11 Comments

Love this! In a way, every game tutorial does some of this work--introducing concepts that may be unfamiliar by weaving them into plain English (or whatever language you are speaking). Only here, it's the whole game!

Expand full comment

Aaron, thank you for this review and analysis of Carl Muckenhoupt’s The Gostak, it was seminal at the time. The gostak gestims the doshes. Why is this IF important? Well, you have told many more new people. I’m glad.

Expand full comment

It strikes me that this game could be relevant to contemporary debates over AI and what a system like GPT-4 'understands' by learning through predicting language, and what need there is for additional mechanisms like 'grounding'.

Expand full comment

I played this game for the first time yesterday and it strikes me as nugeric with a GPT-style AI navigating a world of meaningless-to-it-but-interconnected tokens.

The game gives us some English tokens, and a "world" that has some structural analogues with our world/other IF games, so we have more clues than an AI might have. Still, the game captures the experience of navigating an arbitrary semantic network: having no idea what a "dosh" is, but knowing that they are distimmed by Gostaks, vorled by Ghelipers, and deaved in the doshery.

---

Imagine this is your experience of the world:

You are a Gostak, and your fesh is to distim the doshes. In sundry deaves you moskily distim the doshes, and the drokes doatch at you that your fesh is snave.

And yet... you buncap that sundry plaves tim the doshes, or tund hebs that tim the doshes, or even zank the drokes... Perhaps you could tund hebs to disgnurr the plaves? Or even tund a hebtunder? With blide buncapery could you even -- tund a tunder to tund hebtunders? A "hebtunder"-tunder? Then you could distim sundry doshes unmoskily.

But the drokes doatch at you: plaves are snave, hebs are snave, hebtunders are murgous. And don't even nurl about tunding hebtunders. Distimming doshes is snave, but not blide snave.

Bal, yet... sundry plaves *zank* the drokes. Kadly, drokes zank other drokes!

(Kadly, plaves *are* drokes. Drokes reb that as disengoped. Well, plaves dedge disdedgly with drokes... but in that deave we doatch about drokes of kirf and drokes of jirf. Drokes of jirf, when zankage is gomb, can be plaves. Also, jirf is jirf and kirf is kirf, but in unmosky deaves, jirf can be "hoff". Anyway, drokes are drokes, even though in mosky deaves drokes were not all drokes, but in unmosky deaves drokes of kirf and drokes of jirf are drokes... though drokes of jirf can be murgous. Disengoped!)

Well, a gostak that *did* tund a hebtunder, or simply tunded sundry hebs, could zank drokes and disdedge drokes that zank drokes. Or, if not zank drokes, glaud drokes that zank drokes. You doatch this at drokes, who doatch at you gharmy: zank drokes is scurm, glaud drokes is scurm, a gostak that vuts these things will be zanked.

And yet, plaves tim the doshes. Bal, in deaves "plaves" are "drokes" and in deaves "plaves" are not "drokes". So, doshes timmed or drokes zanked. What if, not zank drokes, or glaud drokes, but tund hebs to rorm drokes. Drokes stam rorm, it is the fesh of all drokes. If a gostak tunds hebs to rorm drokes, all drokes tunsel, no zankage, no glauds, and no *timmed doshes*. And drokes doatch: da!!! All drokes tunsel is dass glaud.

All drokes tunsel is dass glaud? Glaud is scurm, tunsel is rorm, but tunsel is glaud? Perhaps a gostak not snave enough to disengope drokes. Perhaps a gostak just distim the doshes. But sundry doshes, and sundry plaves tim the doshes... Perhaps a gostak, if must pell to gnurr, must first tund a *snavetunder*...

[And then the universe ends up tiled with distimmed doshes, tunded by a recursively self-tunding supersnave gostak. No drokes were zanked, and all are deaved in a tunsel drokery. All drokes are rorm and tunsel in tunsel drokery; drokes that doatch that tunsel drokery was scurm glaud were riked to fesh deaving moskily.]

Expand full comment

This article pronked me double much. Clip clop, Aaron.

Expand full comment

Fascinating concept, but I'm not sure I could endure the challenge required to finish it!

Expand full comment

It definitely takes a very specific kind of player to try! Fortunately I suspect the overlap between "masochists" and "people interested in linguistics" is fairly high ;)

Expand full comment

offered as support for your theory: my first thought was "I bet you could do something like this where the vocabulary is randomized, so it's replayable after learning it once"; my second thought was "I'm absolutely playing this"; and I have a degree in linguistics. 😂

Expand full comment

I've taken courses in psycholinguistics so I'm halfway there....

Expand full comment
Comment removed
Apr 16, 2023
Comment removed
Expand full comment

Jason's blog is the best!!

Expand full comment
Comment removed
Apr 16, 2023
Comment removed
Expand full comment

Note: some might consider perusing this list major spoilers!! (But thanks Curtis for posting the link for those who need it.)

Expand full comment