Friday, October 28, 2005

Article: "How to prototype a game in under 7 days" on Gamasutra

Here's a great little article on prototyping. It is rare that you see such a dedicated prototyping effort described in such detail. At times I feel like those who write about game design are anthropologists wandering deep in the Amazon looking for signs of some highly elusive tribe. Most activities are done in secret, under the cover of darkness. and when you actually get a chance to a design ritual in action, you should whip out your notebook and frantically take as many notes as possible. :-)

One of the suggestions that came up was to make a 'toy' first and then add goals like scoring. This concept of toys has come up repeatedly. Chris mentioned it in his book and numerous other authors have attempted to make the distinction between toys and games. Each is describing a different slice of the same basic puzzle.

A toy has been described as a simple fun activity that 'feels good'. It also has been described more formally as an environment that allows players to create their own goals. Both of these can be categorized as low level core game mechanics that do not have higher level meta mechanics.

A toy is really a simple risk / reward system. Do A and reward B occurs. The risk is often minimal and can be described in terms of opportunity cost or irrevocable decisions. Rewards do not need to be numbers increasing or stars. A little animation, a new graphic or a sound effect are all simple rewards. The article describes this as 'juiciness', which is a delightful term that speaks to the inherent reward in many tactile activities.

We often forget that mechanics like score are meta game mechanics layered ontop of simple tactile mechanics. The coin collecting in Mario is a layer on top of the simple activity of jumping. The score in asteroids is a layer on top of the highly rewarding act of shooting asteroids and watching them explode.

In games like that Sims that are described as toys, there is a simple risk / reward to watching your Sim perform an order. There almost always is an existing core game mechanic. Where many people get confused the nature of such a game is that they expect a plethora of higher level goals to direct their actions. They look for things like a score or a mission or other familiar mechanics that have been used in their favorite games.

None of these elements are needed really. All they do is explicitly reward and enhance a particular play strategy that emerges from the players interactions with the existing game mechanics. Here's a thought experiment. Imagine playing Space Invaders. A particular player enjoys sitting in the right hand side of the screen as a method of controlling the space around their character. He has a series of implicit goals about not moving too far out of his zone.

This player is creating his own rules, punishing himself when he breaks his rules and feeling satisfaction when he executes them well. This emergent goal creation is evident in almost every game that I can imagine. The mere fact that you have choice means that you must choose a play style and build up internal goals.

So the boundary between games and toys is remarkably fuzzy. All games have emergent goals. Most games that are described as 'toys' have risk / reward sequences at their core. At best, I see the term 'toy' as being useful to describe the general flavor of the gameplay to an existing experianced gamer. More often than not, it gets you very little practical information for further refining your design.

I lean towards a term like 'highly layered' when I judge a prototype. A title like Civilization is has a large number of concretely defined strategies that are explicitly rewarded and expanded upon with complex layers of meta mechanics. A new game prototype will have one or two of these risk / reward layers and the player will be forced to make strategic leaps without much guidance.

Measure the player strategies
Here is a useful method for adding additional layers to your game prototype that the article touches on very briefly. Start by building up a game is to create a core game mechanic with a focus on an enjoyable 'juicy' activity. Then watch what the player does. What activities are repeated? Which goals does the player set for themselves? Often this takes the designer's touch at the early stages since it can be difficult to get players to explain their momentary urges.

Once you've identified these primitive strategies, try to create risk / reward sequences around them. When you are popping bubbles, perhaps popping three in a row gives you points. When you are jumping on something and yelling 'Gotcha!', wouldn't it be nice if the thing you were jumping on broke and maybe gave you something for your efforts.

take care


  1. Garry's Mod 9 is going to be released in a few hours, and one of the major features of the release is LUA scripting. Some of the games beta testers have been making really show off the system's flexibility. You aren't limited to any of the normal FPS conventions at all.

    It seems to me that Garry's Mod with LUA is the ideal candidate for demonstrating prototyping. There's the existing Source/Half-Life 2 framework to build on, a very wide user base and it's all serverside so once they have the mod itself end users don't even have to download anything. What do you think? I know I'm going to be giving it a whirl. ;-)

  2. The reason that I kept playing Killer7 was to find out what was going on. I wanted to find more information about the world and characters, and I enjoyed piecing together the fragments and making sense of the story. Essentially, theorizing about what was happening and what it meant became part of the game to me.

    Of course, the problem is that the game has no way to interact dynamically, to reward me for correct inferences or punish me for not paying attention (Trace Memory actually has multiple choice quizes at the end of each chapter, but it doesn't work very well). It's a form of gameplay that can be enjoyed by the involved player, but is not easily communicated as part of the games appeal, as well as not tying in to the game's win conditions. It does however have metagame aspects, as I've seen quite a bit of analysis and speculation on message boards and whatnot.

  3. Great article Danc. I've been enjoying your site ever since stumbling onto your Revolution controller thoughts.

    I've found success with the same design process. You build a toy and play with it. You find some actions more fun than others. You create goals to encourage and reward those actions.

    My comments on the "How to prototype a game in under 7 days" article are over at my site. I elaborated on the lesson that constraints enhance creativity.