Etech 07: Raph Koster on magic

Raph is high information in short time bursts, so I've taken staccato notes on this one. He usually puts his slides online after a talk, so check over at his site for such nuggets of gold.

[UPDATE: He's posted up his full slides here.]

Things that work have underlying structures.

Art, social media, physics.

Sometimes they get crazy elaborate: Solar by Miles Davis. But it follows a blues grammar.

We can identify this by the ear.

The best computer programs doing harmonic analysis have trouble, but humans don't.

This structure is grammar.

Grammar is underlying all ways in which we communicate.

In grammars the logic is fractal, the structure is fractal.

Blues structure has recursive quality to it. All of music is about the leading tone.

This is deep structure.

What it means is that songs... I’ll get off music soon... are made of smaller songs.

Even complicated stuff like a fugue by Bach is not only made up of smaller songs, but overlapping songs too.

Harmony is overlapping voices.

This grammar carries through to absolutely everything.

See van Gogh: the golden section: fractal math!

Games are made out of games.

Look at the latest big cool game. It’s a giant game. With small games you play within it.

Designing web 2.0 techs, the same principle applies.

Site, page, box, button.

Frogger: game of getting a hi score: game of getting frog to five spots: one frog to one spot: game of crossing the road: etc…until you reach the interface button. Hopefully you can't lose at that.

The atomic grammar reaches its bottom level.

Games, and each individual game, has to be FUN.

We game designers obsess about making all of those micro bits entertaining.

Clearly designers of typical application productivity software don't do this.

So how do we do that?

Fun is something moderately well understood.

It's chocolate and orgasms: a drip of morphines down your brain stem.

We are in a sense drug pushers.. but fun arises because of characteristics:

• Understand a pattern,
• Poke it,
• Learn how it works,
• Master it.

That’s how games work.

Then when you master it, that moment... WOO HOO. Then you get your little drip of drugs.

There is interesting empirical research on what kinds of fun are.

This is from Nicole Lazzaro:

• Hard fun: tough problems for you to solve.
• Easy fun: aesthetic delight. "that's really pretty".
• Visceral fun: rollercoasters, where your stomach falls out.
• Social fun: games are always played in the company of other people. Schadenfreude. GLOATING! It’s fun.

Games mostly focus on hard fun.

You can find lots of hard fun games lacking in the other bits, e.g. Go against a computer. Fun, compelling, but not social, etc.

So there's a grammar to how these things work.

You can decompose a move in Go.

You can make this action more or less fun.

We're talking about users interacting with a system and getting a feedback response.

We’re talking interaction design!

Front page of Amazon: like a game. Nested objectives, and parallel objectives.

Think of moment to moment fun.

This is a really daunting list!

It’s like an endless chain of logs and turtles without much feedback.

In a game we would try to find a way to do that.

Buy a book: that's your goal. You have to pursue all these challenges in parallel...

You might buy multiple books; you meet the same challenges multiple times.

Is it as fun buying the fifth book as your first book?

Here are the magic ingredients:

• Where should matter.
• When should matter. What page you came from previously. What your past      interactions are.
• What you are buying should feel different.
• What you do it with should matter... what tools you bring to the problem. It      matters whether you’re coming to this with a sword, or a bow... users should feel the difference.
• What you're buying it FOR should matter. You need variable feedback. Did you do it well or poorly?
• In games it matters sometimes that you don't get what you want. That’s what drives return visits...
• Finally, if you fail, the game needs to smack you on the head and say, you idiot. Fun comes from LEARNING. Failure is very important.

A verb is an objective. The goal.

In a game it might be destroy, or capture.

Commerce, it might be buy.

Social, it might be connect.

But this has to be a repeatable activity.

If people don't care to come to it over and over, then it will fail.

It has to involve skill. You need to be able to do it better or worse. Purchasing on eBay is compelling - you figure out tricks! Sniping. Evaluation. In order to learn, you have to feel like you're growing more competent.

Fun comes from a growth in competence.

As you come to accomplish it, there need to be variant challenges. Connecting to a CEO on LinkedIn vs. connecting to the pr dude = different.

What you want is for the game to acknowledge the fact that it's tougher to get on Reed Hoffman’s linkedin rather than someone who sells ads.

Social media is about cooperation, but the core of games is competitive. As soon as you give people a ladder to climb, they'll climb it.

Ratings. Metrics of contribution. Other people need to see it to measure against it.

Everything you have done must matter.

The system needs to remember that.

Amazon needs to know whether you came from search or direct link. They do know, but are they telling you?

The last interaction you had should matter.

When interacting with tech, it should be like chess. Whatever the feedback was from the device should change the situation. Can you tell what you did, did anything? The feedback?

