This post is part of an ongoing series annotating my book They Create Worlds: The Story of the People and Companies That Shaped the Video Game Industry, Vol. I. It covers material found in chapter 5 on pages 70-73 and Chapter 8 on pages 126-134. It is not necessary to have read the book to comprehend and appreciate the post.
Now that we have examined some of the controversial differences in memory between Nolan Bushnell and Ted Dabney, its time to turn to something far less controversial yet still incredibly murky: the sequence of events that led to the creation of Computer Space as a scaled down version of Spacewar! manufactured by Nutting Associates. Unlike some of the controversy surrounding who did what in the creation of the game, the question of when various milestones occurred is not a topic Nolan and Ted ever fought over, yet they still have different recollections of the timeframe of the project. As Computer Space occupies a historically significant place as the first coin-operated video game, its worth taking a moment to figure out to the best of our ability when and how it actually took shape.
First, the basics. The idea for Computer Space was entirely Nolan’s. This is not in dispute. After seeing Spacewar! at Stanford, Nolan decided to recreate the game as a coin-operated amusement running on a minicomputer. He enlisted Ted Dabney and programmer Larry Bryan to help with this project, and they discussed forming a partnership together. Eventually, Syzygy Company was formed sans Bryan, but work on the game ceased when it became clear a minicomputer would not work. At some point, Nolan had the epiphany that they could just recreate the game in hardware, so he and Dabney completed a scaled-down version of Spacewar! missing many key features such as two-player competition and the central star with its gravity well to create what they originally called Cosmic Combat but eventually reached the market as Computer Space. Sometime in early 1971, Bushnell learned of Nutting Associates and pitched the coin-op manufacturer on the game. He then finished the project at Nutting as an independent contractor despite also coming on board as the company’s chief engineer so he and Dabney would retain the rights to the technology. Before the end of 1971, it was released.
While these basic facts have largely gone undisputed, creating a timeline out of them is complicated by a lack of much hard documentary evidence. Before the last decade, the only vague chronological details had to be drawn from the recollections of the participants. Dabney provided an account of the early history to Marty Goldberg and Curt Vendel for their book, Atari: Business is Fun:
The first meeting was held at Larry’s house in October of 1969, where they set about trying to map out a plan for the project. First and foremost was a $350 startup contribution by each that would be placed into a checking account Ted was going to open up. Subsequent meetings were then held at either Nolan’s or Ted’s house, and it was during these meetings that they tried to hash out a name for their new venture. Originally Nolan and Ted had been considering B&D or D&B for Bushnell and Dabney, depending on who felt their name should come first. The problem they found with these though, was that both sounded too similar to the initials for Black & Decker or Dunn & Bradstreet. Larry then suggested the name Syzygy; a name used in astronomy to describe three celestial bodies aligned in a single straight line. […] With that, Syzygy was born.
While this is not a direct quote from Dabney, Marty Goldberg later explained to me in an email correspondence in February 2015 that the timeline for Syzygy and Computer Space as established in the first chapter of their book came from him. Based on Ted’s recollections that Nolan approached Ted “about a year” after starting at Ampex, Business is Fun concluded that the idea for the game and the proposing of the Syzygy partnership occurred in Fall 1969.
While not a point of contention between the partners, Nolan Bushnell provided a different timeframe when interviewed by Morgan Ramsey for Gamers at Work:
Ramsay: When did you talk to [Dabney] about starting a video-game company?
Bushnell: It was about 18 months [after joining Ampex]. I really didn’t talk that much about the video games that I had played in college for awhile. The catalyst was seeing an ad for a $5,000 computer come across my desk. The ad was for a machine called the Data General Nova 800. I thought that was a strong enough computer at a low enough price. If I could build a unique monitor—a cheap monitor because computers in those days cost about $30,000—then I would have a business.
Leaving aside the fact that Nolan did not play any computer games in college as previously discussed in this blog, Nolan gives us both a timeframe, 18 months, and a catalyst, seeing an ad for the Data General Nova. Note that Dabney’s “about a year” and Bushnell’s “18 months,” do not hugely conflict with each other, but they do introduce a discrepancy of roughly six months. So who is right in this case? This time, it appears to be Bushnell.
First of all, even if Dabney is correct about that timeframe of “about a year,” Nolan himself makes clear in his 1974 deposition in the Magnavox patent cases that he graduated college in December 1968 and joined Ampex that same month. Therefore, there is no way he and Dabney could be discussing a partnership under Dabney’s timeframe in Fall 1969. Second, two Atari corporate documents have since emerged: a 1972-73 fiscal year financial statement prepared by Arthur Young, and a prospectus created in early 1975 when Atari was considering going public. The financial statement claims the Syzygy partnership was formed in January 1971, while the prospectus explains it was formed in December 1970. While this does constitute a slight disagreement between the two sources, clearly they establish this event as happening well after Dabney’s claim of Fall 1969.
