, right? So we called Sears Tower in Chicago. I mean, can you imagine. We got through to a lady, the operator. She looked through the catalog and found that they were selling the Magnavox Odyssey in the sporting goods department, Department 606.
She passed us through to the buyer, Tom Quinn, who happened to be there. “We're this company, Atari,” we said “We have this thing called Pong.” Now, this buyer happened to be the guy that was selling the Magnavox Odyssey at Sears. Magnavox should have owned this business. Right? But they didn’t. They were the old-line. They were manufacturing vacuum tube TVs, television with huge cathode ray tubes in big wooden cabinets; that was their mission. They didn't care about the fact that people wanted these little Japanese sets. It was a fluke that they were selling a videogame.
The Odyssey video game was just an excuse to sell televisions. Magnavox wanted the customers to think that their video game only worked on a Magnavox set, even though it would work on any set. So they only sold it in their retail stores. And Sears convinced them to let them sell it but only in the Sears catalog, not the Sears stores because they felt it would compete against Magnavox dealers. Tom knew this was a big market there, but he was frustrated.
Now we show up and “Hey you can have this.” So Tom response, “Sounds interesting. The next time I'm in California I'll stop in and see you guys.” Three days later he's on our doorstep eight o'clock in the morning, at our factory in Los Gatos. Now, none of us showed up at eight o'clock in the morning! He was in a suit and tie because he was a businessman at Sears, you know--professional. None of us were in a suit and tie. I was the first one in about nine o'clock, and Tom probably wondered, “What the heck is this? What have I gotten into with all these young people? There was nobody over thirty in the whole place. It was crazy. And there were all these video games on the factory floor--arcade games. We then showed him the home Pong prototype.
Sears and Pong was a cultural mismatch, but we clearly had the product that he wanted, and now he had to sell it into Sears. But that was his problem…convincing the management at Sears that these crazy guys from California who had never done a consumer product, let alone a consumer video game, could come through. Could this small company get past all the FCC rules, all the testing, all the safety, all these other issues? We learned by trial and error.
Nolan didn’t want to have Sears as our only customer. “We want to this to sell this to other manufacturers.” Sole-sourcing to Sears could be risky. “If Sears shuts us down we're going to be in horrible shape”, was Nolan’s fear. So Tom said, “Well, no, we're really good, you should go with us.” Instead Nolan decides that we should take it to the Toy Fair in New York. In 1974, February I believe it was, we took Pong to the Toy Fair in New York City. Having never been to the Toy Fair, not knowing the toy business at all; we figured that would be the marketplace, like any other normal trade show. Wrong! Toy Fair is not that way. The majority of the business takes place about a week before the Toy Fair for the public. And it's by invitation only; you can't get in; it's not free market. So we go to the Toy Fair with a booth next to a guy selling magic sets and Crazy Glue, and here we had Pong--consumer Pong--that would be the hit of that next two or three years.
All the buyers from all the department stores--Macy's, all of the top guys--came by and saw it. But we did not sell one unit. Because, as it turned out, in that industry all the buying was done for Christmas, the week before we arrived at the Toy Fair. Just minor products were left and the buyers were uninspired. No one would take that risk. So now we were forced to being sole-sourced at Sears. I remember Tom came by and said, How's it going? “Oh, it's going great”, we replied. But here we were in New York City; we were having dinner by ourselves, there was no interest… nothing. We hadn’t done our homework. What was marketing? Nolan's attitude was, “let's just throw it out there, we'll learn, fast. Getting to market is more important than getting it right the first time. We'll just keep our eyes open as you go in, make mistakes and react quickly to those mistakes.” That was our approach.
If we had had anybody in our company with experience in the buying and selling of toys and games, he would have said, “That's not how you do the Toy Fair. You go to the Toy Fair, you rent an office in the Toy Building for a whole year--you have to rent it--and you only use it for two weeks of the year. But even then you have to go to your potential customers and invite them.” So that industry is still to this day set up to exclude innovative new people – to the detriment of their business. For example, Mattel, in their booth, when they went to the Game Show, the E3 show, they had a big booth at one point and they were the only player that you could not get in to see unless you had an invitation to the booth. And I asked them, “You mean, you spend like a quarter million dollars for this huge display and you're only letting your existing customers in? Why?” They would reply, “Well, we have secrets in here.” I said, “Really? Can you tell me what the big secret you had last year was?” They had no idea. I believe that secrets can be a very destructive, damaging force. It prevents you from getting valuable criticism that could improve your product.
