Tuesday, June 9, 2009

Perception As Intelligence

I've been a big fan of Douglas Hofstadter ever since "Godel Escher Bach", and one of his ideas is that perception is not only important for intelligent beings, but it's the greatest part of intelligence. His projects often focus on creating a program that can perceive the essential character of a situation. After the situation is well understood, finding a reasonable course of action is often trivial.

I think that's a great piece of perception on Hofstadter's part. How do you figure 24 times 4? You might perceive that 24 is one less than 25, so 4*24 must be four less than 4*25 or one hundred. After you understand the unusual nature of 24, it is a simple task to see that 4*24 is 96.

Martial artists tells many real life stories of wizened old masters easily defeating opponents because of their great perception. Where I might only see a big guy running at me and freak out, a master would immediately see, "He's overreaching and leaning too far forward," or "He's leaving a big opening on his left". Check out this demonstration of a Judo expert comfortably defeating a powerful and skilled opponent. He just perceives the balance of his opponent clearly.

I wish artificial intelligence in games would focus more on perception. The tendency I've seen in games like Halo (And I admit I haven't seen very much recently- I'd like to study more of the small pay-for-download games on say, X-Box live) is to make a very large list of scripted actions and pre-determine the important characteristics of a playing field. An intelligence based on Hofstadter's Copycat analogy engine, for example, would be much simpler and have surprising results. (And be able to be used on custom fields without modification.) I feel that today's smaller games must focus on "emergent" qualities rather than a huge database of scripted actions. An artificial intelligence focusing on perception would be much simpler and still have very clever human-like action.

Thursday, March 26, 2009

Ignoring the purpose of natural reactions

There is a way that games have of being non-intuitive that needs to be addressed. They find an interesting image or experience, and use it for a cheap effect without working it into the game's mechanics. I know that early games couldn't be very realistic, but modern games have no excuse. I'll give three examples that I think are particularly ugly.

The Silent Hill games use a really cool blood and rust theme in many of the level designs, complete with a gross squishing sound when your character walks on some surfaces. But as you play the game, you start to ignore the grossness because it's not important to the gameplay. Why is grossness worrying in real life? It gets you dirty, and bodies and stuff tend to have diseases that you should avoid. But in the game, your character has no possibility of getting dirty. Characters will often jump in freezing water, then be dry and warm the moment they jump out. If getting dirty really does become meaningful in the game... Even if your character could "look" dirty, or if your character sinks into certain surfaces... Any attempt to be true to the underlying reason grossness is "horrifying" will emphasize the horror and provide a better player experience.

The survival horror games tend to abuse this a lot. They use a very frightening or stressful image and ignore the real reason that image is stressfull. (sharp weapons- but they hurt you the same way blunt weapons do. Gigantic enemies-but they hurt you the same way small characters do. Even the knock-back animation ignores the size and mass of the characters involved. Creepy sounds all the time- but it's just the background music, no relationship with real things in the game)

Another natural reaction that is ignored is size and mass. Every fighting game pits tiny characters and huge ones together, and the tiny ones have just as much of a chance. They even fight with the same style as the large characters. A little girl can pick up and throw a 300 pound sumo wrestler. She can punch the sumo wrester and knock him backwards. It is natural to be frightened for a small fighter and be joyful when she overcomes the odds and wins, but these games abuse that natural reaction and ignore the real reason for it. The player's intuition becomes so dull that it doesn't seem unusual anymore.

(By the way, remember Trinity's jumping crane kick in the opening scenes of the Matrix? She jumps straight up and lands straight down, but the cop she kicks flys across the room into the wall... where did all that forward momentum come from?... I guess she was blurring the physics of the Matrix, but I have a sneaking suspicion the fight choreographers are used to getting away with too much.)

One last example that came to mind while slapping this blog entry down is Grand Theft Auto. Why is crime so shocking? Criminals tend to go to jail. Or die. (Live by the sword, die by the sword, and just about half the verses in Proverbs). But the cops in Liberty City seem to be too busy eating donuts to corroborate witnesses and figure out who's hijacking all these cars.

I know that games should give players a sense of freedom and power, but I would argue that acknowledging the natural laws that govern players' reactions will give the players a more satisfying experience. Stretch the laws, don't just ignore them.

Saturday, February 28, 2009

Entertainment Books

I found The Art of Game Design: A Book of Lenses by Jesse Schell at the library, and I think I'm going to buy it. Every page has been an enlightenment and inspiration.

The author worked in juggling and magic shows, huge Disney theme experiences, and many varieties of video game. His love for entertaining audiences of all kinds shines through. The book is very entertaining itself, I guess Jesse can't help himself.

I love reading books by masters of entertainment. I've read several great books by movie directors and short story writers, and they communicate a real joy and love of life.

Tuesday, February 24, 2009

Roshambo, math stuff

Rock, paper, scissors, or roshambo* is a really cool game, and a nice way to add some flavor to a conflict system. Everyone probably knows the game, but I'll describe it for completeness' sake. Two players each choose a "method of attack" independently without seeing the other's choice (usually by doing it quickly at the same time). There are three choices, if both players choose the same attack, it's a draw. otherwise,

Rock crushes Scissors,
Scissors cuts Paper,
Paper covers Rock.

*You can spell it Rochambeau if you want to be utterly pretentious. Or French.