Bushnell’s deposition further clarifies the situation:
Q: I show you a document which has been marked 39-3 and 39-4 and ask if you can identify those?
A. That’s a listing that came from one of the trade journals, and I don’t remember which one it was, which listed all the mini computers that were on the market at that time, their approximate costs and how fast the cycle time was and what the architecture of the machine was. It was sort of a thing that we went through to see if there wasn’t a cheaper system that we could buy that would do essentially the same thing.
Q. Essentially the same thing as what?
A: The same thing as the Data General unit that we felt probably was as good a buy on the market at the time for what we wanted.
Q. I notice that those two documents bear the dates August 1970. Were these documents that you were considering after the date of January 26, 1971 or prior to that time? [the importance of January 26, 1971 will become clear later in the post]
A. Well, it was prior because, you know, obviously we had made a decision at the time this letter was written as to which computer we wanted and we had been looking at this quite a bit before August 1970 and was [sic] very happy when they published this because it made us evaluate a lot more units.
Q. You said you were looking into it quite a bit before August of 1970. I gather from your prior testimony that all of your activities were during the year of 1970 with respect to the building of this?
A. That is true, in terms of actually putting any hardware together or, you know, drawings.
While this passage is vague on when work actually began on the project, just placing it before August 1970, Bushnell provides more detail later on in the deposition:
Q: When did you first decide that you Wanted to use a raster scan cathode-ray tube display system in that apparatus?
A. Probably it was coincident with the time that I decided to pursue this on an active basis.
Q. What time was that?
A. It was the early spring.
Q: Of 1970?
Now, Bushnell and Dabney’s timelines are adding up a little more. If, as a thought exercise rather than a statement of fact, we assume “early spring” to mean April, then this would be 16 months after Bushnell joined Ampex, which is also fairly close to the vague estimate of “about a year” given by Dabney. While not entirely on point, this seems reasonably within a margin of error when dealing with trying to remember a series of events over forty years after the fact. When coupled with the documents indicating Syzygy was established around December 1970-January 1971, the totality of the evidence indicates that Business is Fun is in error.
Even if 1969 is the wrong year, could Dabney be remembering events correctly, but just be off by a year? Quite possibly. After Business Is Fun was published, Marty Goldberg tracked down the third person who was originally supposed to be part of the Syzygy partnership, Larry Bryan. In an interview with Marty conducted in March 2015, Larry said the following about the timeframe for his participation in the project:
Well, you can sort of find out by sub…it was like six months or so before Nutting and Nolan got together
While this is also a vague recollection dredged up forty years after the fact just like Dabney’s false memories of 1969, it does have the added benefit of fitting the documented timeline quite well. In his 1974 deposition, Bushnell discusses his departure from Ampex and his move to Nutting Associates:
Q. When did you commence your employment with Nutting Associates?
A. I think it was in March or April of ’71.
Counting back six months from March 1971 gets us back to September 1970 for Bryan’s involvement, right around when Bushnell would have seen those August 1970 sell sheets for Data General computers. This also happens to be the Fall, the exact season Dabney remembers all these events happening. While most of these remain estimates based on recollections, its amazing how nicely they all play with each other once we assume Dabney is merely off by a year.
Now that we finally have a timeframe for when the Syzygy partnership was proposed and then formed, we can now turn to the actual work on the game. For this, we must turn once again to Bushnell’s deposition, in which several early schematics for what became Computer Space were presented into evidence. While these were sadly not dated, Bushnell did attempt to establish when they were created:
Q. Do you know when you drew these documents which I just enumerated?
A. I’d say it was probably around July or August 1970. It might have been as early as February for some of them, but I think the ones that I drew in February were rougher. These are more detailed as to interconnections.
Once again setting aside that Bushnell is likely taking credit for some work actually done by Dabney, this July-August timeframe for sketching out a system also appears plausible and actually fits really nicely with Bushnell’s later recollection that he started talking up doing the game around 18 months after he joined Ampex. This also fits well with Bryan’s recollections of being recruited in the early Fall, because both Bushnell and Dabney agree that he was a later addition to the project. It makes sense that Bushnell and Dabney would first make sure they could even interface a monitor with a Nova in the way they wanted before asking a programming buddy whether he could create a timeshared version of the game. Absent contradictory evidence, we will take Nolan’s word on this one for now.
So if work started in Summer 1970, when did the first iteration of this project end in failure? Bushnell provided this part of the story to Tristan Donovan for his book, Replay:
By Thanksgiving 1970, Bushnell concluded the project was doomed to failure. “I got frustrated and decided to abandon it,” said Bushnell. “But I kept worrying about the problem and thinking about it and then I had that ‘a-ha’ moment where I thought I’m going to get rid of the computer and do it all in hardware. From that point, it just flew together.”
