Garry Kitchen - Retro Gamer

Garry Kitchen’s name won’t be familiar to everyone, but he’s made some big contributions to gaming. He’s worked on the impressive conversion of Donkey Kong for the Atari 2600, creating his innovative GameMaker in 1985 and made a number of successful games for Activision. Here he looks at his early footsteps in the industry.

When did you first encounter a computer?
My first exposure to a computer was in college in the 1976-1977 timeframe, when I took a FORTRAN programming class. We had to write the program on a timeshare terminal and submit it to the computer lab, where they would run your code and give you a printout with the result. I didn’t get much out of the course and it certainly didn’t encourage me to focus on software. That happened later, ‘on the job’, as I found that programming assembly language on microprocessors was a whole lot more fun that FORTRAN.

You then got yourself a scholarship to study?
Matsushita/Panasonic wanted to help electrical engineering students who were attending college while working to pay their way through college, as I was. Any EE student in that situation was encouraged to submit an application for the scholarship. Matsushita reviewed the students’ applications, grades and I was fortunate enough to be chosen.

Did you always plan to work with game hardware?
While in college I was offered a part-time job with a small engineering company, Wickstead Design Associates. With no experience, I was ‘low man on the totem pole’, and started at the bottom, learning how to be a technician, soldering circuit boards, building prototypes etc. Being a small company, my responsibilities quickly increased to the point where I was doing hardware and software design. Our initial work was in general consumer electronics (digital photographic timers, digital instrumentation, digital clocks, etc).

However, in 1977, Mattel released their Mattel Football handheld electronic game, and in 1978 Space Invaders the arcade game showed up in the local pizza parlour, and I was hooked. We shifted our focus to toys and games, working first in the electronic toy space (Wildfire, Bank Shot for Parker Brothers) and then in the videogame space on the Atari 2600. Since 1977, my focus has remained on interactive electronic entertainment.

An early handheld game Garry worked on.

What was it like designing handheld games?
The two Parker Brothers products were LED-based, not LCD. We preferred an active light source (LED), rather than a reflective light source (LCD). At the time I thought that LCD displays lacked vibrancy, clarity, brightness and colour. Needless to say, they’ve improved a lot since then.

I started programming games on a 4-bit microprocessor (the AMI S2000 or the TI TMS 1000), driving a display made up of a matrix of LEDs (I believe 76 in the Bank Shot billiards game). In addition to handling all of the gameplay logic, the code had to strobe the LED display to keep it lit; unlike a bitmap display, you couldn’t just turn an LED on and have it stay on. Each LED turned on for a split second, followed by the next and the next etc, under software control. It happened so fast that your eye didn’t notice the flicker and it looked like a solid display. Scanning the display in this manner was mandatory if you didn’t want to use six AA batteries every hour.

When the Atari 2600 took off and started to seriously hurt the handheld electronic toy market, I jumped over to that platform, which was based on an 8-bit processor (6502/6507). Once again, without a bitmap, the software had to run synchronously with the display, drawing every individual pixel on the screen. It was not that different than the standalone, handheld games that I had worked on. Looking back, I enjoyed designing both electronic toys and videogames. Similar, but different challenges.

In 1980, you reverse-engineered the Atari 2600 to see how it worked. How long did that take?
It took about six months. My brand new Apple II computer was key to the effort; it would have taken much longer without it. While I led the effort, I had help from a brilliant programmer with a PhD in Physics, who worked with me to decode the software routines.

Were you surprised when your first Atari 2600 game, Space Jockey sold a million units?
While Space Jockey was a modestly fun game, it was very simple in concept and gameplay mechanics and was really just a test bed for me during the reverse engineering process. I had to write code to test out how the system worked so I picked a simple game demo with multiple horizontal bands of objects, tailor-made for the 2600 hardware. It was also a 2K game (2048 bytes), so it was pretty limited in graphic and audio complexity (the things that take up more memory). It was a decent effort given the fact that it was my first game on the platform but it’s certainly not a masterpiece. Not to be self-deprecating but I think the majority of those units sold for less than full price.

This is Garry’s senior year collage project. It got him an A.

Then you got to convert Donkey Kong to the 2600…
Donkey Kong 2600 was a monster to do. I loved the arcade game when it came out and I was thrilled to get the project to adapt it, but it was a daunting task. After the critical reception that Atari’s Pac-Man cartridge had got, I felt great pressure to deliver on this next high profile coin-op adaptation. While I had no help from Nintendo I did have direct access to an arcade machine. I could play the arcade game but I didn’t have any access to the game resources; graphic frames, sound samples or code.