This system of defeats is cool for several reasons. It's completely symmetrical (one other option will beat you, one can be beaten, for every choice) It's "non-transitive". That means that if A beats B and B beats C, A won't necessarily beat C. (Imagine the relationship "greater than". If A is greater than B and B is greater than C, then A is always greater than C. That's a transitive relationship.) Transitive relationships always have a neat hierarchy, but non-transitives can have loops.

These mathematical properties mean that there's no best choice. There is never an easy rule for winning. That's a good recipe for an interesting game.

Another much richer system with some similarities to Roshambo is the Chinese elements. There is a constructive and a destructive cycle which both use five elements:

Constructive Cycle:
Fire makes ashes which are Earth
Earth gives the ore to make Metal
Metal forms Water droplets through condensation
Water feeds Wood
Wood burns to create Fire

Destructive Cycle:
Fire melts Metal
Metal cuts Wood
Wood draws up Earth
Earth dams Water
Water quenches Fire

This system is also completely symmetric and non-transitive, and the mnemonics for the cycles are very suggestive. The five elements are rich with meaning from the traditions of Feng Shui. But it is a bit harder to remember than the three "element" roshambo.

Another thing I might mention, symmetry is not really an advantage. It tends to make the system boring for the users. Only by the symbolism of the elements is a symmetric system made interesting. A slightly unbalanced but still non-transitive system without any dominating element should be more interesting, at least in theory. How about this one? (Called "Undercut", introduced by my hero Douglas Hofstadter).

Undercut:
Two players each choose a number from one to five. The highest number wins the difference between the two choices, unless one player undercuts the other's number by exactly one. Then the undercutting player gets the sum of both numbers. Of course a tie gives no one any points.

This system has no dominating number (any number can be beaten by another number) but a strategy of playing each number 1/5 of the time isn't the best idea-certain numbers have a better payoff and should be played more frequently. Can you guess which number should be played the most?

The system of four card suits with one trump is also surprisingly interesting. The trump is usually restricted to only be played after nothing else can be played legally for a certain user. (the trump would beat any card, so it has to be restricted some way.)

Usually small wrinkles added to a symmetric system can add a surprisingly large amount of strategy, interest, and replay value.

As usual, I'm disappointed by the element magic systems used in most RPGs. it's usually not very meaningful in the game or useful only in a certain place. And the magic systems are usually very boring. Making a system that is interesting and simple enough to remember easily is a great challenge.

Hit points

Why does everybody create games that use hitpoints??!?!? Because it's easy, I suppose. But what real-life conflict has anything resembling hit points? I can't think of one. Certainly not physical fights. Injury is either minor or debilitating. Injuries cause changes in the ability of the attacker.


Another thing that hit points takes away is the mystery of a new opponent. If you are fighting someone you haven't seen before, you have no idea what they might be capable of. You might not know if an attack you just landed had any serious effect. You might not know for sure if an opponent is killed or knocked out. The "hidden power" that every anime hero possesses isn't there!


What happens in a RPG fight against a new opponent? "Whoa, he's got a million hit points!" or "Whoa, he's level 120! (As I can see by the numbers above his head)" We need numbers and names to be written above an enemy's head? Or your attitude might be, "He'll be tough because he's a boss (As I can see by the word 'Boss' in his name)" or you might just fight and let him kill you once to see what he's like, 'cuz you can just rez and fight him again. RESURRECTIONS?.. I know you don't want to throw away a player's hard work, but there's gotta be a better way....


What happens in a more natural conflict (such as a physical fight)? You don't see any numbers, so you look for more subtle signs. He's 250 pounds of pure muscle. or He's really energetic. or he doesn't seem nervous, maybe he's done this before. Caution and estimating your opponent are completely missing from traditional RPGs. The result of your attacks is also judged by more subtle signs (I mean, not really subtle, but more subtle than a flashing sign over your head). And fighting has more serious consequences, so you need to study and estimate unknowns.


Adding subtle signs that can be seen on the screen and adding more serious consequences for your actions might be very difficult to do well, but those changes will drastically improve the game's quality.

Friday, February 20, 2009

Just what are you trying to say?

First post! I have a lot to get off my chest so I'll start right off with some things I feel very strongly about. This post might have a bit of a ranting feel to it.

1.If you want to make money, I hear sales is a good field to get into. Artists of any kind must not have the ultimate goal of making money.

2.The artistic medium of video games has greater potential than any other available today. A game can obviously mimic any of these media: Movies, extended mini-series, novels, graphic art such as painting, musical pieces, training simulations, educational shows, certain social interactions. The video game "dominates" these other media, it can do anything they can do and more.

3.Games should never be designed with the object of wasting time or trapping the user into spending more time or money. (See point 1.) Cliffhangers suck. Fantasy should enrich life, not substitute for it.

4.Spaceships should not travel at the "speed of drama". Are you afraid that reality is so boring that making your art too realistic will ruin it? Defining things in your narratives makes it more difficult to keep consistency, but the richness this adds is worth it. The TV series Lost sucks.

5.Of course I'm being very simple-minded with these rules: artists need to make money and I don't think they need to starve for their art, a video game shouldn't try to copy a different medium's idioms, games should be somewhat "hooky" to entice users, and drama is so important that realism is usually stretched a little in even the most realistic art. But why isn't anyone even trying? Games are assumed to be only timewasters and ways for game companies to profit off of stupid users. They should be more.