But people talk about how important trade show response is. Here we had the hit product of the next few years and we couldn't sell one of them at New York Toy Fair. What does that really mean? Was it a problem with the product or a problem with the industry? You know, so it's a really important lesson about what people will buy.
Because of the television, we had to deal with FCC. As it turned our, FCC compliance was one of our biggest challenges. I got into the project I thought if I could get a custom chip to do Pong, everything else will be easy. It's the only miracle that had to happen. The chip was the hardest thing; get it done and all you have to do is just make a plastic case. It's all tried and true. But that is not how it turned out.
The FCC regulatory environment was almost impossible. The history goes back to 1972 when Magnavox came out with the first electronic device to hook to a TV antenna terminal besides the antenna, the Magnavox Odyssey game. It was back in the day before televisions had analog input jacks. In fact, most of the chasses made in those days were AC-hot just for cost purposes. So you didn't want to connect anything to the chassis ground. You would get a shock off the metal chasses. We had to go through the antenna terminals. If you just hooked it to the antenna terminals and someone left the outside antenna still attached you could interfere with neighbor’s reception.
So the FCC wisely made rules under Part XV of How to Construct and Build a Device that would limit how much radiation a TV device could. I believe that the rules were created with the Magnavox Odyssey product in mind. Odyssey was essentially an analog device while Pong was the first truly high speed digital consumer device. The rules stated, I don't remember the details, that above 10 megahertz the allowable radiation was dropped to about one microvolt per meter at two meters from the device, which is at the limit of detection of the best spectrum analyzer you could get at that time. Pong had a 3.58 megahertz crystal oscillator created by a TTL chip that had very fast rise and fall times that generated lots of energy up to and beyond 100 megahertz. It radiated like stink and it became quite a challenge to suppress without raising the cost too much.
Eventually, with Sears help, we got the thing to comply with those very severe regulations. But a few years later, when personal computers, none of which were in compliance with these FCC regulations, were coming on the market they caused a lot of interference and many consumers complained. The FCC began a process to change these rules to make them more appropriate for digital devices. We were against the changing the rules because, even though the rules were very difficult for a digital device to meet, we knew how to make to do it and that was an important advantage for us. In fact, there was a request for comments, from the FCC, for this proposed rule-making, and they requested a personal computer from Apple, TI, Atari and Commodore. And, and we, being really clever, submitted our computer to, and got type approval. In those days you gave it to the FCC and they would do the testing back in Maryland. We got approval. So we thought that was a very nice submission. We didn't actually submit it to that committee. Instead, we said, “Here it is. We've gotten type approval.” The question facing the committee was whether this could be done or not. We proved that it could be done. In the final report, it became a footnote that Atari submitted a unit that met the regulations. They just ignored it and came to the conclusion that it couldn't be done. So they changed the regulations. “What happened?” was my reaction.
Now, once we got the chip to work, we just build as many as we could. Now we're interfacing with semiconductor companies. We had our own chip, but the guys from the semiconductor companies told me later it was a terrible chip. And they kept telling me, “Al, we could redesign this part out and get the yield way up.” Semiconductor companies are in the business selling chips to manufacturers, so they sell them for maybe a hundred percent markup, or whatever their markup is on the thing, and they're happy. The way they make money is to drive their costs down. But we were in the other end of the business. We were a manufacturer that was selling these video games--these home consumer Pongs, for like 70 or 80 bucks. Out cost to goods manufacture on these things was like twenty bucks. We had a huge margin in this thing. Getting them out in the numbers was more important than stopping, for three to four months, to redesign it. That would have been crazy. We could have shaved $2--we were paying like three to five bucks for the chips; what did we care, you know. The semiconductor people kept telling me that it was a bad design. “Screw it”, was my response. “You can make them. Just make them.” Synertec, our best supplier actually modified its manufacturing process to work better with our crappy design. They were smart enough and got their yields up that way. The normal way is to fix the design to match the process. They changed the process. They were smart; they were one of the best vendors out there.
