When I was growing up, my family was fortunate to have a computer. This alone may sound odd to folks who have grown up with a tablet or smartphone in their hands at all times, but the 1980s saw a slow and gradual expansion of computers in homes.
People didn’t really understand their wide-ranging utility, and both hardware and software were pretty user-unfriendly. To run programs, you had to understand at least a tiny bit of code yourself—at least the very basic C prompts from MS DOS, just to run programs (even games). You needed to load the program from an audio cassette tape, just like the kind on which you’d create a mixtape.
This clunky interface made people who played games on computers necessarily a bit more knowledgeable about code than, say, today’s gamers who can just log in and start playing.
Another common way to play video games on your home PC was to type the code in directly and just run the program. You could buy computer-specific magazines that had the entire game’s code printed in there over several pages; all you had to do was type it all in without making a single error, and—poof! easy as pie, you could be off and playing in just a few short days.
For my part, I very strongly preferred to play games from cassettes, although I did eventually dabble in writing my own BASIC code to run a few programs of my own.
All this is just to set the stage for this game I really enjoyed playing. It was a turn-by-turn role playing game for the computer, based on the Star Trek universe. The program was incredible to me because it “understood” natural language in a very simplistic and limited way.
At age eleven or so, I was very keen to probe these language limits. I quickly surmised that the inputs needed to have a vowel and a noun in them, such as “shoot Klingon” or “grab phaser.”
After playing this game relentlessly for hours (and probably over several days), I began to realize that you didn’t actually need to type the whole word in. Maybe I made a typo myself, or maybe my dad (who helped me learn to code a little bit, and who also really enjoyed Star Trek) helped me figure this out—this is lost to the dustbin of my memory, I’m afraid.
The program generally worked well to prevent you from skipping ahead, instead limiting the potential actions you could take to the few realistic options available. It was well thought out! But of course, there were ways to get around this.
The plot eventually necessitated you, James T Kirk, to beam down to the surface of whatever planet the Enterprise was orbiting. There was a lot of lead-up to this step that you’d have to go through every time you played the game.
Shortening the words you needed and fumbling around with lots of experiments yielded a magic incantation: go surf.
This meant “go to the surface”, or “beam down to the surface.” You could do this from the Enterprise at pretty much any time, so I eventually skipped ahead in the game whenever I started a session and wanted to beat it quickly without repeating the same tedious beginning.
This kind of experimentation with the most efficient language strikes me as especially relevant today. With LLMs (Large Language Models), the idea is to just say what you want the AI to do. Seems simple, right?
It is and it isn’t. It’s simple because, yes, for the first time, you can ask a program to do things in the same language you might use with another human. It’s complex because language means different things to different people, and what you think you’re saying might not have the universal meaning you think it has. It’s also complex because the AI will often be very good at either creativity or accuracy, but probably not both at the same time.
C prompts and MS-DOS were seriously nothing like this. The arcane code you needed to create programs (which you can do in LLMs by just asking for them) was tedious, took time to understand and learn, and was antiquated by the time I got any good at it.
I was intrigued by the Star Trek game because it allowed me to use natural language that I understood, in order to get the computer to do things in novel ways. You might call this prompt engineering practice.
What’s the “go surf” of talking with generative AIs? I’ve personally found quite a few shortcuts here and there, and I’ve learned a ton of tricks from
over at —shortcuts for prompting that have made it much faster to get where I want to be.In my lifetime, we’ve traveled from having to use punch cards (still in use at the time of my birth) to run programs, to popping a cassette into a machine to load a program, to downloading or streaming a program instantaneously. At the same time, we’ve gone from arcane code to natural language in order to create new things.
Can you see a future that makes this process even simpler than using a natural language dialogue? What does that future look like?
Did you ever play any turn-based RPGs like Star Trek when you were a kid? What were your favorites?
In jr. high (1971!), the math teacher set up a computer in the lab, the kind that used 0s and 1s. A couple of the math geeks programmed it to run in a loop, and the printer used up all of the paper that was in a very large box.♾️
I had an Atari 800 in the early 80s and remember well the cassettes. I had more fun programming it than playing the actual games.
Later, I got on several BBS and had a great time “surfing” those. The early days sure were fun, weren’t they?🔚
Thanks for the memories!
Oh man, I remember coding basic stuff in, well, BASIC. I even remember drawing entire pictures by programming each line at a time. I believe you had to first tell it the coordinates of the screen as the starting point, and on the next line tell it which direction to draw and when to stop. Then off to the next "brush stroke." Took ages, but I was so damn proud.
Can't say I've ever played the Star Trek version but I definitely played games where you could type basic "natural language" commands. Have the same memories about it as you do. Voodoo magic.
And I've always been a huge fan of turn-based games. My favorite turn-based strategy to this day is the X-COM series. And my favorite turn-based RPG was hands-down the Fallout series (although I started out at the Fallout 2 stage and never played the original).
I appreciate the shoutout, too!