D&D General The DM Shortage

[Reynard]

You just need a set of rules for determining DCs. That's easy to code.

The only thing that's challenging for a computer is creating engaging story and characters (and even that might be close). You don't need that for an OSR game, that's the players' job, insofar as it happens at all.

My point was the OSR style game doesn't use DCs the way a modern game does and the GM does a lot more... GMing.

 

[Fanaelialae]

You just need a set of rules for determining DCs. That's easy to code.

The only thing that's challenging for a computer is creating engaging story and characters (and even that might be close). You don't need that for an OSR game, that's the players' job, insofar as it happens at all.

It depends on kind of action you're setting a DC for.

For example, for a jump check you might set the DC at 10 + distance in feet. That is easy.

On the other hand, setting the DC for more open ended possibilities is far more challenging. Something like, does my character know if trolls have any vulnerabilities? You could have a set of DCs based on how rare the knowledge is, but unless someone creates a database where they designate the rarity of knowledge for every possible question a player might ask, this is far more difficult to implement, and much more prone to error.

I'm no expert, but as I understand it, AIs basically use a mathematical algorithm to derive the probable correct response from a large data set. So you'd need a suitably large D&D specific dataset with things like DMs settings DCs for all kinds of player requested actions for the AI to have a reasonable chance of deriving a reasonable response.

 

[Paul Farquhar]

My point was the OSR style game doesn't use DCs the way a modern game does and the GM does a lot more... GMing.

Makes no difference. There are rules for NPC reactions, and everything else is either thief skill %, auto-success or auto-fail. It's all easy to code, and has been, right from the gold box games.

 

[Reynard]

Makes no difference. There are rules for NPC reactions, and everything else is either thief skill %, auto-success or auto-fail. It's all easy to code, and has been, right from the gold box games.

If that's how your old school games ran, no wonder you think of them as nothing more than adversarial hack and slash.

 

[Paul Farquhar]

Something like, does my character know if trolls have any vulnerabilities?

Computer games, such as Solasta and the Owlcat Pathfinder games, have already solved this. Doesn't take an AI.

 

[Ruin Explorer]

I can only relay that we did not have a high body count in any edition. At least not higher than we wanted.

The game has always been what you want to make it.

Agree completely.

2E, in practice, we didn't see a huge body count, because we used virtually every optional or suggested rule that would reduce it, and my experience at least was that most DMs did similarly and/or had house rules that effectively did that.

Plus people fudged. I'm sorry we're too late in history for me to pretend. I'm 44 and I've been running RPGs for 34 years at this point. I can admit it. We fudged, and a lot of DMs fudged (some in one direction, some in both). Most DMs rolled behind a screen back then and that really helped with fudging.

And as well as fudging, you could just have enemies not kill the PCs. Often it made sense, too.

 

[Paul Farquhar]

If that's how your old school games ran, no wonder you think of them as nothing more than adversarial hack and slash.

That's how the book you cited in the OP told people to run them.

That and lots of procedurally generated stuff. All so simple that a computer could do it in the 90s.

 

[Celebrim]

My point was the OSR style game doesn't use DCs the way a modern game does and the GM does a lot more... GMing.

Not really true. OSR style games just have less well-defined skill systems, with the result that each encounter has its own ad hoc generated skill system. You can see this in plenty of the modules of the day, with each room given guidelines for how to resolve fortune when the players attempt to do various things. There really is no more or less ad hoc GMing required when the players purpose something not covered by the guidelines, the difference is only where guidelines come from, how robust and transferable they are, and whether there tends to be a uniform approach.

If you read a published module from the era, you'll find lots of "there is X% chance to this as you cross the room" and "calculate the percentage chance of success as follows" and "force the PC to make a save versus wands/paralyzation/etc to avoid the trap" or "have the PC roll under his DEX on a D20 to succeed" or "Make the PC make a bend bars/lift gates check to succeed". That's all setting the DC very much like a modern game, it's just hidden under all the clutter and typically hand crafted for each scenario.

But in my practice with GMs, they develop tool kits and tend to apply answers uniformly simply out of habit if nothing else, at which point you could actually write down the rulings as rules and you'd have a table of DCs similar to the guidelines of modern systems.

 

[Fanaelialae]

Computer games, such as Solasta and the Owlcat Pathfinder games, have already solved this. Doesn't take an AI.

Open ended questions? I don't think so. A pre-determined list of questions the programmers anticipated, sure.

Text based adventures like Zork have existed for decades, but were always limited in that they can't do anything the programmers didn't explicitly code for.

The advantage of an AI is that they can (potentially) handle scenarios that haven't been explicitly coded for. Which makes them more like a real DM, as opposed to a Choose Your Own Adventure book. But they still need a large data set in order to compute a probable correct response.

Otherwise, there's no reason to use an AI. There are plenty of CRPGs that can handle a broad number of (programmer anticipated) player choices. But they're not actually the same as playing with a real DM, who can actually improvise when a player attempts something unanticipated.

 

[Oofta]

So, like, next year?

Specialist systems are one thing, you would basically need a generalized AI. Like fusion power, that's at least 5 years away and has been for decades.

You could get a module generator and probably an AI that runs combat, but that's only part of the game. You'd have to have an AI that could convincingly replace a real DM ... that's not going to happen any time soon.

A dungeon crawl generator? Maybe. A generalized intelligence AI? At least 5 years away and likely will be for quite some time. Besides, if someone really does figure it out, there's a pretty good chance finding a DM will be the least of our worries.