Dynamic Quest Generation

So, while I'm thinking about it, I'd like to talk about something called "Dynamic Quest Generation". Many people have heard of it. Many others have asked for it. I'm here to discuss why it is feesible, and why it is not.

So, as far as making a system to do this, I can say with 100% certainty that it not only is possible, but it could be programmed by an undergrad student at university. In fact, a group of students in the CS Department here did do just that. I did that at one point as well. The core of the system was these classes:

1. Quest class - Represents a quest, its goals, its preconditions, postconditions, rewards, etc. Can also encompass multiple quests (quest chain, etc).
2. Goal - Represents an abstract goal type for a character to achieve. Can encompass complex goals, such as doing multiple parts of a quest, etc.
3. Precondition - Any precondition required for beginning the quest (level, certain item, etc).
4. Postcondition - Any artifacts that occur in the world as a result of the quest (level, area change, changes in people, etc).
5. Rewards - I made this separate of postconditions, in that rewards are what only interact with the completing characters. This includes things like experience gain, item gain, alignment shifts, etc.

Now, using these, it is possible to create a completely random and logical quest system. But does anyone see the main problem with this? If you look hard, we have everything to dynamically create quests, except for one thing: Dialogue, Narrative, etc.

So, you may think that is the easy part. But in truth, for a system like this, that is the hardest part to accomplish. For example, it is very easy to create something like this:

"Go kill ten flying monkeys. Then fetch my ruby slippers. Then come back to me. I'll give you ten gold and a shovel. Then help me bury Toto. I'll join you."

However, it is much, much more complicated to create something like this:

"Oh, I really need your help! The flying monkeys of the horrid wicked witch have gone too far. They killed my dog, Toto! Even worse, they ran off with my ruby slippers, and I am stranded here without them. Could you please go and kill all ten of those horrid monsters and bring back my slippers? I'll reward you with all the gold I have and the Tin Man's shovel. Please, I'm begging you!"

"Oh, thank you for returning them to me. I am ever so grateful. If you could be so kind, could you do me one last favor? I was going to, but I just don't have the heart to bury Toto."

See the difference?

So, in all of those, I bolded the actions required to complete certain stages of the quest. You need to kill the ten flying monkeys, fetch or bring back the ruby slippers, get the reward, and then bury Toto. The specific objectives I underlined, as with the number and name of the creature to kill, what you needed to fetch, and what you received as a reward. All of that content could be replaced very easily in both instances. However, the problem goes with randomly creating the second piece.

If every NPC had that exact same dialogue for this type of quest, it could get really old, really fast. So, I could expand it to being 2 dialogues, then 3, then maybe X per race, Y per gender, etc. The point is, no matter what numbers I choose, eventually the system will become stale. Hence, we really need to randomly create quest narrative to solve this problem, but also do it in a believable manner.

Hence, the idea of random quest generation is really solving 2 problems:

1. The first is to solve the problem of building the quest.
2. The second is to solve the problem of building the narrative.

The harder of the two is most definitely the latter. To solve that, you would need a narrative engine for creating your quests. Even more so, you would need it to be flexible and update-able with new quest types, new dialogue and item entries, etc. Lastly, everything in it would need to make sense, contextually, for the game. And even after all that is done, you would need to update the narrative options, as any system is bound to get stale after a while. Add on top of this multi-language support, etc, and you can see the issues with this.

S, that is my rant of random quest generation. I think the idea has great promise, but at the same time, as players and designers, we need to really note what is the meat of the problem. It is not in assembling a quest, but in making the quest believable.