Tim Skelly's History of Cinematronics and Vectorbeam

Giant List archive, June 1, 1999

In mid-May, some extensive comments from ex-Vectorbeam programmer Dan Sunday were posted to a vector game enthusiast mailing list. Some of this information was new to me, and other bits were obviously wrong, but it started shedding some light on interesting parts of the tangled history of Cinematronics and Vectorbeam, two makers of vector coin-ops in the 1970s. In response, Tim Skelly, the designer whose name comes to mind when Cinematronics is mentioned, wrote up his personal, definitive history of Cinematronics and Vectorbeam. This is the first time many of these details have been exposed. The rest of the words on this page are Tim's.

Okay, now the details of the whole damn [Larry] Rosenthal / Cinematronics / Vectorbeam saga, based on hearsay, reliable witnesses, my own experiences and legal documents I have been privy to over the years. I will try to identify specific sources of information as I go.

Larry Rosenthal and Space War

Until Larry gets to Cinematronics this is just legend to me, but it all sounds reasonable and I have never heard a different version. From Larry's first contact with Cinematronics and up to the time I met him, my sources are Jim Pierce and "Papa" Tom Stroud, at that time the co-owners of Cinematronics.

Either while a student at MIT or shortly thereafter, Larry developed the TTL-based "vectorbeam" board and prototyped a coin-operated version of the famous MIT game Spacewar!. I was told by someone (I don't remember who) that Larry bought rights of some kind to Spacewar! from the guy who is said to have first created it, so that game may not be in the public domain as has been suggested elsewhere. Larry took his prototype to just about every game company in the US, with an offer to split profits 50/50 with anyone who would build and distribute the game. This was an unheard of arrangement, and the industry reaction was a big fat NO. Eventually, Larry worked his way down to Cinematronics, a company that had done a couple of cocktail knock-offs and was about to go under. This was at the time of Pong and its early cousins. Pong had no copyright protection, so there were many companies at that time that began by copying that game, right down to the circuit board. Unfortunately, these companies, like Cinematronics, had nowhere to go from there and had to look for other sources of product.

The collaboration was a huge success, but even though they were very happy with their share of the revenues from Space War, Pierce and Stroud were not so happy with their arrangement with Larry. 50 percent was (and continues to be) an outrageous share to go to a game developer. Also, in addition to his cut of the profits, Larry retained the application patents to his board, which he licensed to Cinematronics. This meant that he received additional cash for every game, Space War or otherwise, that Cinematronics manufactured using his technology. I am relatively sure of these details, but I never saw the contracts, only heard them discussed. However, later events would demonstrate that Rosenthal had maintained ownership of the patents.

My recollection of the sales figures on Space War was 30,000 units. This is not unreasonable given that the game was one of the top ten earners for almost three years, starting at number 1 for 1978 and ending at number 7 in July 1980. (Pac-Man would later break the 100,000 mark.) Based on manufacturing and sales figures at that time, a very reliable number for profit per unit on sales of upright coin-op games was $1,000 per game, net. Manufacturing costs were approximately $1,000 per unit and the games were sold to distributors for around $2,000 apiece. This, of course, varied with the desirability of the game, but given Space War's success, we can assume that it earned at least the minimum in profits. (Note: This $1,000 per unit figure was also used at Gremlin/Sega and Gottlieb/Mylstar during the time I worked for those companies, all the way through 1983.) If you do the math, you'll see that even without Larry's licensing "bonus" he should have made almost 15 million dollars from Space war. Even if I am off by a factor of ten, he still did pretty damn well for an individual in the early days of coin-op video games.

Larry Rosenthal and Me (a very short story)

I'll skip my humble beginnings and go straight to the day I interviewed for a job at Cinematronics in El Cajon, CA, just east of San Diego, sometime around May 1978. After talking to co-owner Jim Pierce, I was sent to the tech area to talk to Larry Rosenthal. There he showed me the "development system" he used to program Space War - a piece of plywood with the TTL board, some LEDs and buttons that allowed him to manually punch in Hex op-codes. That scared the hell out of me, but at least I knew how hex and machine code worked. Scarier was that my limited graphics experience had been with bit-maps. I knew next to nothing about vector displays. Larry didn't explain very much and answered very few questions. Whenever I mentioned our possibly working together he was evasive, so I figured I had failed the interview. I flew back to Kansas City, where I lived at the time, and waited to hear back from some other game companies. I was amazed when Jim Pierce called and told me to drive on out. I had the job!

