Mad Libs

Tuesday’s post got me to thinking about how to build quests that are not immersion breakers. Long before I decided that modding NWN2 was not how I wanted to build my next world, I commiserated with Thrym, the lead admin of the NWN persistent world Markshire on about a way to make quests that are not dancing elephants. We were thinking along lines of semi-procedural quests. “Mad Libs Quests” was his term. The idea is generally as follows:

  • You have a pool of quest boss names, from which you draw one randomly.
  • You have a pool of themes:
    • revenge (“bring me a head”)
    • rescue(“my daughter has been kidnapped”)
    • escort(“please take my daughter safely to X”)
    • delivery(“deliver Y to X”)
    • resource(“kill ten rats”)
  • You have a pool of quest locations. You would make lots of nooks and crannies in your world. Then, using whatever tools are available for your engine, you flag certain locations as possible quest pool locations. It is probably wise to share these locations with the random encounter system (assuming there is one. That is a topic for another time).
  • You have a pool of NPCs that are able to give quests. It is not always the same people who are quest givers. A merchant may one day mention that his daughter was kidnapped to a PC who comes to do business. Some NPCs who are “background flavor” NPCs may be sometime quest givers.
  • Sometimes, but not always, the NPCs may use the player’s communication systems to advertise quests. If you have an in character (IC) forum, or an in game notice board, you may create a series of templates for advertising these quests.

We considered ourselves very clever. The only real barrier was the dev time. This kind of solution comes up on a regular basis and whichever designer blogs or forums – be they modder forums, engine forums, etc. you follow always seem to have a thread describing a mad libs scheme every couple of months. Every half baked amateur designer rediscovers this idea at some point

But… all is not well in procedural generation land. Damion Schubert wrote this today on the Mud-Dev2 mailing list:

I’ve worked on autogeneration before, and I now work at a company that champions hand-crafted content, and I can tell you, it is nearly impossible to autogenerate content with the emotional depth and resonance that hand-written content provides. This shouldn’t surprise: after all, we still don’t have algorithms that will write better movies or novels than real humans do. Yet, for some reason we expect this to be true, even though interactive content is far more difficult and subtle to write, and writing for a media platform where consumers are routinely interrupted (logging off, being disconnected, handling another quest, exploring another activity, or helping a friend).

A core problem to solve with the autogeneration is pattern recognition. The best quests in MUDs and MMOs have either interesting stories, interesting activities or interesting characters. One example: a quest in Fallout 3 sends you to pick up the Declaration of Independence, which has a lot of emotional resonance with the player in this post-apocalyptic world. This is not something that a random generator would have created, and if it did, players would likely have also seen quests to rescue the Constitution, the Magna Carta, and the Consumer’s Bill of Rights, all generated with similar explanations and expositions, and that repetition would have destroyed the illusion. Once the players recognize the algorithm, maintaining their emotional hold becomes harder and harder.

The Old Republic is his current project and presumably he means Shadowbane as the game he worked on with procedural content. The man has a point. Nobody has yet created procedural content that could pass a touring test equivalent to what can be hand crafted by a skilled and dedicated designer. If you are building a hardcore deep immersion world, you can get burned by deep, hand crafted quests (as Bioware is doing with the Old Republic and which are always in limited supply and need to be repeated by different players), or once the player has played a few, she’ll catch on to them being created by an algorithm.

We were not nearly as clever as we thought we were and the mad libs approach can kill immersion as surely as everyone repeatedly serially-killing the same bear.


About Dave

I’m a 38 year old American who has lived the past 9 years in Germany and India.
This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

2 Responses to Mad Libs

  1. Edward says:

    And yet mad libs are needed, because a boss should be able to hire a PC to eliminate an enemy. Mobs need ways to accomplish their goals, and this is one way.

    It seems like the themes are key. They should be related to the boss’s actual goals. For example, if the boss has been raided by another boss and can’t simply launch a frontal attack to destroy the offender, he may decide to hire an assassin. If he has agreed to marry his daughter to another boss, but his troops are tied up elsewhere, he may outsource the job of escorting her to the groom (depending on how important he considers his daughter and the marriage alliance).

    Mad libs shouldn’t be limited to PC’s, though. The boss should be just as willing to hire an NPC assassin, assuming the assassin has a good reputation.

  2. Greying Wanderer says:

    I think this sort of thing could be used for generic fedex type quests. They don’t need any emotional connection plus making them this way adds to replay immersion as it makes the game seem less static.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s