There have been several misconceptions associated with Pong. The one that gets me is the popular belief that there is a microprocessor computer involved. Everybody assumes it. They want to know what was the code like for Pong. There is no code because there really was no microprocessor. The 4004 had come out, and it was an interesting device, but it certainly couldn't do anything at the video speeds that we had. So these were all just digital circuits hard-wired to make Pong. It's a digital machine that did Pong and nothing else. And, it's just amazing to me today how hard it is for people to conceive--today's engineers--that you could build anything without a computer or code involved.
R&D at Atari
In the early days of Atari, here was never a business or R&D plan. We were all engineers, okay. Once we got successful with arcade Pong, we were so busy keeping Pong in production that I couldn't do any new designs. I mean, it was like engineering ground to a halt. And yet, we still had to worry about everybody copying our machines. The only way we really could defend ourselves was to come up with another better product that the copiers didn't have. So how was I going to do this and keep us in production at the same time? I came out of being an associate engineer at the bottom of the heap. I was not a manager. I had no leadership experience. So I learned by trial and error.
Our original R&D team was a couple of ex-Ampex engineering friends of ours who quit Ampex before we did, Larry Emmons and Steve Mayer. They had moved to Grass Valley, CA to work for a company called Arvin Systems. Grass Valley was a picturesque gold mining town in the Sierra’s only an hour away from great skiing. Arvin closed within a year of them joining the firm. Larry and Steve then struck out on their own and started an engineering and consulting company called Cyan Engineering. So Larry and Steve had this little engineering operation looking for work and Nolan said, “Let's hire these guys. They’ve set up a little office up in the old Litton engineering building, Mountaintop, in Grass Valley. We'll make them our R&D group, to do advanced engineering.” I was very young. I had to be dragged fuming, kicking to do it. But it was the best thing we ever did.
Initially, I took offense to this. I was as a 25-year-old engineer. I figured I could do everything, and knew everything. Are you going to hire somebody that's smarter than you to work for you? You might look bad. You know, it took some getting used to. I got used to it and I learned. I soon realized that as we grew bigger, people resented those guys in research up there in this ivory tower, “they don't know the real problems”. I had an airplane and would fly up about once or twice a month and meet with them, and thank God we had them. Because all we could do were minor variations on Pong, whereas one of the first things they figured out how to do the driving game.
The idea for the driving game had come out of an article in Scientific American, in the mathematical puzzles section. There was a little paper game of a racing game. You had a grid and you started out with one and the next move you could move two, and you could change your vector. Hey, that's how you drive. The next thing I knew they had a prototype of this driving game--Grand Track--that was a big advance, a big quantum advance. There were times when Atari was on credit hold, and couldn't get parts. So the Grass Valley group used their own account, Cyan Engineering. They paid their bills; we didn't always pay our bills.
We first tried to be market-driven or customer-driven in our R&D efforts. Customers told us, especially the European customers, they wanted soccer. So we defined this complex game with four or six potentiometers, and we gave it to them. We did what they said. We couldn't sell one of them. So we decided to heck with those guys. We were on the edge of technology of what you could do, which is a lingering story, in video entertainment electronics.
Instead we just developed ideas that we thought would work. Nolan had his own list of twenty great ideas. A few of them are really good ideas--a lot of them are crap. Once every few months, we would meet as a group here at Pajaro Dunes, in the Monterey Bay or at Grass Valley, and make a list of things we thought would be fun to build. And so we'd work on them. And that's how R&D came about.
Products after Pong
The first major one was the VCS cartridge game, probably with a premier--the best product--one of the best products of the 70s, and that was a real breakthrough. Steve Mayer and Ron Milner, in Grass Valley, created the VCS prototype. And I found out recently that they did it in about three months, the whole architecture. It was really brilliant. It was a totally different from where everybody else was going. And then we did the Atari computer 800. But that wasn't as big a breakthrough from a technological perspective. And then the last thing I did, once the Warner guys kicked in, was a holographic toy, which was really an interesting challenge technically. We wound up inventing embossed holograms.
