What a 1979 camel game taught us about designing choices for a Garmin watch A design analysis of what a 1979 BASIC text game still teaches about interdependent resources, randomness and voice, and how those ideas hold up on a round Garmin screen with a few buttons and very little memory. [HEADER IMAGE: preracer-editorial-header-1280x720.png] In 1979, a text game called *Camel*, submitted by the Heath Users Group and printed in David Ahl's *More BASIC Computer Games*, gave players a simple job: cross a desert without running out of water, exhausting your camel or getting caught by your pursuers. There were no animations and barely any interface. You made a choice, the game changed a few numbers, and the next problem appeared. A few years earlier, *The Oregon Trail* had done something similar with food, health, supplies and the many ways a journey west could go wrong. [IMAGE: oregon-trail-apple-ii-1985.png] When we started building Preracer, those old games suddenly felt relevant. A Garmin watch has far more computing power than the machines they ran on, but as a place to build a game it brings back some of the same constraints: a tiny screen, limited memory, simple controls and very little room for visual spectacle. If you can't fill the screen with graphics and animation, the game has to become interesting somewhere else. ## Make the numbers depend on each other The interesting thing about *Camel* wasn't that it tracked thirst. It was that thirst competed with everything else. Push harder and you make progress, but tire your camel. Stop to recover and your pursuers get closer. Drink now and you have less water for later. That same tension sits underneath Preracer. You manage four resources: **sugar, water, legs and morale**. They represent different parts of a race, but they aren't four separate health bars. Decisions that protect one resource can hurt another. Morale is where we deliberately went a step further. When morale gets low, the other resources start draining faster. And as things deteriorate, morale itself becomes harder to hold onto. A bad race can therefore become a downward spiral rather than four independent numbers slowly reaching zero. That felt much closer to racing. A pure death spiral isn't much fun, though. If one bad moment guaranteed a collapse, the system would just be punishing bad luck. Some choices restore morale, while high morale slows the baseline drain on sugar, water and legs. We also measure how often simulated runners recover after entering the spiral. The spiral is there to punish ignoring morale, not to end your race the moment it dips. [IMAGE: preracer-impact-fr570.webp] ## Four numbers and a choice on a round screen *Camel* could print as much text as the terminal would hold. On a Forerunner 255 we get a round 260×260-pixel display, where the corners simply don't exist, and five physical buttons. That changes what a “choice” can look like. Every decision in Preracer has to fit into a carousel of no more than three short options, controlled with UP, DOWN and START. No menus inside menus, no scrolling through explanations. If a choice needs a paragraph to understand, it's the wrong choice for the screen. The resources work the same way. You shouldn't have to read four numbers to know how your race is going. Sugar, water, legs and morale each have their own colour and occupy a fixed gauge in the bezel, so you can see at a glance which one is in trouble, the way you'd feel it in a real race before you could put a number on it. On the FR255 tier, the whole game runs within roughly 508 KiB of usable application heap. Every new event competes with code for the same limited space, so content is something we budget, not something we pile on. The same interface also has to work on both MIP and AMOLED watches. A colour that pops on one can disappear on the other, so the palette uses fixed positions as its primary identifier and keeps enough contrast to survive both display types. [IMAGE: preracer-choice-fr570.png] ## Uncertainty makes the same race different Knowing the system shouldn't mean knowing exactly what will happen. *The Oregon Trail* became memorable partly because a sensible plan could still be interrupted by illness, weather or a broken wagon. The route stayed broadly the same; the story of getting through it didn't. Preracer uses the same principle. During a race, events are drawn from different pools. Events can also vary in severity, so the same type of setback might be a minor nuisance in one run and a serious problem in another. The game also avoids serving the same event over and over again. So you can replay the same course with the same basic strategy and still have to adapt. That's important because Preracer isn't trying to create hundreds of different races through content alone. Much of the replayability comes from the interaction between your decisions and what happens along the way. ## Losing should tell you what happened Old resource-management games also understood something useful about failure. Running out of food feels different from drowning while crossing a river. A broken axle tells a different story from disease. Preracer follows the same idea. You can bonk. You can run out of water. Your legs can give up. Your morale can collapse. Or you can simply fail to make the cutoff. During balancing, we explicitly look for situations where one of those becomes the dominant way to lose. If nearly every failed run ends for the same reason, the other resources aren't really doing their job. We check this with a simulator that plays 2,000 runs of each course for each of five strategies. In an early DXT configuration, dehydration accounted for 89% of failed balanced runs. We adjusted the course economy; in the current calibration no single cause accounts for more than 55% of balanced DNFs. A DNF should feel like the consequence of *your race*, not just a generic game-over screen. ## The text is part of the game This may be the part where *Camel* feels most familiar. Its output wasn't written like a spreadsheet reporting changing variables. It had a voice. Push the camel hard and the game announces: *“YOUR CAMEL IS BURNING ACROSS THE DESERT SANDS.”* Give it a night's rest and: *“YOUR CAMEL THANKS YOU!”* Even the instructions send you off with: *“GOOD LUCK AND GOOD CAMELING!”* [IMAGE: preracer-camel-terminal.webp] None of those lines are necessary to make the underlying system work. But they're a large part of why a handful of variables starts to feel like a journey through a desert rather than arithmetic. That's something we want in Preracer too. The scenarios aren't only there to change your stats. They give the race its character. And the short lines that accompany them — *“Over the top on willpower alone,”* *“Your legs are having second thoughts”* — are meant to sound like racing rather than software. Some of them only really work if you've been there. ## Old problem, new course We didn't expect a game from 1979 to be much use when designing for a Garmin watch. But strip away decades of hardware and the problem is surprisingly familiar: not much space, limited input, and no spectacle to carry the experience. You have a small screen, a handful of resources and a player repeatedly asking: *what do I do next?* The rest has to come from what happens around that choice — the trade-offs, the bad luck, the unexpected scenarios and the few lines of text that make a bunch of numbers feel like a race. An earlier version of this article first appeared on [preracer.com](https://preracer.com/notes/camel-1979). Preracer is a free tactical running game for Garmin watches.