I just got sent this screenshot from some rando Facebook group about retro gaming. And I love it and don’t know how this is the first time I’ve seen this.
When it comes to 80’s and 90’s tech and game development, I’m high defensive about it. People who worked on this stuff were legitimate trailblazers who had no funding. “Built in my kitchen” did not have a caveat of “took of 200k personal loan from relatives.” Some of the most important tech and gaming legacies were bootstrapped.
My personal distate for Kevin O’Leary, for example, goes back to him destroying the value of some of the best gaming studios in the world. (Broderbund, MECC)
Those studios were made of people like Ken and Roberta Williams. They were eventually the founders of Sierra Online, to this day my favorite gaming studio and the one that set the stage for almost every game made today.
I hope AI lets us get back to this type of personality and this type of builder. I’m already hearing stories that it is.
Here’s the story from the screenshot in full:
The monitor glowed with a single blinking green cursor. The machine on the desk cost thirteen hundred dollars, and the industry consensus was absolute: personal computers were tools for processing text.
In 1979, the computing environment was strictly mathematical. A consumer machine was an instrument for accounting, statistical sorting, and data entry. If a user wanted entertainment, they interacted with the hardware by typing commands into a void. A player read a paragraph describing a physical space, and typed a directive like “go north” or “take lamp.” The machine responded with another text paragraph.
There were no shapes. There were no colors. The hardware did not possess the physical memory to hold photographs. A single standard image required more storage than the entire Apple II computer contained. To ask a machine to display a visual narrative was to fundamentally misunderstand what the hardware was built to do.
Roberta sat at her kitchen table in Simi Valley, California, staring at the green text. She had no background in programming. She did not know how to write software. She was a twenty-six-year-old mother who had spent the previous month playing a text-based mainframe game called Colossal Cave Adventure. She accessed it through a teletype terminal her husband brought home from his systems programming job.
When she finished the game, she looked for another. She found a program called Zork, but after completing it, the inventory of available digital stories was exhausted. She mapped out the text coordinates of these games on sheets of grid paper just to keep track of her position in the digital space. She looked at the paper maps and wondered why the computer could not simply draw the rooms.
Her husband, Ken, was a systems programmer. He had purchased the Apple II intending to write a Fortran compiler. He explained to her that the machine had a high-resolution graphics mode, but using it for an entire entertainment program was functionally impossible.
At the time, the standard consumer Apple II offered 48 kilobytes of memory. A single digital bitmap image of a house would consume the entire floppy disk for just one room. The limitation was a hard physical boundary, not a creative choice. System architects working in the late 1970s understood that storing pixel-by-pixel images was mathematically incompatible with the era’s hardware limits. The engineering community accepted this parameter as a permanent reality.
Roberta ignored the hardware specifications. She bought a large sketchbook and drafted a murder mystery. She drew seventy separate scenes of an abandoned Victorian mansion. She handed the stack of paper to Ken and told him to make the machine display them.
He explained the math again. Seventy bitmaps would not fit on a single disk. He refused to write the code because the machine could not hold the data. Roberta continued asking. She located an obscure peripheral device called a VersaWriter, a mechanical arm designed to trace physical drawings into a digital interface.
The solution bypassed the memory problem entirely. They did not save the drawings as static images. Instead, Ken wrote an assembly language routine that recorded the exact coordinates of the mechanical arm as Roberta traced her sketches.
The disk did not hold a picture. It held the mathematical instructions of how to draw the picture from scratch, rendering lines in real-time whenever a user entered a digital room.
The resulting images were crude. The lines were jagged and uneven. The characters were literal stick figures because the software could not render human curves smoothly. The game occasionally crashed if the drawing routine overwhelmed the processor.
They worked at the kitchen table for three months. When the code was finished, they lacked money for professional packaging. They bought clear Ziploc bags from a grocery store. They printed instruction manuals at a local copy shop, slipped the floppy disks inside the plastic, and sealed them.
In May 1980, they drove to a MicroAge computer store in Los Angeles and convinced the owner to put four plastic bags on a display shelf. The price was marked at twenty-four dollars and ninety-five cents.
They called it Mystery House. It sold ten thousand copies by the end of the year. The compiler Ken intended to write was never finished. The company they founded in their house eventually employed hundreds of workers and shipped millions of units.
By 1982, the commercial market for text-only computer games collapsed entirely. The original VersaWriter mechanical arm was lost in a subsequent office move. The paper sketches were discarded after they were traced.
Today, the global video game industry generates more revenue than the film and music industries combined. Modern graphical interfaces calculate millions of light reflections per second in photorealistic environments. At the base of that architecture is a set of jagged digital lines, drawn by a woman who did not know how to code.
Source: Roberta Williams, Ken Williams archives.
Verified via: The Strong National Museum of Play, Vintage Computing records.
(Some details summarized for brevity. Source link in the first comment.)