Jump to content
Posted 19 November 2012 - 10:38 PM
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-advisor of the GDC AI Summit
Co-founder of the AI Game Programmers Guild
Author of the book, Behavioral Mathematics for Game AI
Posted 13 March 2013 - 01:56 PM
To point out the obvious, the primary goal of a player who watches YouTube videos to find the "Magic Method" to defeat a boss and to farm it, is not a dynamic, challenging experience. They may want to experience the whole game in a short time period, to prove they can excel at something, or to generate money on eBay. Even back in the early console days, many games came with cheat codes, and it was up to the player whether to use them. Whatever best entertains each player. http://public.gamedev.net//public/style_emoticons/#EMO_DIR#/smile.png
AI in raid encounters use 1 strategy once players find out how to counter it they post it online in a video everyone reads it and spends a few weeks practicing and then the boss is on farm status. There doesn't have to be a "Magic Method" of defeating a boss you've just been brainwashed into thinking that's how it should be.
Posted 13 March 2013 - 10:52 PM
I was have been looking at a game that will be coming out this month and after alot of the game mechanics were shown on the way too many trailers and interviews, it was obvious that by randomizing game factors (plot scene order, weapon/powerup mixes, opponent spawn content, props to effect cover) they could vastly change/vary the players game experience.
That would be good for replay (needed with the shortening play-thru times). That would also be without require much/any costly expansion (expense $$$) for additional choreographed special scenes or flavors of opponents. There are always plenty of generic situation in between where simple combinatoric options could be applied).
But no, apparently nothing remotely like that is to be there, just the usual 'led by the nose' fixed path and scene and opponents.
Posted 16 March 2013 - 04:05 PM
Emergent behavior can be suprising and the game companies dont like spending alot of money on QA. Tweaking coefficients and threshold values is alot simpler than requiring a programmer to fix balancing issues.
Procedural recombination of a scene is especially easy when all the candidate options are already vetted : put object A at one of this list of possible precanned locations. Now the player has to shift where they stand for cover each play thru or what explodable object NOT to stand next to.