Atari enters the home computer market
Steve Jobs worked for me at Atari back in '73, '74. He had this crazy idea for a home computer. And we helped him out. We introduced him to venture capital, and vendors etc. We didn't think it was going to be any kind of success or anything like that. And, gosh, it became a good idea—a great success. But that really wasn't our expertise, building computers. We could do it, and eventually, under the Warner…. Well, actually we started on our own. We, the engineers, wanted to do it because we already knew how to do custom chips. None of these other personal computer manufacturers could do custom chips, and we were pretty good at this. And so we built two special purpose chips for this computer. The microprocessor was 6502. We built a TIA, television interface adaptor, chip to do the graphics. We built a thing called Pokey, a keyboard and external interface, which was built to run intentionally slow so it wouldn't create much interference. The keyboard and all those things did emit a lot of radiation. So we used our skill and experience to build the only personal computer that ever complied with the early FCC rules. And then they changed the rules on us, damn it. And then we had no advantage. Our plan was to be the only personal computer allowed to be sold. That would have been something. We were the only ones that met the regs. And as I said before, the FCC decided that it couldn't be met.
Warner buys Atari
Warner bought us in November 1976. Our revenues were around 28 - 30 million dollars a year at that point in time. We were selling consumer stuff, but the VCS wasn't out until summer of '77. Nobody knew how big it was going to be until afterwards. And then sales were flat until about 1980, when all of a sudden the VCS took off, and by '81 it was up around a billion dollars in revenue. Despite this enormous jump in revenues, I decided that it was time to leave the company. The reason had to do with corporate culture, something that they don’t teach in schools
When companies take over another company, they often fail, because these fat, gray-haired men on the board don't understand culture, and they just see the business relationship. They try to merge two companies together and if the cultures don't work, it's never going to work. And this happens time and time again. It seems great on the spreadsheet or paper.
Warner was a traditional, proper business running--normal business. In Silicon Valley, what we were doing was not normal business. It was a very hit-driven technology business. You had to eat your own young because if you didn't, somebody else would. The new management team did not understand this. Instead they ran Atari like a big, bureaucratic organization that supported it current products. You know, it's funny. They bought us because we created unique products and we built a whole industry, and then a couple of years later they saw us as troublemakers who did not know what we are doing. So they decide to put in professionals, who came out of traditional management…Proctor & Gamble-style of stuff, where they're used to having products that last forever.
They saw potential in the VCS and decided to really spend a lot of money for a marketing campaign. Here was a big difference where we were wrong. We would have scraped the VCS probably in 1980 and had another replacement product. But they didn't. They just put the advertising behind it and made it a hit…it was a huge hit…it was enormous…it was colossal. But in its wake, all invention had failed. All new stuff was dead. And it happens in many, many big companies. When you get a big company that's kind of a monopoly position, they've got to spend money on R&D because that's what big companies do, and if you don't spend the money on R&D the stock market won't like you because where is your future? But on the other hand, these big companies are the last to want a new revolutionary product because it will change the order of things, and the order of things is what keeps them rich.
So the good news is that working for a big company gives you lots of money to do R&D; the bad news is you better not really want to see a product ever get out or impact the marketplace. And that's what happened to me. I like to build new stuff and new things. Once Warner kicked in I realized that they weren’t going to build and ship new products. Nolan was gone, and Joe was gone. I realized that if I wanted to build something new I was going to have to do it myself. So while still with Warner, I decided to turn back into an engineer, leave my mahogany row office, go back into the engineering building, and put a team together of the brightest designers I could find at the company, which I did, just three guys.
In pursuit of a holographic game
The idea was to work on holographic stuff. Back in the early days we could not keep a secret at Atari. People were spying at us all the time and all our plans would come out. Instead of trying to stop talking - because we wanted to talk, I mean, you can't be creative if you're totally secret – our solution was to put out disinformation. So we would put out lies and rumors about ourselves that would hopefully distract the competition. One of Nolan’s favorite stories was that our next game was based on holography even though we had nothing going on in that area.
Since Nolan and the old team were no longer around I decided to return to engineering and come up with a great new product. I decided to try holography because it was new and different and since we had often said we were going to do it why not do it now. The main marketing advantage was that it could ½ the price of an equivalent cartridge game. Holography was technology with which I had little experience. It produced weird images, and I had studied it a bit at Cal Berkley so I understood the concepts. So we went ahead and did it.
