Raph Koster introduces himself as an alien from another planet: a game designer. He's the author of The Theory of Fun. He starts by introducing structure in music and art with some cool audience participation.

There are different dimensions to fun:

• Hard
• Easy
• Visceral
• Social

Hard fun is about solving problems. The problems tend to be mathematical Therefore the grammar of hard games ignore presentation.

He applies the theory of fun to Amazon.com and concludes that it's not structural, not fun. There a lots of sequential steps and don't provide any "fun" or feedback.

The magic ingredients:

• Territory - where you meet the challenge matters. Doing the same thing in different places should make a difference. The topology should affect the available options.
• Preparation - when you buy it, what pages and interactions you saw should matter. Everything you did before should matter. The last interaction you had should matter (think of your opponents last move). Never start an interaction with no context. You must prepare for the encounter.
• How - should involve skill on your part (eBay is more fun than Amazon). The core verb should be repeatable. Tricks and tips should matter. You should feel like you're growing more competent. Competition is an important element and people need to be able to see it.
• What - buying different things should feel different. The same verb must be applicable in many different challenges. A given verb is a hammer. You should present them with lots of nails.
• With - what tools you use should matter. Give users an array of tools to solve the problem. The system should give the user different rewards of different feedback based on how the challenge was met.
• For? - you need feedback. A game that has only one possible outcome is boring. A lot of services drive you toward a given outcome and then they stop. Usually, the best feedback is a grater challenge. Sometimes it's a pleasant surprise. Either way it has to be highly visible.
• Few? - don't always get what you want. Low risk activity for high reward is bad for fun. You need to drive users to challenges at the edge of their ability.
• Phooey - fun comes from learning. Failure is important to learning. Making a wrong choice has to be a set back. Fun won't exist if there's no consequence.

All of this can be quantified. You can give people ratings for how hard it is to accomplish a particular task. What is the challenge rating for buying a particular book at Amazon?