After a fast packing job and a four day drive, I was shown to my office, which was the same tech area I had been interviewed in—except now it was empty except for some office furniture, a legal pad and a pencil. I met Jim and the rest of the Cinematronics employees, who informed me that during the four days I was on the road, Larry Rosenthal and Bill Cravens departed to start their own company in the Bay Area. Oh, yes, they took with them EVERYTHING that might have been necessary or useful for developing games using the vectorbeam board. Cinematronics still had the legal right to use the board (as long as Larry got his licensing fee), but now they had nothing except me, a legal pad and a pencil to get them a new game to build and sell. Bill DeWolf and another couple of techs were there when I arrived, guys who mainly worked in testing and service, but who later did a great job with custom sound boards, controls and even the modifications to Larry's board that made the graphics in Sundance possible. Soon after I started, Dennis Halverson was hired to create a macro assembler that we ran on a DEC machine. Dennis handled only system stuff and utilities; he later wound up at Atari.

Obviously, things eventually worked out. I managed to crank out Starhawk in time for a winter game show in London. But what still angers me to this day besides being put on the spot like that, is the fact that more than a hundred employees were depending on a new game to maintain their livelihood, and I was clearly chosen as the guy who couldn't come up with one. I think you can see that, under the circumstances, there was no love lost between the Cinematronics and Vectorbeam camps.

Cinematronics and Me

I certainly had no 50% deal. My salary started at 15K a year and after three years had risen to 30k. I received 2 $1,000 bonuses. I've calculated that my games sold at least 59,000 units, total. Was I screwed? You do the math. Here are some events that I witnessed or was party to myself preceding Cinematronics' purchase of Rosenthal's and Cravens' company:

Like I said, the first thing I did was Starhawk, which I programmed first on legal pads in machine code, then on a teletype machine, then finally with Dennis Halverson's development software. It and Space War were the only 4K games. We immediately went to 8K for all future games. Jim Pierce designed the cabinet, which we later found had to have a cinderblock placed in the back or else it would tip forward onto the player!—rather typical of Jim's design talents. The company that silkscreened the side graphics did the cabinet art. The indestructible joysticks, later used in Warrior as well, were handmade at Cinematronics.

Starhawk was enough of a success to keep the doors open, so I began my second game, Sundance. This game was an oddity in more ways than one. It had a vertical screen and a switch which could be set to display Japanese rather than English, the only game I ever did that had that feature. The controls were two matrixes of buttons (3x3, or 4x4. I don't remember which), one set per player. I would have to explain the whole game to tell you what they did and why. The biggest difference about this game was the addition of more levels of intensity for the vectors. This required a daughter board and lots of cut-and-jumpering. As a result, this game was very fragile and few lived long.

Somewhere around the time I was finishing Starhawk, we hired Rob Patton as a second game programmer. He stayed busy learning the system while I was working on Starhawk and Sundance. One day Jim Pierce walked into the lab with a Mattell handheld football game. This was the first handheld game and extremely popular, despite being incredibly simple, with just a few LEDs for a display. Jim thought we should turn it into a video game. I told him that it would certainly stink as a video game and would probably mean a law suit from Mattell. He forgot about it for a while, but when it became clear that Rob had run out of things to do, Jim talked me into letting Rob program it strictly as a learning exercise. That game was Blitz, later Barrier. To make Jim happy, we put it out on test. It did very poorly, to put it nicely, and we stuffed it in the closet.

I started work on Warrior, my one-on-one sword fighting game. Late at night, while waiting for code to compile, I'd go down to the production floor and set a new high score on Speed Freak, the first Vectorbeam game to rise above the radar. On their breaks, the production crew would beat my score.

Cinematronics, Me and Vectorbeam

Speed Freak was a step above other driving games of that time, although like others, the player's point of view (and therefore, car) didn't rotate. Instead, the trick was to slide back and forth like a stick shift moving dropping through a slotted board. Even so, this simulated the driving experience fairly well. It was a good game, but Vectorbeam wasn't selling enough to keep the assembly lines going. They needed something to build and sell, soon.