While Donovan did not quote Bushnell directly in the book on the date, this lines up with what Bushnell said in later interviews, such as his oral history in 2017 with the Smithsonian:
This was over the Thanksgiving holidays. My idea was I was going to finish the design over the Thanksgiving holidays. I started literally Wednesday night and got up on Thanksgiving Day, worked all morning. Before dinner at 3:00, I abandoned the project. I say, “Time’s not right. Math doesn’t work.” I don’t know whether it was the tryptophan of the turkey dinner or what have you, but you always do this after Thanksgiving dinner, you always lay down and have a nap. During my nap, my hind mind came up with the solution.
Screw the minicomputer. Get rid of it. Do it all in hardware. Make the game out of this collection, just make it a simple state machine. And the minute that happened, it was like knife through butter. Not only did I get the cost down, but what was budgeted for $1,500 worth of minicomputer, the whole damn computer cost me less than $300 in glue parts. So, I knew that I had something.
This makes for a good story, but it sadly appears not to be accurate. Just about the only documentary evidence we have in this entire mess is a letter that Nolan Bushnell penned to a salesman at Data General dated January 26, 1971, in which he planned to order several Data General computers. Back to the deposition:
Q. Just to try to fix that date as firmly as possible, Mr. Bushnell, I show you Atari Exhibit 39 about which you did give some testimony in January. A part of that exhibit is Page 39-2, a letter from you dated January 26, 1971 to Bob Washburn at Data General. At this point in time had you made this decision to drop the software approach or not?
A. No. At this point we had not completely ruled out that we would use the software approach.
Bushnell provided more detail as to what changed his mind about using a computer later in the deposition:
Q. I hand you Atari Exhibit 40 and ask you if you can identify Document 40-1?
A. This is a letter from Bob Washburn who was the sales engineer in the area for Data General. We had kind of been stringing him along because we weren’t ready to commit the dollars and we had sort of told him, “Yes, the order is coming. The order is coming.” I think this letter is to just sort of jack us up and trying to push us into a close. It was during this period that I had pretty much decided that I was not going to go the direct computer route but was going to go to a single stand-alone unit.
Q. During what period was this that you just referred to?
A. Between the time of composing that letter to the time that I got that. Because it was–I was almost ready to go but I just wanted to go back and check to make sure that the system as I had configured it made sense. I wanted to make sure the thing was doable, and so I wanted to get closer– I had found a place where I could rent a Data General computer and I had gotten a little bit closer to a guy that was there who was trying to sell me some time on the machine. He pointed out something that I had failed to take into consideration on my initial calculations and it scared me into thinking that maybe I wasn’t even going to be able to get four monitors to go. So at that I point I decided that I really needed to change one of my design at that time and that pushed me into the thinking of just doing it all hardware and not doing it software with the computer.
Q. And the period during which this occurred that you are referring to was the period between the dates of 39-2 and 40-1?
Q. What was the date on 40-1?
A. February 16, 1971.
So according to Bushnell’s testimony, a computer was not totally abandoned until sometime between January 26 and February 16, 1971, dates fixed by actual documentary evidence. So how does this track with Bushnell’s Thanksgiving recollection as well as the simple fact that Bryan had warned them off a computer solution months before this? Well, Bushnell has claimed in several interviews that they initially tried to solve the computer problem by moving some functions to specialized hardware to take some strain off the main machine. From his Smithsonian oral history:
I started working on the interface logic, the thing that would take the raw data from the minicomputer and give it in video form to the monitors. I just kept running out of time. There was just not enough compute power. But my solution for that was I would take over certain aspects and just do it in hardware, because another thing about right timing is the TTL Caterpillar chips that were the logic building blocks of flipflops, AND gates, OR gates, decoders, what have you, had just dropped in price by two orders of magnitude. What used to be fifty bucks to do a flip-flop was now fifteen cents. Talk about right timing. It was really great. Doing little circuits using digital logic and Boolean algebra, it was trivial. I mean, it was very cost effective.
The first thing I did is I took over the star field, which was part of the space, the background. Then I did the score, in which I just decoded a certain segment of the screen. Maybe I should back up. I started out knowing that the computer had to be crystal-controlled so that we would have a clock speed that was equivalent to the horizontal position of the raster, and so I determined that we would have an active screen area of 256-by-256, 256 pixels horizontally, 256 pixels vertically. So that just turned out to be eight bits. [Laughs.] So, I had eight-bit counters horizontally, eight-bit counters vertically, and it just worked. It worked out.
So perhaps augmenting rather than replacing the computer was his Thanksgiving epiphany? At this point we are well into the realm of speculation, but its as good a guess as any. Obviously, its also possible his Thanksgiving recollection is simply incorrect.