In terms of interaction design, you should never start a given encounter without there being context. What has the user already done?

The user should be able to prepare for the encounter.

They need to be able to come to it to prepare for ways of taking on the challenge. Tooling up before battle, if you like.

When you come to a social media site, it should matter... arriving with different personas, profile info, and seeing how the system responds.

Doing the same thing in different places should make a difference. Users are seeking novelty and new scenarios. In a fighting game, you're doing a flying kick. It matters where you start from. Similar situations should arrive in interaction design. Buying a book off the search page, the topic page, the author page. It should matter.

The topology should affect the outcome. The things that surround the task should affect how you do the task.

You should be able to do this task on many different challenges. Buying a book, making a friend, finding a partner?

A given verb that you supply, like buy, is a hammer. You need to give users lots of nails. If they like buying, they might apply it to all kinds of different things. Buying membership. Buying a one day experience, a friend, an event.

There are lots of kinds of hammers, too... users should come to any form of interaction design, and choose from an array of tools with which to solve a problem. Connecting to someone on a social networking site: lots of ways. Intro emails, referrals, forums, groups, a multiplicity of ways of reaching that goal.

The system should give different feedback based on how you did it.

"You have accomplished this. You have bought the book".

How different it would be if Amazon said, "good job finding this book from the search results!"

The path by which users come should be acknowledged.

The interesting key point for you guys is that all of this can be quantified. You can measure how hard it is for users to come in via a particular path.

You can do this in game design, you can have stat measures on how hard something is to achieve. A challenge rating for your interface. We do this: we graph as you move through game levels. We track the difficulty rating as it goes to see how many people will make it to the end.

A game that only has one outcome is boring.

An interesting thing about a lot of services is that they drive you towards succeeding, you succeed, and then it's the end.

Games have learned that this is boring.

When you buy something it should tell you alternative endings. Greater challenges. Secret discounts!

Sometimes you should get a pleasant surprise.

Games do this really well, and social media sites do this really badly. Get a free book on your 10,000th book ... and let everyone else see it!

Bottomfeeding is bad for fun. Think link farming on orkut.

Low risk activity for high reward is bad for fun.

When we build services, you really want users riding right at the edge of what they can actually accomplish.

Flickr photo challenges, for example. Or buying the wrong shit on eBay.

It’s important for fun for you to realize you did the right or wrong thing.

The absence of any of these features makes something less fun.

Comments

[Schmitt]

If Amazon told me I did a Good Job finding my book, I would kick it in the nuts.

[sporb]

Schmitt: your comment tells us more about you that it does about ideas on interaction design at Amazon...

-s

[cazart]

I'm with Schmitt. That's patronizing.

[Alice]

I think Raph's point was, if Amazon rewarded you for achieving something, how different would the experience be... (answer: different!)

You're taking the wording a little too literally, I think.

[Vincent]

I have no clue how you manage to transcribe stuff this dense, but perhaps that's down to my density. Good stuff.

It struck me right at the start that Raph's point about the nature of games - see a pattern, poke it, learn how it works - is exactly what I'd been hoping someone would say about stories at GDC, SXSW, etc.

Stories work the same way, except you have to watch someone else do the poking (that'd be the protagonist in the story and, yes, in certain kinds of stories the poking would be literal as well as figurative).

Surely that should mean there's a better natural fit between games and stories than people assume, beyond the old linear vs. emergent debate?

The pattern in a story is the theme. A story's theme might be 'crime does not pay'. Now stick that theme into a game. Suddenly, the story isn't about plot points a, b and c. Instead, it's a pattern the player has to uncover.

Give the player a Grand Theft Auto-esque sandpit. Give them an inciting action - get $1million to some gangsters before sunset or you're dead. How the player goes about achieving that is up to them, but they are constrained by the story, which stipulates that 'crime does not pay'.

Every time they attempt to get the money by criminal action, sooner or later, things go wrong. Only when they find a legal option do things go right. For a cinematic analogy, see 'Run, Lola, Run'.

It's still Skinner's pigeon figuring out that pecking the lever is a behaviour pattern that will cause food to drop down that chute, only it's metaphorical, rather than mechanical. Story is just a metaphor and a metaphor is just a way of expressing a pattern. The power of story isn't about saying 'crime doesn't pay', it's about illustrating what that means.

A game is potentially far more effective at illustrating this meaning than a book or a film, because those media force you to identify with the choices someone else makes in order to learn the pattern. In a game, you get to make the choices for yourself.

Sorry, rant over. Maybe next year someone will talk on this at GDC and then I won't feel compelled to write these off-topic comments.

[Prokofy Neva]

I can't believe Raph has fallen for all this Chomsky stuff. Yeesh.