Even a hard-pressed, overworked design team is quite capable of setting up quests that are far better than anything the computer can come up with itself. A full-blooded quest system should therefore unquestioningly allow designers to add new material to the game as it runs. To do this effectively, they must have available to them a set of tools to take out the grunt work. Integral to any online component of this could be an interface to the automatic system that I’ve been describing.

This works by allowing quest designers to set the goals and partial plans of mobiles to create situations that would not occur naturally. For example, consider a game where town mayors might be able to issue newbies-friendly edicts such as offering to pay 1 gold piece for every 10 rat tails brought to the town hall. A designer might take advantage of this by setting up a mobile who plans to run a covert rat-breeding farm. This would be a one-off quest that the game almost certainly wouldn’t be able to generate on its own. However, once the basic goal is in place, the mobile’s planning system can take it from there. Thus, a player going to the mobile for a task might be told “cause the mayor to issue this edict” or “take this money and this note to the mayor” or “kill the mayor then come back for further instructions”. Once the rats-for-gold edict has been made, the mobile could then ask the players to capture live rats and bring them to the secret farm, or get them to kill farmed rats and take them to the town hall for money.

The planning mechanism is a way for designers to set up quests without having to flesh them out in full. The designer can assign goals and create plans to whatever level of abstraction is necessary for the desired effect, but the mobile can flesh out what happens after that itself. What they do depends on the level of the player (or party of players) asking for the quest and on the nature and personality of the mobile itself.

In other words, you can have the best of both approaches if you do it right.

Some loose ends: problems with quests

The system as I’ve described it sounds, on the face of it, pretty neat. Assuming you can create enough actions (ie. quest components) then mobiles should be able to string them together in interesting ways to solve complex, context-sensitive plans and generate meaningful, achievable goals for players.

Even granting that somewhat large assumption, that doesn’t mean dynamic quests are a breeze. If it were only one player playing, then all would be well. However, the game world is shared with Other People; they don’t act like they’re supposed to act...

If I’m wandering through the mountains and come across a bunch of savage ape-men, I may well consider it my civic duty to send a fireball their way lest they build up sufficient numbers to mount an attack on the local village. I don’t necessarily have any inkling that there’s a daring knight heading this way already. The knight on the quest is going to call the game buggy when he arrives at the ape-men camp and finds no signs of life, but as a passer-by I would call the game buggy if I tried to kill the ape-men and it gave me no good reason why I couldn’t (especially if I could subsequently watch someone else slaughter them with impunity).

My having hijacked someone else’s quest, there’s then an issue of who gets the reward. Can the knight immediately go back to the village elder and claim that he should be paid? After all, the ape-men are all dead. Or can I claim the reward myself because it was me who killed them? Should the answer depend on the personality of the quest-dispenser? Or are all quests somehow “public”, and whoever gets back first after the event can claim the reward?

What about partial quests? If the real aim was to rescue the village elder’s father but he fried in my fireball, could I or the knight nevertheless claim part of the reward? What if the knight did rescue the elder’s (even more elderly) father, but when he returned he found I’d accidentally killed the elder who’d given him the quest?

There isn’t really an answer to this. The problem can be alleviated in part by building rewards into the required actions, so if you kill the ape-men you get their treasure but only if you bring the father back alive do you get the extra money the village elder is promising. This still isn’t going to cut much ice with the party of 10 who have tromped across the desert to wrest the golden Eye of Wisdom from its serpent guardians, only to find that a gang of 40 (player) bandits raided the snake-god temple the day before they arrived.

On the whole, though, players must learn to accept that in multi-player games you can’t have action without interaction; it is, after all, the whole point! Shouting “it’s not fair” at customer relations staff when someone else steals “your” quest is a very short-sighted thing to do. Sadly, though, many players only look to the short term (one reason why the mere concept of persona death is so unpopular), which may mean that sophisticated, multi-level quests are something for the future, rather than the present.

This concludes my series on quests. I'll probably return to mobile AI in a later set of articles.

For the moment, I shall be taking a few months' hiatus to finish a book I'm writing about virtual worlds. I was supposed to have completed it in November, but still have three chapters to go — hence it's “all hands to the pumps” time (unlike Jessica Mulligan, I can't write a book and keep up a series of fortnightly articles at the same time!).

I'll be back (hopefully) in the Spring.

Richard