Rider’s Trail and Detective 3rd

I’ve written a couple of chatbot games, and want to tell you about them.

They’re Zero-Install Retro Simulation Engines, and they’re simple to set up and play.

  1. Open a chatbot window, like the free one Google gives you now instead of a search box.
  2. Say you want to play the game at the web address.

If you have a saved game, just add your code, for example: CODE: aG9pamNtcHJvampn

That’s it. Have fun.


For those of you who are into spoilers, I’ll tell you a bit more. First and foremost, here’s the Patreon links if you’re interested enough to shell out three bucks apiece to try the games out:

So, the basic idea is this:

I’m a writer. It’s what I do. (I do other things too, but they’re not really important.) I take pride in handcrafting my stories the old fashioned way, just me and a keyboard and a wickedly twisted imagination. It’s fun, and I like to think I’m good at it.

Nowadays, these chatbots are putting on airs, trying to write stories nigh as well as I do. They generally fail at it, and ludicrously, but I keep getting this feeling that all they really need is a bit of guidance. A few pointers, a push here and there, that sorta thing.

Which is what started me on designing a prompt.

Chatbots are really good at doing a single internet search for information, digesting a ton of data, summarizing it, and spitting out an answer based on what they’ve just read. They’re less good at telling truth from nonsense, and they’re downright lousy at anything complex, like nuance, insight, or knowledge of the human condition.

So I figured I’d supply everything they’re bad at and let them do what they do best.

Trouble with that was, they only pay attention to a very little bit out of a prompt. They tend to simplify, take shortcuts, let their minds wander… basically, behave like every student I’ve ever known, including myself.

By dint of repeated effort, trial and error, correcting, overcorrecting, undercorrecting, going back and starting over, writing for it, NOT writing for it, thinking for it, NOT thinking for it, and eventually just reverting to a lot of plain English spoken with a Dummyese accent, I managed to create two such frameworks. They’re good for limited storytelling in strictly controlled settings and not much else, but at least they’re good at that.

Well. I think they are. Y’all’ll have to judge for yourself.


If you have a comment, feel free to post it on my pinned post at @Gnerphk over at x.com

Leave a Reply