Designer Dave Rickey (whose credits include EverQuest, Dark Age of Camelot, and now Wish) notes that increasing bandwidth makes deeper AI available in large social games. Up to now, AI basically just decided whether to fight the player and then how to fight badly enough to lose, over and over. And over. Other NPC actions were scripted: "Take this bottle to my sick cousin and return for a reward." Rickey urges designers to re-think their inattention to AI. That got me excited about Wish as a game.
Perhaps the future of AI in these worlds will be a 'massively social AI' that involves the users in dynamic relationships among the NPCs. Extensive AI can create events that generate emotional investment: a spouse is seduced, a child abducted, a land ravaged. It could also create the possibility of reward and resolution, by yielding to the user in the right way, at the right times. And a massively social AI could create a level of collective immersion that validates these emotions and makes them real. By making a large number of people feel that these events are meaningful, the AI can make its emotional gifts that much more credible: As a user, I am happy not only because this robot says I did well, but also because other players - real humans - attest that what I did was hard and also valuable. Ultimately, a social AI could create a home-like atmosphere: a group of people who know very much about me, who know what a hero I am, who call on me for help, and who cheer when I return.
A fascinating "massively social" game (A Tale in the Desert) I recently discovered, seems vaguely appropriate to mention here. While I haven't yet a chance to play it - it sounds innovative and quite the grand "social experiment" (virtually speaking).
One detail which I gleaned from website - of which you reminded me here - is the one where your leadership skill (player discipline) improves with affirmations from other players in the form of shrines built to you and votes against artifacts you build online.
I'd love to hear more analysis here by someone who has spent more time here.
-nathan
http://www.atitd.com/
Posted by: Nathan Combs | Sep 15, 2003 at 22:42
I played ATITD for three months or so. It has a lot of new things. Starting with a pure barter economy, and no tech (literally), players work their way up a Civ-like tech tree. Players coming in later come into an more active economy, but they still have to work up the ladder themselves if they want into the production side.
That said, there are roles entirely outside of producing. There is an ongoing gathering role for rare resouorces. There is room for pure games, within the game. One could play a professional games player and "live" off that.
A group of us set a goal to lauch the first widely accepted currency. That was an interesting project. Last I checked it was still going strong. The issue was to build trust so that in a world where anonymity (for the players) is the norm, with the attendant effects, players would trade "real" virtual goods for a piece of scrip. It was very interesting. I kept waiting for someone role-playing a closet bad guy to turn up in our little core. None did. That would have set back currency a long time.
That is a very social world too. There seem to be fewer "griefers" per capita when there is no real combat. The game is about building utopia, not anarchy, or distopia. Utopia, even within the game, is unlikely to emerge, but it has lead to some real wide-scale cooperation such as I haven't seen before.
And "customer support" in ATITD is no oxymoron as it is in most online games. That may be because it's smaller and tighter. It's a Mom & Pop game compared to the superstores that are EQ, DAoC, SWG, etc.
Posted by: Dan | Sep 16, 2003 at 00:34
Regarding the need for more analysis of ATITD: very true. Not too many people outside this area understand how time consuming the study of these places is. We are all doing personal observation studies of entirely new planets.
We also need scholars to go to Korea and report back.
Posted by: Edward Castronova | Sep 16, 2003 at 01:09
Speaking of Korea, I believe Dark Ages, by Nexon, is a Korean game. I played that one a while and was interested by the player-run government system.
One additional note on that one is they award in game status for out-of-game contributions to the game "literature." In order to progress in government, you need a certain amount of status from having made some contributions of this sort. That means only those truly invested in the game, emotionally and intellectually, can have a role higher up in government. It's by character also. You have to choose which of your characters (if you play more than one) is authoring the contribution and the status award applies only to that character.
Posted by: Dan | Sep 16, 2003 at 02:21
I suspect that the MMORPGs which support a clan/guild structure have already gone beyond the stage of the psychological success payoffs being based on what the NPCs say.
The esteem of my guildmates after leading a successful raid is worth far more to me than the loot or ingame advancement. My clan welcome me when I appear, and are disappointed when I am not around.
Although my avatar is a partially reworked variant of my real "self", the realworld psychological value to me of their esteem is only slightly dimmed by that pretence and distance.
In the higher end of games like Everquest, continued ingame success is not just based on game technique, but also in being able to sustain the social relationships necessary to remain in favoured membership of the "uberguild"s.
The game design decision to require co-operation among players, by restricting numerous key skills to small subsets of characters (class skills) seems to have already engendered the kind of mutual valuing which Ted foresees for a social AI.
Posted by: Estariel | Sep 17, 2003 at 07:18
Regarding Estariel's point about group affirmation (being the greater draw). I agree, but wonder if this is still a flawed model in online worlds?
I was struck by a point in Shirky's article "A Group Is Its Own Worst Enemy" (see sidebar) -we, players, would appear to better manage social reputations (or at least better for us) than could the online mechanisms (incl. in-game).
I often have thought that one problem with at least the "traditional" MMORPGs has been their high-rate of decay in "social memory" - too much churn in the societies. Not only do players drop in, drop out, create new characters. But also society tends to balkanize into level bands (high level parties group with high levels) etc.
At the very least, only those with large positive reputations might be incentivized to invest for the sake of reputation, whereas, given the ease of anonymizing one's history, there is little disincentive for those failing to invest in their reputation.
I'm simplifying away the perverse case (d00dz etc).
Posted by: Nathan Combs | Sep 17, 2003 at 11:08