From here, the rest of the timeline becomes much simpler. Bushnell has explained in several interviews that he made contact with Nutting because he talked about the project with his dentist, who connected him to another patient of his, Nutting sales manager Dave Ralstin. As previously indicated, he joined Nutting in March or April 1971. According to his deposition, Bushnell and Dabney had a working prototype of the spot generating and motion hardware even before the January 1971 letter to Bob Washburn, and he further indicated in his deposition that there was a rudimentary playable version of Computer Space running by “April or May of ’71.” Cash Box confirms the game debuted at the Music Operators’ of America (MOA) trade show on October 15, 1971, and the first ad for the game appeared in the November 27 edition of the trade publication. This dovetails well with Bushnell’s recollection in his deposition that the game actually entered production in December. And thus was the video game industry born (well, not quite, but that’s another story).
One final strange wrinkle. In 1983, a magazine called Video Review featured a timeline of early video game history. Its an interesting document because its clear that the author must have conducted some interviews and examined some documents rather than just regurgitating what had been previously recounted in the press. This is most evident in the section on Ralph Baer and his Brown Box, which clearly draws from primary source material held by Baer that was not generally made available until decades later. For our purposes here, the most interesting claim is as follows:
On August 23, 1971, Bushnell signs an agreement with Nutting licensing Computer Space from Syzygy.
This is a claim I have never seen anywhere before. It could be incorrect, but having a specific day like this lends a certain amount of confidence to the claim. Could it be that Nutting did not actually agree to do the game when they hired Bushnell? It makes some sense seeing as all parties agreed Computer Space was a project independent of Bushnell’s employment with the company. Perhaps Bill Nutting wanted to see how the game turned out before committing to it. As Business Is Fun and Benj Edwards’ article on Computer Space for Technologizer indicate, the game existed in a mostly completed state by the Summer. This date does make sense if Nutting was hesitant to embrace video games until he could actually see a living, breathing example of one. More corroboration is needed before I take this claim at face value, but it is interesting nonetheless.
So where does that leave us? Below is a rough timeline of how Computer Space came together. Note that unlike my similar Spacewar! timeline, this one contains a lot more theorizing and speculation. I would not take every entry as gospel, but its the best we have that reconciles the various recollections and documentary evidence. If nothing else, it gives us a rough sequence of events.
December 1968: Nolan Bushnell graduates from the University of Utah with a degree in electrical engineering and is hired by Ampex Corporation as a junior engineer in its Videofile division. He is assigned an office with Ted Dabney.
Spring 1970: Nolan Bushnell sees Spacewar! for the first time at the Stanford Artificial Intelligence Laboratory and starts pondering how to commercialize the game.
Summer 1970: Nolan enlists Ted Dabney to help make his Spacewar! coin-op game a reality. They begin sketching out a monitor interface and other components and evaluate the available computers on the market, settling on the Data General Nova.
Fall 1970: Needing a programmer, Nolan turns to his friend and co-worker Larry Bryan. A meeting is held at Ted Dabney’s house in which the three discuss forming a partnership named Syzygy to create the game. Larry spends a little time looking at the problem of timesharing three or four games of Spacewar! on a single Nova computer — a necessity for the venture to be profitable — and determines it cannot be done. The partnership falls apart.
November 1970: Nolan has an epiphany, possibly over Thanksgiving weekend, that the project may still be viable if he moves some of the game functionality out of software and into hardware. Bushnell and Dabney start constructing a prototype in the bedroom of Ted Dabney’s daughter.
January 1971: With a basic feasibility prototype finished, featuring an exerciser standing in for the Nova computer according to Nolan’s deposition, Nolan Bushnell prepares to order several Nova computers from Data General. With the game finally starting to come together, Bushnell and Dabney finally form Syzygy Company.
February 1971: Nolan realizes that the game will still not run on the computer even with the specialized hardware after an employee at a computer center with a Nova computer points out some errors in his calculations. Nolan realizes, perhaps with a nudge from Dabney, the exerciser can just be expanded to replace the computer entirely.
March or April 1971: Nolan Bushnell joins Nutting Associates as chief engineer. The company agrees to consider his and Dabney’s video game project, which will be completed on an independent contractor basis with the duo retaining rights to their technology.
April or May 1971: A basic version of Computer Space is up and running on the prototype hardware.
Summer 1971: A prototype of Computer Space is tested at the Dutch Goose, where it performs well. A subsequent test at a pizza parlor goes less well.
August 1971: Nutting Associates possibly only commits to building Computer Space at this time and licenses the game from Syzygy on the 23rd of the month.
October 1971: Computer Space premiers at the MOA show in Chicago.
December 1971: Computer Space enters production.
They Create Worlds: The Story of the People and Companies That Shaped the Video Game Industry, Vol. I 1971-1982 is available in print or electronically direct from the publisher, CRC Press, as well as through Amazon and other major online retailers.