The twilight bark of Nintendogs is loud these August days (e.g. [1.], [2.], [3.])...
I knew after Satoru Iwata's keynote at GDC 2005 that we would likely be talking about them and that I could end up with one or three. What was less clear was why.
Clive Thompson might be onto that answer when he wrote (emphasis his) :
I think the appeal is much subtler -- and weirder. If we love Nintendogs, it's not merely because they're so adorable. It's that they're so needy... The thing is, this precisely inverts the normal logic of artificial intelligence. Back in the '70s, everyone assumed we'd eventually have super-smart robots as servants -- guarding our homes, managing our schedules and bringing us a beer. That never happened. Nobody really wanted robots like that, because robots like that are kind of scary. Nobody wants a Terminator hanging around the kitchen.
I'll go one step further for the sake of discussion. To what extent is the games AI malaise a product of a myopia to create Terminators rather than Nintendogs? From a games AI perspective, would it be easier - and would easier first steps encourage more sure-footed second steps - to design virtual game worlds which are in some sense more needy than confrontational?
Part of the advantage of a needy AI may be you. You want to believe. Empathy turns to sympathy and much is forgiven by the player: you work harder. In contrast, if that bot is glaring at you with its trousers-inside-out, no sympathy, no respect -- "what dumb AI."
Perhaps.
The danger is that this needy revisionism places the cart-before-the-horse: shouldn't game design be the dog that shakes the AI tail, and not the other way around. Yes, but can this flip on occasion? Are there scenarios where the AI is so engaging that it can reach out and grab you - thus carrying a mediocre game world and its cast of denizens? Rome burns while you are having a great time wrestling the Labrador...
Can the tail ever wag the dog?
...If it is true, as Andrew Stern suggested, that games AI has a long slog ahead of it (see comments, here):
I'd suggest game AI is not in a rut, per se - instead it's standing on a slope, part way up a massive mountain, looking up at the distant peak, gone as far as it can go on foot, and now has to break out the mountain climbing gear and prepare for some very strenous, painful rock climbing.
If it is going to be a slog, a few victories upon the incline would be helpful for all of us. Here is a toast to needy AI and all the associated Bow Wow...
Recently, game design has become subdivided into numerous methods, not all of them requiring specific game design skills. Companies like Id show that inventive game design is not always necessary - that one can simply craft games around new technology (although some design skills are needed to make this happen). AI is in the same camp, I believe - if the audience will accept games which distinguish themselves only on superior technology, then it follows that it is possible to build a game which draws in via AI.
It does seem perfectly logical, as you suggest, that robots that need us are more appealing that robots that do not. Our dogs do not threaten us because they depend upon us and love us unconditionally - successful domestic robots would do well to follow the same social interaction.
As for the draw of 'needy' AI - the audience (at least as far as our current model suggests) is already broadly divided into those players for whom agon (competition) is an essential play need and those for whom emotional engagement is an essential need. The market for the former is fractured and oversubscribed, while the lack of product competition for the latter (read: the success of The Sims) suggests there are opportunities there one could take advantage of with the right game design.
The problem? The evangelists who promote games to a wider audience are generally more motivated by agon than emotional engagement (There are exceptions - but generally not enough of them). Perhaps this reflects the 'boys club' nature of the specialist press.
Posted by: Chris Bateman | Aug 30, 2005 at 03:56
I don't think needy is the differentiating characteristic. Recall how many Tamaguchis were abandoned over the years.
I'm not a psychologist, but it is probably our desire to relate to something familar and safe, like the play we had as a child with a pet. Emotional engagement may be the right world for this.
I recently watched the anime series Ghost in the Shell: Standalone Complex that bring up these issues. One question that it raised in my mind that relates here is what happens if your mind and soul was placed in the body of a terminator robot or a dog or a box. How would a stranger react to your appearance and behavior? This has been the basis of many stories and is interesting.
Anyway, I think AI currently are smart enough, they just need to have more soul, or "ghost" as used in the anime series. We can program AI knowledge, but it's the tacit application of knowledge and wisdom that's the new frontier in AI.
Frank
Posted by: magicback | Aug 30, 2005 at 10:41
To what extent is the games AI malaise a product of a myopia to create Terminators rather than Nintendogs? From a games AI perspective, would it be easier - and would easier first steps encourage more sure-footed second steps - to design virtual game worlds which are in some sense more needy than confrontational?
I agree with Frank that this is more a matter of "emotional engagement" than neediness. Creating emotional and emotionally engaging AI may or may not be easier than creating strong tactical AI, but I think it's likely to be much more satisfying to a broader variety of people than what we see now. I believe this will also enable new kinds of gameplay that we don't really see now -- including melding the achievement vs. socializer or competition vs. relationship camps, something we could definitely use as Chris Bateman notes.
As for Nintendogs, it's unclear to me how long they will remain engaging. At what point will you sigh and turn off your pup for the last time? With such narrow interaction, our attention and loyalty can be capricious and fleeting. (I'm reminded of the last line of "The Truman Show," just after Truman has heroically left his sheltered life to the stunned stares of the audience. Without missing a beat, one long-time viewer says to another, "see what else is on.")
I suspect that to make emotional AI truly engaging over the long haul, it's going to have to include signficant, emotional interactions between AIs (beyond barking and playing) and more importantly between humans. I also believe that this is going to change how we view games at least as much as did 3D or persistent online play. At that point, in terms of AI and game design, it becomes unclear to me which part is cart and which part is horse (or which is tail and which is dog).
Posted by: Mike Sellers | Aug 30, 2005 at 11:37
Nintendogs, Tamagatchi, Pet Rocks, Furby, Imaginary Friends, I kind of lump them all together into a single progression, driven by technology. They all fill a temporary need, one that seems to become popular every five or six years, but then things go back to "normal", such as it is.
In my opinion, emotionally engageable AI is useful for strong narrative and immersion, but as such, no more nor less important than strong tactical AI. They are sort of the Ying/Yang of immersion to me. I've got to believe what the AI is doing (ludite?) and why it's doing it (narrativist?) to feel truly immersed. While they require very different specialties and talents to get right, it's up to the Leads to identify the balance and manage it lest we get games with unbeatable combatants or overly high maintenance neighbors :)
Posted by: Darniaq | Aug 30, 2005 at 16:48
Wasn't there Mutz and Catz once upon a time?
As for the draw of 'needy' AI - the audience (at least as far as our current model suggests) is already broadly divided into those players for whom agon (competition) is an essential play need and those for whom emotional engagement is an essential need.
Well, I don't know that these are mutually exclusive. I'm thinking that the previously discussed Neopets for example works by playing on both or either factor as the player desires.
But I think there is something to the possibility that "needy" NPCs have been extremely successful as a narrative hook, and inspire some of most intense modding efforts. Just getting paid off by someone who hires you is less satisfying than rescuing someone you care about.
Posted by: Kirk Job-Sluder | Sep 01, 2005 at 16:14
I think that Nintendogs is a really, really fun
game! But I really wish you could buy as many
dogs as you like instead of just 8. But
still, I have a reason why this is the best
DS game ever: ALL of the people that I know
who have DSes, all have Nintendogs. Like I
said, Nintendogs is the best DS game known to
man and DS kind!
Posted by: Morgan | May 23, 2006 at 14:16