I know this because I was in the room when Bill Cravens visited Cinematronics, looking for something to build and sell, soon. Cinematronics sold him Blitz/Barrier and we all laughed our asses off. But soon after that a very strange thing happened. Cinematronics purchased Vectorbeam, which at the time was building and failing to sell Barrier. It wasn't until much later, after I had seen the legal documents of the sale, that I was able to figure out this bizarre business move. It was true that Vectorbeam was in trouble, and therefore a bargain, but why purchase a losing company? Cinematronics didn't need product. The short run on Sundance had me hurrying to finish Warrior, but that game was ready to go with time to spare. Though it wouldn't see release for a long time, Rob Patton had started War of the Worlds. Also, by then we had hired Scott Boden, who was already up to speed and ready to start his first project.

So, what did Cinematronics have to gain by buying Vectorbeam? Nothing. Admittedly, the company did gain Tailgunner, but it didn't need it. Cinematronics shut the doors on Vectorbeam as soon as they finished building the game I had developed at Cinematronics, Warrior. Cinematronics may not have had anything to gain from the purchase of Vectorbeam, but Jim Pierce and Papa Tom Stroud were set to gain plenty—Larry Rosenthal's patents. They had been paying (or were supposed to pay) a licensing fee to Larry for every game sold. Now they wouldn't have to. But it gets better. Jim and Papa Tom purchased the patents under their own names, not Cinematronics. Now every time a game shipped, Cinematronics had to pay them, personally, as did other companies that later licensed the technology.

I didn't learn about this until months later. So let's go back to a week or so after the purchase, when Papa Tom's son Tommy had taken charge, and I was tasked with evaluating Vectorbeam's software assets. I flew up to Oakland with some of the Cinematronics techs. This was when I first met Dan Sunday saw the game that was to become Tailgunner, and the demo that inspired Star Castle. The spaceship shooter that was near completion (and might have saved Vectorbeam if it had been finished in time) became Tailgunner when Tommy Stroud had Dan and Larry reverse the moving starfield. This was a clever decision that gave the game a stronger sense of purpose and set it apart from the "attack" POV that had been around since Atari's Starship. Dan and Larry had designed the game so that the player lost a "life" whenever an enemy ship managed to escape the player's shots. Okay, going forward or backward didn't make a lot of difference, but to my mind the "miss a ship, lose a life" game play is a defensive position, not an attacking one. I'm attacking and I miss a ship, so what - just one less dead. We'll swing around and kill him later. But if I'm the battle cruiser's first line of defense and an attacker gets by me, we're all in trouble. I'm not knocking the original version, but wanted to give Tommy Stroud this tiny bit of credit.

On that first visit of mine, I asked Dan what else they were working on and he showed me an odd demo. On the screen was a Space War type ship surrounded by a couple of rotating rings of rectangular blocks. The player controlled this ship, and the rings moved as the ship moved. What I saw in addition to that was a flock of what looked like giant snowflakes. These moved towards the player's ship with increasing speed. When a "snowflake" collided with a brick, the brick disappeared. Eventually, enough bricks were knocked out and enough snowflakes got through to destroy the player's ship. The End. Game Over. That's what I saw. I tried to talk Dan into staying on, but he said he planned to quit and go back east as soon as he finished work on Tailgunner.

Over the next few days I thought about the revolving rings. The game play of the demo wasn't very good, since eventually the player would be overwhelmed by sheer numbers - no real defense strategy except rotate and shoot like hell. Worse, because of the size of the rings, if the player moved the ship to attack or dodge, it was likely that he would move the shield blocks right into the attacking "snowflakes." So, I figured we got Tailgunner out of the deal and let it go at that.

Because Warrior was ready to go and Vectorbeam needed product fast, Cinematronics decided to build it at Vectorbeam and sell it as a Vectorbeam product. Once Tailgunner was finished, it would be built and sold through Cinematronics. My contribution to that piece was limited to designing the cabinet with its blue Plexiglas and the cabinet artwork, which was executed by Rick Bryant.

My only other contribution was subtractive. I made one more trip to Vectorbeam. When I got there Rosenthal was working with Dan Sunday on Tailgunner. They were adding the old Space War starfield to the background and Dan's initials at the bottom of the screen. Regardless of motive, these were bad things to do for one simple reason: Larry's board had a watchdog circuit that reset the program counter to zero if it wasn't hit every certain fraction of a second. This was used to prevent a runaway program from letting the cathode ray shoot the side of the display tube. (Bad mojo, believe me.) This scheme also initialized the program counter. The program counter would usually start up at zero, but nothing in the hardware guaranteed that it would. If it came up at some random value, in a very short time the watchdog would reset it to zero. Sometimes the counter would start inside a bogus loop that hit the watchdog continually, but that was very rare and could be fixed by turning the machine off and back on again.

