One of the holy grails of games are random quest (mission) generators that do not seem repetitive. Part of the problem with this, is that there really are only a few types of quests. Its just the dressing that changes. Quest types Go someplace and then: Interact with someone/something Find something/someone and return Kill/destroy something/someone Protect something/someone Deliver something/someone Survive Video games have a few variations, such as go kill x creatures, but there really is not many more generic quest types. And if you come up with one I bet it would most likely fall into one of the above quests. So how do you take these 5 generic quests and make them into an interesting system that generates random quests? I’ll talk about this in a few steps. Step 1, Mix up the types. This creates a series of quests that have two or more objectives from the above list. Go from one main goal, to two, or even three main goals. Example: Protect someone, while you Kill something, so that they can interact with someone/something afterwards. Step 1a. Create some quests that are multi-tiered. Don’t make every quest go here and do these 1-3 things. Make some quests be Go here do 1, then go here do 2, and finally go here and do 3. At any time you may use steps 2 and 3. This will give you many types of quests. Step 2, Toss in the random surprise. Basically have a random chance that the mission will change while in the middle of the mission. Don’t do this too randomly basically a 5-10% chance (once every 10-20 missions.) Basically take the given mission and then switch it to one of the other sets of conditions. Example, You are going to kill someone/something in a dungeon, and then someone hits self destruct, you now have to survive the situation. Step 3, Link SOME Quests together. This is accomplished by tossing in a tag on certain missions to generate a “bad guy” in the background. This guys name is randomly generated, but is tagged and then logged into the player’s logs (account.) The next time this player should end up with another “bad guy” quest, the game instead tosses in this guy’s name, and then logs the tally of the number of quests completed against this bad guy. After a predetermined (either programmed or randomly generated when the badguy is first tagged) number of missions where this guy is tagged, the player will be offered a mission that includes as one of its objectives (kill someone) where the someone is the generated bad guy. Give this quest a higher than normal (say 10-20%) chance of having the random twist. If the twist does occur, do not kill the bad guy but the quest count against him is reset, and then starts over. This will give the player the appearance of an arch villain. Step 4. Random, instanced, dungeons. In my opinion this is a must. And it must be worked well. The system that is used in COH isn’t bad, nor is the one from Anarchy online. I think they could be better, but for now I’ll stick with this. Step 5. Sometimes ignore step 4. This makes it feel as though the mission system is part of the game world, and not just a random add on. Step 6. This is the hardest part, but probably one of the most important. Get some good writers, and have them creates some decent story pieces that can be randomly tossed together to form a basic story. They don’t need to be nobel prizes in literature, but they do need to be more than the same 5-6 stories with a name change. Step 6a. Take the story and place it in the mission. Do not just give the mission in an over-view at the very beginning. Put NPC’s in that help push the story along. If you put these six steps together I think you will have a large system with a huge amount of quests. What do you all think? Random Mission Generators usually end up being a large portion of a players experience because they give limitless content, but always seem to feel as if they were slapped together and tacked on to the game at the end. They need to have more done to them.

I guess pretty much every game designer thought about this at least once. The problem is, that many small objective don't make sense when they are put together.

Killing the dragon can make sense with saving the princess or retrieving the treasure, but it will take a purposed story line to make a kill the dragon and deliver the gift mission. Random mission generators can not create such story lines.

You suggested to get writers to write many story parts, and than place those parts in the missions. I think you missed the idea of random mission generators.

Instead of quests types, you could give *every* object attributes to describe quests, based on which actions are being performed on the object. When the actions described in the quest attributes are done, you complete the quest.

That way, your quest types would only be limited by the things that can be done with your objects.

About multiple objectives(step 1): personally, I slightly dislike when objectives are added after I complete current objective, specially if the quest is boring. When I complete the quest, I expect my reward, not "you have to do something else before I give you the prize". Unless I get rewards for each step... however if you give me all the objectives right off the bat, I'm cool with it.

Now, also another thing could be that quests could have more than 1 outcome. For example, when you wrote about having to activate a self-destruct mecanism. In that case there could be more than one way of escaping, and you could get different rewards depending on the path you choose, discover new lands, or even get another quest.

There are two basic elements to any quest or mission: context and purpose. The purpose is the easy thing to specify: go to Strongbadia and defeat the evil dragon-wizard Trogdor. What you've covered here is a very good look at purpose, which is typically what the vast majority of mission-generation systems focus on. Mission generation is attractive to developers because generating a purpose is very easy; as you noted the options are limited and most activities boil down to a few core actions.

However, this is a siren call. Missions are not compelling because of the purpose; the purpose of a mission is an incidental, a time-consuming thing that in and of itself is not really interesting. A great place to see this is in RPGs, where many players complain about "grind" - the process by which you do a lot of purpose but for no clearly defined reasons. Open-ended "sandbox" games like GTA or the X space games also suffer from this problem. Dynamic mission generation has a severe problem in that it tends to ignore the more important half of content, which is context. The whole issue stems not from there being too few basic actions for the player to do, but from the fact that the player doesn't feel any compelling reason to do them.

The solution I think here is not to think more about generating purpose; the solution is to think about generating context. Purpose is the actions the player is meant to do, their reason for being a part of the quest. Context is a larger thing - it is the reason for the quest existing in the first place. Context tells us why things are in their current state, as well as giving us a clue as to what state(s) would be better.

