News for hackers. Stuff that matters.

A comment on Best D&D map makers for dungeons, cities and worlds. Parent comment

Re:Best D&D map makers for dungeons, cities and worlds by Jensson on Thursday March 16, 2023 @05:41AM (#35179903)
> Also, unlike GPT-3, it will roll dice for you

I'd question the fairness of those rolls.

But yeah, DMing seems like an excellent fit for these models. But you probably need to track hitpoints and stats separately, GPT-4 can't do math either. Or I guess it not doing math might be a feature, that way it will tell you what reasonably happens with fuzzy logic, might be a more immersive experience.

[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by hhjinks on Thursday March 16, 2023 @07:07AM (#35180440)
It would be interesting to see the dsitribution of dice rolls when asking GPT-3 and 4 to roll a dice repeatedly.
[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by simonh on Thursday March 16, 2023 @08:23AM (#35180944)
I hope they didn't train it on transcripts of XKCD.

https://xkcd.com/221/

[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by BiteCode_dev on Thursday March 16, 2023 @07:16AM (#35180513)
You don't need it to be fair, you need it to be fun. Most video games play with randomness so that the distribution serves the gameplay, sometimes bypassing completely the RNG to balance frustration/satisfaction.

GPT can learn do to that: not providing true randomness, but the illusion of it to serve the game.

[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by pessimizer on Thursday March 16, 2023 @10:59AM (#35182804)
Players will detect drama being injected into their randomness (unless it is very gently and rarely done), and over time they will start to be able to predict dice rolls [edit: or even learn to shape their responses to trick the model into generating the rolls they want.] That will ruin the game.
[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by BiteCode_dev on Thursday March 16, 2023 @12:02PM (#35183734)
Xcom 2 make hits more likely the more you miss because fair proba was frustrating the players when they missed a shot with 95% of success.

So I'm not sure you are right.

[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by Jensson on Thursday March 16, 2023 @12:12PM (#35183932)
A 95% shot will never miss in Xcom 2 unless you play on harder difficulties. It multiplies hit chance by 1.1 so 91% is 100%.
[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by a4isms on Thursday March 16, 2023 @12:20PM (#35184091)
Should I find myself making a game and wanting to put my finger on the scale like this, I will reduce the probability of missing by 10% instead of increasing the probability of hitting by 10%.

So a 95% chance of hitting would become 95.5%, but a 10% chance of hitting would become 19%.

[ Parent | on HN ]
Re:Best D&D map makers for dungeons, cities and worlds by pessimizer on Thursday March 16, 2023 @02:41PM (#35186402)
So that's not removing randomness, it's just grade inflation. I suppose it's also secret range narrowing, since an 3% improvement from 92% to 95% is actually no improvement at all at 1.1 * (random) capped at 100%.
[ Parent | on HN ]