Posts
This point is something I have really struggled to get my head around. I would just love to know how in Left4Dead they account for four people being in a team, all with their own stress levels. I developed a single player game, a battle against an AI. This allowed me to adapt the AI itself, making it smarter or dumber, faster or slower, luckier or unluckier as required. It's easy to see how and what can be adapted in a single player game, but in multiplayer the issues are vast: differing emotional states, differing play motivations, differing skill levels, and differing styles of play. Player motivation determines whether a player wants to rush through a game to get to the boss fights, or the good loot, or whether they are more leisurely in their progress, values exploring the world, reading quest logs, studying loot etc. Skill levels obviously changes how quickly a player gets bored or stressed out of their mind, how do you deal with that in a multiplayer setting? Styles of play is something ahoodedfigure talked about at my GiantBomb blog, whereby players can be very destructive, and all they try to do is break the game, whereas others just play it within the frames of what's intended.
I truly don't know how to account for all these things when attempting to adapt to the player's emotions. Doing an average could nullify the emotional states altogether, keeping nobody happy. Giving one priority over another will make that player happy and the other frustrated with the game. It's also important whether the game is coop or a pvp game. In a coop it's easier as the players should (in theory) all have the same goal, to progress together. But then again, some people are all about the me, me, me, and others are quite the opposite. In a PvP situation you can't adapt the fight one way or the other, as that would make the more skilled player feel hard done by, and the fight would not really be fair.
Man, I find this bit tricky! Any ideas floating about out there?
Should really ask Valve about this... :P
Maybe you could identify the levels individually and then apply some kind of average for the group? There will be always somebody above or below the average, but that is what usually happens in groups.
ReplyDeleteYou could identify the levels individually, but you might be able to construct responses that suited several of the individuals simultaneously. e.g. In a co-op, enemies could primarily attack the players who wanted tougher fights (or at least spawn nearest them), and the players with lower skill levels would be more able to 'stand back' (or have something constructive but non combat-based to do.) I think this is feasible as long as every activity that a player might want to engage in (tricky combat, finding loot, exploring the game world, exploring the story) in some way contributes to the group's success ('soft' rewards like exploring the world/story would yield experience points, useful hints, treasure, or something else.)
ReplyDeleteI guess there's also various ways to tailor a co-op game to only the individual's experience (players who enjoyed difficult combat might find their game taking place in a more foreboding atmosphere, with less ability to hear enemies coming and more distractions.)
What kind of game are you developing? I think adapting the game to each player style may be a very difficult goal to achieve. Is there no psychology involved in this? I mean, do players not accept that they are in a certain situation and if the game is relatively balanced (and the concept appealing) they will enjoy themselves?
ReplyDeleteIf you ask Valve I think you would probably get the answer "playtesting, playtesting, playtesting", which is probably the only reliable way to find a good game balance. I'm willing to bet that "the director" AI is based on playtesting and then some.
Yes, of course play testing is extremely important, and I'm sure they have done loads of it, but to play test, you need an algorithm to test first. And it is this algorithm I am interested in. I am sure it was both altered and tweaked throughout play testing, so I'd be very happy to hear about the end result, in terms of algorithm for dealing with the various people in a group.
ReplyDeleteI think adapting the game to each player is a very interesting idea, and it wouldn't surprise me if L4D does this to an extent.
I also loved the idea of guiding the AI to focus more on the people who can cope, and less on the ones that can't! :)