Every moment of frame time was precious. How much we could draw on the screen was based on the worst case. We had to hit that damn watchdog and we couldn't do it while a line was being drawn. If something on the screen wasn't one hundred percent necessary to gameplay, it had to go, and so went the additional stars and Dan's initials. I was sorry to have to delete his initials because giving credit is very important to me. Initially, it was for technical reasons like this (not enough memory was another) that we designers weren't allowed to put our names on the screen. But that continued long after this technical problem was fixed. My first Gottlieb game, Reactor, was the first coin-op game to have the designer's name on the screen, and I only got that because I was freelance and negotiated it into my contract. The other Gottlieb developers weren't as fortunate.

Back at Cinematronics I had started work on Rip-Off, but Scott Boden still needed a project. Soon, I came up with a way to make Dan and Larry's shield of rings work. I put the enemy inside of the rings and anchored it to the center. The player would have a free moving ship, much like those in Space War. The player's goal was to shoot through the rings and hit the enemy. This was made much more interesting by the fact that the enemy's shots, very accurate and deadly, could not pass existing ring segments. This meant that by shooting out the shield, the player was shooting away the one thing that was protecting him from that nasty gun in the center. I cut the shield segments down to lines instead of blocks (fewer unnecessary lines) and added more rings. The player scored points by hitting and destroying ring segments and other enemy bits and pieces. A hit on the central cannon resulted in an extra life. Besides the central cannon, danger to the player came from "space mines" that originated in the center and then hopped outwards from ring to ring, giving the player some time to anticipate their approach. Destroying a ring segment that held a mine would free the mine to float towards the player's ship. It would continue to seek the player until it was destroyed or had timed out. Whenever a complete ring was destroyed, a new ring was regenerated from the center and other inner rings moved outwards to fill the gap. You can see at work here an attractive tension between the player's motivation to shoot and score, and the fact that his success made it easier for the enemy to destroy him.

Except for the rotating shield rings and a "player" in the center, not one of these game elements was present in the demo I had seen at Vectorbeam, and even those were altered substantially in form and function. No effort was ever made to examine or even save the demo code. And remember what I said about non-functional playfield elements? For Warrior I had used a half-silvered mirror to superimpose playfield art on the screen. For Star Castle I came up with the idea of the colored screen overlays to highlight the rings. Oh, yeah, did I mention that I designed the cabinet art (executed by Rick Bryant) and named the thing too? Scott Boden wrote every single line of incredible, economic and elegant code, and invented many small features that helped add up to the game Star Castle. That is how Star Castle was developed. Period.

When Vectorbeam finished its run of Warrior, the doors were shut. Exidy purchased the rights to build the sit-down version of Tailgunner, which was known as Tailgunner II. As far as the game business is concerned, that was the end of Vectorbeam and Larry Rosenthal.

Tailgunner did well, but not as well as other games. This was not because it wasn't a good game, but because of the failure of the analogue joystick. Chosen by the original Vectorbeam techs, it used a conductive plastic material instead of wires and brushes. This made it extremely durable, a very important feature for an arcade game. (On the plane flight back from Oakland, we stomped on the sample we were given. We couldn't hurt the damn thing. It was a very cool little joystick.) However, what the manufacturer hadn't told anyone was that after x number of movements of the stick the plastic lost its conductive qualities. For other applications, that had been no problem, but for a video game that ate up about a zillion moves a day? As a result, many fewer units were sold than might have been, since the game would die at unpredictable times and the stick had to be replaced on a frequent basis. The sit-down version used a more rugged conventional joystick.

My Life After Vectorbeam and Cinematronics

My next game, Rip-Off was the first two-player [symmetric] cooperative video game. In June and July of 1980 it was rated fourth in RePlay magazine, headed only by Asteroids, Galaxian and Space Invaders. In October of that year it was still rated fifth in Playmeter. Rip-Off was still being built and sold after I had finished Armor Attack and had left the company. I worked briefly for Gremlin/Sega before moving to Chicago in October 1981. There I did three games under contract to Gottlieb/Mylstar: Reactor (Dave Thiel, sound), Insector and Screwloose. Only Reactor was built, but it was done in time and earned enough to jump start that pinball company's video division. This was 1981-1983, and the video game business was about to go down the toilet. The rest is an even longer story, but it doesn't include vector games.