CASA ::

I'm surprised that a model made an assumption.

Mr Creosote wrote: Sun Mar 01, 2026 8:59 am This one comes from Z.ai's "GLM":

It fails fairly quickly as you cannot teleport from the headmaster's garden (location 4) to location 204. You can do it the other way around, but.

Very interesting! I fed this into several different models. Each of them found the winning condition with ease. But when it came to actually writing down a step-by-step solution, one model just laid out generic, high-level descriptions. Two others said "OK" and started printing commands, but ...

The ridiculous number of locations aside, those QEGS games seem to be very sparse on objects, possible interactions etc. That makes them easily parseable, I guess.


Yes, they are long on location and short on interest or puzzles. Which presumably explains why they were unsolved.

In QEGS 2 ...

Mr Creosote wrote: Sat Feb 28, 2026 7:56 am Would you have the "UNGACed plain text file"? Sorry, I'm not so familiar with these formats, but I'd love to experiment a bit on my side.

Sure:
https://www.dropbox.com/scl/fi/7ym2jgw7 ... pitpp&dl=0

And to be honest, having looked at these games in the past, a human shouldn't have to be forced to solve them! Though, of course, my input was necessary.

Agreed. I was similarly surprised at its success. It did really well at QEGS I. Although, if you look at the solution, it's barely a game at all really. However I was impressed by the understanding of the use of the teleport over the standard map.

I decided against keeping the transcript because ...

Part of the problem (for me) was Copilot's undeserved (over)confidence.

"I will produce a precise, complete walkthrough".
"I will give you a perfect, step‑by‑step solution with no guesswork".

Umm... No. But what was provided was very useful.

So I tried QEGS II https://solutionarchive.com/game/id%2C7377/QEGS+Adventure+II%2C+The.html and it did not fare well, unable to map the game at all.
However, it did summarise the objectives and how to achieve them, which enabled me to solve it very quickly.
No, it wasn't fun. But I'd decided a few ...

I asked Copilot to solve The QEGS Adventure ( https://solutionarchive.com/game/id%2C7376/QEGS+Adventure%2C+The.html ) by uploading an UNGACed plain text file. It did pretty well and with my occasional corrections did just that. I've uploaded "our" solution.

I don't think it will be the most ...

Avior https://solutionarchive.com/game/id%2C912/Avior.html
and
Avoir https://solutionarchive.com/game/id%2C2523/Avoir.html
are duplicates.

They both have a solution and map.

I think Grimble https://solutionarchive.com/game/id%2C2879/Grimble.html is a duplicate of Crimble https://solutionarchive.com/game/id%2C2651/Crimble.html, given that Crimble is a.k.a. Grimble.

Please delete Der Rote Modstein https://solutionarchive.com/game/id%2C5 ... C+Der.html as it is a duplicate of Der Rote Mondstein https://solutionarchive.com/game/id%2C9 ... C+Der.html. Thanks.

Please delete the Atari 8-bit version of Pharaoh's Curse Adventure https://solutionarchive.com/game/id%2C1 ... nture.html.
It is a duplicate of https://solutionarchive.com/game/id%2C7 ... Curse.html

Please delete Golden Oldies Volume 1 as it is a compilation of several games, only one of which is an adventure and is already listed (Software Toolworks' The Original Adventure).

https://solutionarchive.com/game/id%2C1 ... ldies.html

Thanks.