There was no initial product concept other than “Let's do something holography.” This, of course, was the wrong way to do a product. This is absolutely "ass-backwards". But I had Harry Jenkins and Roger Hector, some of the best product designers I knew. I had several types of holograms made that we could use for mock-ups. We asked ourselves “what could be done” and we tried various ideas.
The goal was to use holography in some novel way. And we finally came up with this idea of a cartridge game system whose cartridges cost less than the VCS cartridges and the base unit would also cost half the VCS cost. The problem was how do you get the cost out of cartridge when a cartridge has only one chip in it? The answer was no chips. And it was a simple LED, light emitting diode, game of something like Space Invaders. You looked at it through a 4x5 inch hologram illuminated by one of two light bulbs inside to get two different images out of the hologram.
As I got halfway into this project, I discovered that holograms were, at that time, all made by exposing a light sensitive material laser light reflected off an object. That meant that the holograms would be at least $4 or $5 each. There was no way you could do high volume production this way.
I found some very clever physicists, holographers. One of them thought that we could emboss them. “Oh, that's interesting”, I thought. We just stamped them out. When you use a photographic film you're making light and dark areas on a flat surface. Holograms can be made with phase modulation using differences in thickness or height. People had made phase relief holograms by then but they were harder to make ones with photographic film. But there was this company back in Baltimore Maryland called the Diffraction Company that made circular diffraction grating embossed on Mylar, as jewelry; And so this one holographer, Steve McGrew, said, “I think I can do this, and he had been working on the process on the side had yet to create a hologram on a nickel master that could be used to make sub-masters produce thousands of holograms.
We put a lab together and he prepared some nickel masters. We sent them to Baltimore where they made production sub-masters. Steve and I flew out to Baltimore to watch the first embossed holograms come off the press. They had this homemade funky press about the size of a big Xerox machine that, "kachink", would hot stamp a hologram on a sheet of Mylar in less than a second. The first 4x5 inch holograms that we made were so bright that they hurt your eyes in the sun. I can see them, kachink, kachink, kachink coming off the press. We had just made an important breakthrough. I remember in an IEEE Proceedings, there was a special issue on autostereoscopic imaging This overview basically concluded that it was very expensive to make holograms. I knew I was on to something when I had already done something that Proceedings said really should have been able to have been done.
Well, I ran into another lesson. Management really want to rock the boat that much. So when I proposed this product, they didn't really want to build it, because it was new and different. They were busy building Atari VCSs. First they told me that I had to have a business plan. I replied, “I've never had a business plan before.” So I did a business plan with the help of a guy in marketing, Conrad Judson.
The plan showed that we would lose money the first year, after which we would make money. And the new management said, “We can't do that; we're going to lose money the first year.” So then I thought, “This is stupid.” I wasn't a businessman; I was an engineer, but I knew this was stupid. So I proposed another product--Product X, which lost money for two years. And again they said, “We can't do that.” So I got on the VP of Marketing's phone, in his office and called up Manufacturing. The manufacturing guy was Paul Malloy. I told Paul, “Let's shut down the line building Atari VCS because this guy says we don't want to build it.” That Project X used the historical numbers of the VCS product. “Okay, okay, responded the VP of Marketing. So I got past that hurdle. Then Manufacturing said they couldn't build it because they were too busy building the VCS. I said, “Put that in a memo please.”
I then found an outside vendor who would build it better and cheaper than Manufacturing would. Then Manufacturing decided to build it. There were more excuses. Finally at the end I'd overcome all of them and we had our Cosmos prototypes and we took them to the January CES in Las Vegas, and we ran our own booth; no marketing group wanted to touch it, because why would you want to sell something when you were sold out of VCSs and people were bribing you to sell them more VCSs? Why would you want to actually have to sell something when you didn’t have to?
Eventually Ray Kassar said, “No, we're not going to build it.” That was it. And that's when I left, because I realized that this company was never going to release a new product, and it would surely die in this business. They had Allen Kay, some of the great R&D people in the world, people far smarter than I am, but they weren't going to release anything that was being done; they just let them play around because we have to have R&D. So that's why I left. I had an enforceable, non-compete employment contract with Warner which lasted until 1983. So Nolan, Joe and I had to stay on the beach until 1983--we weren’t allowed to compete. They were paying us handsomely not to show up, so, we called it on the beach.