Purpose tells us to go cast a fireball spell onto an ice statue of a wolf in the High Mountains. Booooring. Context tells us that this wolf is, in fact, the pet dog of the player's character, who has been taught to speak, and can reveal the secret weakness of Badguy the Archwizard. The wolf was frozen by Badguy's ice spell, and the only way to get him out without killing him is to counter it will a similarly potent fire spell. Hey, now we've got a reason to go do this thing, and suddenly it isn't a stupid mundane task anymore.


Dynamic mission generation succeeds very powerfully in generating purpose. Where it tends to break down is in generating good context. For instance, in the game X2, I developed two missions: a generic "assassination" mission, and the so-called Xenon Invasion mission.

In the first, a Wild-West "wanted poster" type message is posted to space stations in the universe, and the player can contact someone to contract as a killer for the target. The purpose is to go fight a heavily-armed fugitive or something. But the context is virtually non-existent. There are five or six canned backstory explanations for the whole mission, but the actual mechanics are always the same, so after a few runs the context is totally useless. Most players notice this quickly, and it eventually ruins the entire mission. Over time, the same purpose (kill something) could be used thousands of times to great effect, but it gets boring after a dozen runs without sufficiently rich context.

The "Xenon Invasions" were added as part of a downloadable update to give players something to do with large fleets of capital warships. Basically, a bunch of aliens cross a border into frontier space, and start shooting. The player gets paid a bounty for each kill, and gets to take part in a relatively huge space battle. The purpose was good, and in fact many players have remarked that it extended the enjoyable life of the game significantly for them. However, again, the context was weak: it was always the same aliens, always the same basic fleet of ships, the same tactics, the same small frontier border areas that got invaded. The context was static, and so the dynamic and spontaneous nature of the invasions became rather pointless after three or four runs.


My personal opinion is that we've got perfectly good technology for solving the problem of generating purpose. Where the focus should really go is into generating interesting contexts. After all, the separation between boring, machine-generated quests and human-crafted ones isn't defined by what the player spends their time doing - the difference is in why the player feels like they are doing it.

One thing is that there is a good enough technology to do something and another that it is actually used correctly most of the times...

Since ApochPiQ is more interested in the reasons behind the existence of the gameplay... I would enjoy some silly reasons(please rescue my cat, I fell on a banana and I dropped it through this hole, which is full of monsters), but I wouldn't like it if most reasons are.

Also I do not mind if I do some quests just for the reward(a random NPC just told me that if I fetched him 10 cat tails, he will give me 100 gold coins), but I wouldn't do many of those because I want to know if my effort is worth it. In real life I could ask the NPC why he wants the cat tails, and if he doesn't wants to tell me it would be fine. But if people kept asking me to do stuff without a reason, while I do not ask(because in a game I cannot ask an NPC to do something for me), I would eventually stop asking for reasons, because even if I do they aren't gonna tell me, which would give things a more silent and mysterious view. Then a person from another country which is more lively(a new player who comes from real life) just came to my country, it would feel uncomfortable with the enviroment for that person.

Now here's another thought about random quests: encourage players to give quests to other players, and that those quests are interesting. While most players won't give interesting quests, it will make things less boring and some players might actually be good at giving interesting quests.

Quote:

Original post by someboddy I guess pretty much every game designer thought about this at least once. The problem is, that many small objective don't make sense when they are put together. Killing the dragon can make sense with saving the princess or retrieving the treasure, but it will take a purposed story line to make a kill the dragon and deliver the gift mission. Random mission generators can not create such story lines. You suggested to get writers to write many story parts, and than place those parts in the missions. I think you missed the idea of random mission generators.

If every game designer has thought about it then why do they still suck?

The thing is that they can work together. But that is where context needs to come in. Obviously you are thinking take a gift to the dragon and kill it. But perhaps its take a gift to someone in the dungeon who (a. opens the door) or (b. gives you the trinket needed to let you hurt the dragon)

Yes I mentioned writers. They are ESSENTIAL to a random mission generator that is going to take alot of time in a players game. I didn't say they needed to write the missions. I said they were going to write stories for missions. BIG Difference.

Have you ever played ANARCHY ONLINE? Basically every mission of the same type has maybe 2 or 3 story covers providing you the slightest amount of context for the mission. What Im suggesting is basically story templates that can be tossed together with the objectives and missions to give the thing context.

This changes the random mission from "Go to position X and Kill all Y enemies" to "Help! (player) the (location) is being robbed by (enemy name)! You must stop them before they get away."

Now if you're going to have multi-step quests the story needs to be a little more fleshed out. But thats what the writers are there for. In a stand alone game, you need to have enough variety to keep the player going for a long time. For an MMO style game you need to constantly update both the map sets and the stories to keep it fresh.

So yes I understand the concept of RANDOM. You apparently did not read and comprehend what was originally written.

There is a system for generating random adventures just like you describe in the Dungeons and Dragons 3.5 "Dungeon Master's Guide" book. They provide a table with scenarios for different quests and a table for "twists" in the story. Then you can use other tables for random locations, random enemies, random items etc.

True, but it is still up to the Dungeon Master to make up the story.

There's a table for that too! I'd post an excerpt if not for copyright infringement but they're mostly like the OP suggests:

They don’t need to be nobel prizes in literature, but they do need to be more than the same 5-6 stories with a name change.

I agree with the point of this thread though and I think using computer algorithms to play the role of a Dungeon Master or Game Master (ie. creating adventures for the player) could be done a lot better than they are at the moment.

What you need to realise is that you are trying to make a computer do a creative process. A quest generator is like an automatic novel writing machine. The best you can hope to achieve is to find a good compromise between repetiveness and non-sense.

You might want to look into "scenario-based" artificial intelligence. It is a great concept that is under-used.