What made it difficult?
The ramps – the ‘background’ graphics on the 2600 were incredibly crude, only 40 blocks across with each block made up of four horizontal pixels; ie big and blocky. Even worse, they weren’t even 40 unique blocks. The hardware defined 20 blocks (which covered the left side of the screen), and the right side was either a copy of or reflection of the left side. Historically, a 2600 programmer would set a mode switch for copy or reflect, define the 20-block pattern for each line of the screen and forget about it. The hardware would display the right side of the background automatically. If you look at a game like Combat you see that the background pattern is either symmetrical or reflective.

My problem was the slanted ramps that the barrels rolled down, requiring a completely unique graphic pattern from the left edge to the right edge. I originally tried flat ramps but it didn’t look good enough to me; ie it didn’t look like the arcade game. I eventually rewrote my display kernel (the code loop that draws each scan line of the TV image) to load up one 20-block pattern for the left side and a second 20-block pattern for the right side before the television raster reached the centre of the TV screen. By changing the background graphic registers on the fly I was able to create a non-symmetrical display with slanted ramps. It was not easy! I worked long and hard on the ‘feel’ so that it played like the original. I obviously worked long and hard to make the graphics look as close as possible to the arcade game.

Fans have commented on the missing levels…
To clear that up once and for all, given the schedule and the size of the ROM that was available to me, it was impossible to add the two other screens. Despite the fact that bank-switching technology was available and the game could have utilised an 8K ROM, Coleco (the publisher) insisted on a 4K ROM to save money. Despite my best efforts, I was not able to dissuade them. Also, even if I had had an 8K ROM to work with, it would have taken a couple of months more to program the other two levels. I did the entire project in record time (three months), spending the last 72 hours straight (no sleep, no breaks) sitting in a cubicle in Hartford, Connecticut with the owners of Coleco standing over my shoulder waiting for the finished game.

I’m proud of the Atari 2600 version of Donkey Kong. I think I did as good a job of converting the arcade game as any programmer could have done, given the limitations I had to live with.

A prototype for Space Jockey.

What inspired your next game, Keystone Kapers?
After Donkey Kong, I was inspired to do another ‘little man’ game. I was also interested in doing a funny, whimsical character. Many times on a machine with limited graphics capability, the theme is driven by what you can display that looks good. The Keystone Cop worked well graphically because he was recognisable wearing the signature English Bobby hat. I also had the concept in my head of how to write 2600 code to make an animating escalator. I was always trying to make realistic animations of mechanical contraptions (eg the oven and conveyor belts in Pressure Cooker). Once I had the escalator I had to build a game around it and the combination of the Keystone Cop and the escalator lead me to a multi-floor department store and an escaped convict (who also displayed well in a horizontally-striped outfit).

How long did Pressure Cooker take to create?
About eight months. I was standing in a Burger King at lunch watching the burgers move by the flame on a conveyor belt and had the idea.

The jingle in Pressure Cooker is one of my favourite parts of the game. People asked me how I made it sound so good (ie in tune). The trick was simple. Most of the tones that the Atari 2600 could generate were out of tune and it drove me crazy when designers used them in game music; there’s nothing worse than out of tune music. I decided I wanted good sounding music so I brought in an electronic keyboard and went through each tone on the Atari 2600, marking the keys of the keyboard when I found a tone that matched an in-tune note. We then hired a professional jingle writer, sat him in front of the keyboard and told him to write a song, using only the marked keys. His first reaction was ‘you’re kidding me, right?’ but he soon got into it and wrote the Pressure Cooker jingle. With no sour notes it ended up sounding great.

Was Activision a good place  to work?
It was great in its time. The smartest thing that Activision did during that time period was to hire brilliant creative people and leave them alone to create. That strategy only works with the right people, of course. But when it worked, it worked very, very well. No marketing department or corporate ‘suits’ and bean counters were going to come up with Pitfall!, Chopper Command, Olympic Decathlon or any of the other great games that Activision made during that time frame.

It amazes me today how that lesson has been completely lost. I know many game developers that work at large, well-established companies who describe a creative environment 180 degrees opposed to the Activision model. Coincidentally, these are the same companies that also fail to innovate, putting out the umpteenth first-person shooter or clone of yesterday’s best-seller. Innovation happens in small environments where passionate people relentlessly pursue a vision, without having to justify the project’s existence at monthly review meetings. Yet large companies (who you would think know better) insist on constraining the creative process with levels and levels and levels of opinions, lengthy design specifications, approvals, green light processes, concepts, ideas and tweaks from a committee of people who have no idea how to make a fun game, all in the pursuit of the ‘next Angry Birds’. Well, Angry Birds did not come out of a large, constrained corporate environment and I can guarantee that the ‘next one’, whatever it ends up being, won’t either.