Comments on: Procedural Generation and User-Generated Content, Part II: Storylines, AI, and Emergent Gameplay @Claire Blackshaw - great, I am glad you found it interesting. :) I would be happy to read over your draft when it is ready, just send to gavanw at gmail dot com. @Claire Blackshaw – great, I am glad you found it interesting. :) I would be happy to read over your draft when it is ready, just send to gavanw at gmail dot com.

]]>
By: Claire Blackshaw/2011/04/06/procedural-generation-and-user-generated-content-part-ii-storylines-ai-and-emergent-gameplay/#comment-2542 Claire Blackshaw Sun, 10 Apr 2011 15:25:40 +0000 The simplest example I can think of is two nodes representing butter factories, linked by 3 nodes in between. A mercenary is able to travel 1 node at a time, and burns say 2 butters a day plus 3 to move 1 node. The butter factories are racing to reach 100 butters total, and can put out private contracts for damaging the other factory so the mercenary would figure, move_attack :- biggest (bounty subtract cost to do so) > sitting still unless cost to do so would kill myself and butter factories would figure, cost to keep mercenary away versus potential to win faster The simplest example I can think of is two nodes representing butter factories, linked by 3 nodes in between.
A mercenary is able to travel 1 node at a time, and burns say 2 butters a day plus 3 to move 1 node.
The butter factories are racing to reach 100 butters total, and can put out private contracts for damaging the other factory

so the mercenary would figure, move_attack :- biggest (bounty subtract cost to do so) > sitting still unless cost to do so would kill myself

and butter factories would figure, cost to keep mercenary away versus potential to win faster

]]>
By: Gavan Woolery/2011/04/06/procedural-generation-and-user-generated-content-part-ii-storylines-ai-and-emergent-gameplay/#comment-2463 Gavan Woolery Thu, 07 Apr 2011 21:37:36 +0000 Roy - Yes, humans are very specialized at acting like humans :) Most likely, even in MMOGs, you will have NPCs, because there are certain roles humans just don't care to play. All of the concepts I have laid out are actually fairly easy to implement, and the AI techniques to do so have existed for several decades. The AI makes deductions through propositional logic and backwards chaining (as with Prolog), and takes action based on what expected result will yield the highest potential "score" towards their ambitions (each ambition is weighted differently, i.e. staying alive has a very high weight, whereas something like eating food would have a much lower weight - unless they were near starving to death). Generating good fiction involves a deeper understanding of the English language, but we are approaching that as well (as seen by IBM's computer playing Jeopardy). What I am speaking of here is only generating a coherent set of actions that make sense according to a characters ambitions, and allowing fragmented dialog that is basically an abstraction of these actions and the AI's database. Roy – Yes, humans are very specialized at acting like humans :) Most likely, even in MMOGs, you will have NPCs, because there are certain roles humans just don’t care to play. All of the concepts I have laid out are actually fairly easy to implement, and the AI techniques to do so have existed for several decades. The AI makes deductions through propositional logic and backwards chaining (as with Prolog), and takes action based on what expected result will yield the highest potential “score” towards their ambitions (each ambition is weighted differently, i.e. staying alive has a very high weight, whereas something like eating food would have a much lower weight – unless they were near starving to death).

Generating good fiction involves a deeper understanding of the English language, but we are approaching that as well (as seen by IBM’s computer playing Jeopardy). What I am speaking of here is only generating a coherent set of actions that make sense according to a characters ambitions, and allowing fragmented dialog that is basically an abstraction of these actions and the AI’s database.

]]>
By: Roy Zhang/2011/04/06/procedural-generation-and-user-generated-content-part-ii-storylines-ai-and-emergent-gameplay/#comment-2460 Roy Zhang Thu, 07 Apr 2011 21:00:55 +0000 All the things you're asking of this hypothetical A.I. (i.e. gimme a quest, talk to me, surprise me, make rational decisions) definitely veers towards the things humans are especially specialized for. That's why mmo s with functioning economies exist, and why D&D requires a dedicated storyteller. Heck, we can't even use A.I. to generate halfway decent linear written fiction yet. (citation needed) Then again, ambient A.I. chat kinda works surprisingly well, like in this articlehere http://discovermagazine.com/2007/brain/i-chat-therefore-i-am/article_view?b_start:int=1&-C= All the things you’re asking of this hypothetical A.I. (i.e. gimme a quest, talk to me, surprise me, make rational decisions) definitely veers towards the things humans are especially specialized for. That’s why mmo s with functioning economies exist, and why D&D requires a dedicated storyteller. Heck, we can’t even use A.I. to generate halfway decent linear written fiction yet. (citation needed)
Then again, ambient A.I. chat kinda works surprisingly well, like in this articlehere
I think this is a very interesting idea. Assuming you handle balance well using some simple metrics like making sure that the wolves only do so much damage to the mercenary so you don't create frustrating situations (might even be able to adjust conditions based upon difficulty selected) this is a pretty neat concept. I suppose it depends on how many different scenarios and combinations that could be created, but you might have the possibility of different plays resulting in differing levels of quality depending on what options are available to the player based on the scenario they get. I suppose lots of care just needs to be taken into account when setting up the algorithms. Something like this would probably need to be tried out on several smaller scope games first just to see what the difficulties and surprises of this approach would be. But it would be very cool to see the results! I think this is a very interesting idea. Assuming you handle balance well using some simple metrics like making sure that the wolves only do so much damage to the mercenary so you don’t create frustrating situations (might even be able to adjust conditions based upon difficulty selected) this is a pretty neat concept. I suppose it depends on how many different scenarios and combinations that could be created, but you might have the possibility of different plays resulting in differing levels of quality depending on what options are available to the player based on the scenario they get. I suppose lots of care just needs to be taken into account when setting up the algorithms.

Something like this would probably need to be tried out on several smaller scope games first just to see what the difficulties and surprises of this approach would be. But it would be very cool to see the results!

]]>