Players are really good at spotting patterns, so you have to be careful of the drudgery of having them quest to kill X tribbles for $reason all the time.
Anyhow, good luck with your game. I'm sure there are a lot of ways to make good procedurally generated content that haven't been fully explored yet and I hope you find some of them :)
In addition to procedurally generated, constantly evolving environments.. we've got procedurally re-composed music (my background is in algorithmic music composition). Re-composed, meaning the music is based on human-written themes, but the arrangements are done in realtime by the computer. We're also playing with procedural generation of enemies, to create a lot of variety with relatively little effort. That is, they'll look and behave differently depending on the outcome of the procedural generation.
If there's anything specific you'd like to know I'd be happy to email you about it, or notify you when we release it. If you want to email me, you can use this address: [email protected]
I gather games can be procedural by adding graphics and storylines. Simple missions like escort is easy with random positioning but complex linked ones can be created in tangent with name changes, speech style differences but encapsulating each quest part. Programming can largely go towards procedural generation.
The best part is that if a procedural generation game can be encapsulated pretty well, it can be a different game by supplanting different content and stories. Lots of potential. IAAWG :)
The usual problem with random quests is that they tend to become meaningless over time. I'm not saying it's always that way, but that's a very real trap people can fall into. The issue is that the player loses motivation after the Nth quest to kill X things for $reason.
A better system would work to create a strong motive for the player to perform the quest. It could be as simple as the monster doing something to really piss the player off and making them want to go after it for revenge.
But the problem is that you don't want to keep using the same tricks too often. If you can find a good way to combine N tricks into N! reasonable combinations, that's great, but every time you pull the same trick twice, you cheapen it a little, because players will see the patterns eventually. Seriously, there are crazy people who play games using spreadsheets to plot out exact damage formulas and stuff. I suspect those kids will grow up to be accountants :)
Anyhow, my main point was that procedural generation is a very powerful and very cool tool, but it needs to be used to augment your game, not to dilute it. There's a temptation among some people out there to use it to avoid having to make any real content, which creates the "empty" feeling I spoke of before.
I believe that the people who avoid that temptation and create undiluted games with procedurally generated content will be the ones whose games are played for decades to come.
That'd be wonderful. The "empty" feeling of dilution, and the absence of real content, are issues our team are tackling head-on, with full awareness of such pitfalls. As an example, I was raised on the early Final Fantasy games. A huge, huge part of their appeal was the music — beautiful themes with expert execution by a masterful composer (Nobuo Uematsu). Even now, 20 years later, I have friends who grew up playing the same games who still remember the music. We can hum the themes to one-another, and reminisce about our youth beguiling away many an afternoon absorbed in the stories of these virtual worlds.
But if the stories were procedurally generated, or the music, or any other "hook" that will be remembered by the player, the shared experience is suddenly lost. People may play the same "title", but they haven't played the same game. This is a big, big challenge — how do you leverage the power of dynamically generated content, without breaking the consistency of experience within your game and the commonality of experience between players. How do you create shared memories?
It's easy enough to take a game like Diablo, where the maps are random, or Boarderlands, where the weapons are random, and discuss the game with someone else, arriving at a shared impression and similar memories of play. But if the maps, weapons, quests, characters, sounds, music, rendering style, dialogue, plot, gameplay rules, setting, and even game genre are in some way "generated"... how can you still have a shared experience?
Whoever comes out with a solution to that will be doing some fascinating work.
Also, from what I've heard, the next Elder Scrolls (Morrowind, Oblivion) game is going to have some aspect of procedurally determined quests. I'm excited to see how they do that, as it's not so often that a big studio like Bethesda takes on such an unproven concept.
Well, you can do static generation. You use a known seed and a PRNG to generate stuff that's the same for everyone. Usually, you do that for background stuff, but some people do it and create big, empty worlds that they then try to fill with other stuff. But often, it's for background stuff like trees and clouds, where it really doesn't really create a shared experience one way or another.
There's also the sort of hybrid approach from Dwarf Fortress, where people can share seeds/worldgen info, but I think you already know about that. So people can all share that broken embark with the adamantium space elevator, but their fortresses will likely be totally different and have totally different stories. Maybe one will have their last dwarf go axe crazy while on fire, while another will drop 300+ goblins from the highest z-level all the way down into hell (literally...), or use those crazy weaponized bathtubs they're talking about now.
I think there may be some sweet spots in there with handcrafted stories layered atop procedurally generated worlds, but you're right that more work needs to be done on it.
It sounds like you understand the weak points of procedural generation quite well, so I think you'll be able to use it to full effect and create a great game. Good luck! :)