I was playing on one of the original Atari console systems yesterday, as part of a retro TV show, and it made me think about how computer games have changed over the years. Until you actually go and try out the video games of ten years ago you don't appreciate how much everything's advanced. Although, it must be said, it hasn't necessarily all changed for the better. There's a distinct lack of originality in the slew of new games, packaged in attractive glossy boxes on the store shelves.

One subject I do want to talk about is how much more interactive recent games have become. Through new hardware and software techniques the number of 'things' you can do is astonishing. This brings up the topic of this column; as game players we are the virtual puppeteers, bringing characters on screen to life with our own actions. Players do tend to be fairly distinctive too, sometimes you can tell who you're playing just by their style and technique. At the same time there's still so much untapped potential here for games. Imagine being able to control everything about your character on screen - hand and finger movements, arm waving, facial expressions, etc. Imagine hand-to-hand combat where you actually control each arm and leg yourself to make your own moves, or a lightsabre battle where you really control all of the swings and parrys.

This concept reminds me of the scenes in Being John Malkovich, where the puppeteer gives life to a marionette through a complex system of controls. These skills take a huge amount of manipulative skill and dexterity to master, but translated to a computer the possibilities are endless. Your online Avatar could have real expression and emotion because you would have the power to give them it. While bandwidth for this would be tricky, we already have the raw processing power to do this kind of complete control over a computer character.

One reason why we only see limited control, is that software houses don't want to make their game too complicated, otherwise they risk scaring away newer players. I remember John Carmack speaking on the subject of more control, and he drew a comparison with flight simulators that need you to read a long manual before even starting the game. He commented that with Quake 3 they wanted to make a game that was easier to jump in and play. I'm sure there's a happy compromise where new players can begin with the simplest controls to play a game, but then learn more advanced techniques as they progress. The learning curve is evident in all games already - would there be any harm in extending the top of that curve by allowing advanced control over your character? One consequence would be that experienced players would be a lot better at the game than newer ones and you'd have to play someone of similar skill for a good contest. But isn't that the goal of any game? To make the player want to improve.

The other big limiting factor is the hardware interface you use to control your character. How do you faithfully represent a complex system of controls? The real puppeteet uses a system of strings to control their marionette, but what do we use to control our computer generated avatars? While the keyboard and mouse approach works well for today's first person shooters, how do we control subtle facial expressions and a hundred other movements?

One line of thought I came up with stemmed from looking at the optical mouse technology that's becoming popular. They use a tiny camera to 'look' at the mousepad thousands of times a second. By comparing the separate pictures, the mouse can tell how far you've moved it between snapshots. It wouldn't take a huge jump in technology to make these sensors work in three dimensions. Instead of just looking at motion on the surface of the mousemat, theoretically you should be able to work out if the mouse is moving up and down too - by comparing how far away the mousemat surface is. This will most-likely need a special marked mouse surface, but there's no reason why it can't be done.

Now here's the good part. Imagine putting a sensor on each fingertip of a glove. All of a sudden you have five individual three dimensional points to control at the end of each finger. The number of control combinations from this kind of set-up is almost unlimited. As part of our lives, we generally have excellent co-ordination in our hands, so it wouldn't take long to get used to The Glove as a new controller. No doubt some big company will read this and steal the idea, but at least I said it here first!

Maybe one day the movie industry will consist of computer generated actors and us virtual puppeteers will be controlling them for the silver screen. I bet you never imagined playing Quake would lead to a Hollywood career. We can but dream.