Minmaxing Life

For those who have not heard of him, Randy Pausch was a professor of computer science at Carnegie Mellon University working on subjects closely related to video games. He was one of the creators of Alice, a software that tries to teach children programming while having fun. He was a brilliant speaker and you should not miss his last lecture if you haven’t seen it yet. He was a hell of a minmaxer with excellent tips on time management (his best video in my opinion). He was all that and much more until today when he finally lost to pancreatic cancer. He will be sorely missed.

Randy found out the hard way that it was later than he thought but he nevertheless kept having fun and enjoyed the game until the end. So - as one lucky enough to still have a ticking clock - the next time I hit a wall, I will knock it down in memory of Randy Pausch.

Anyone who’s been around me long enough has eventually had to suffer through one of my long winded tangent on the societal impact of technology and the probable advent of the Singularity. So, of course, Ray Kurzweil’s book’s “The Singularity Is Near” takes the place of the bible on my bed stand and it was with great anticipation I headed to South Hall yesterday morning to hear his keynote on “The next 20 years of gaming”.

However, it hasn’t been as exciting as I was hoping for since he basically gave the same speech I heard him give years ago at Siggraph. It’s a quick overview of his central idea - the law of accelerated returns - and its corollary that technological development follows an exponential growth curve. So the keynote went along these lines:

Exponential growth is very surprising and people usually don’t think that way: they make linear projections.

His interest in technology trends came from his desire of becoming an inventor. Timing is essential in building products so he makes mathematical models to project the evolution of trends.

It’s hard to predict the future on a specific project but much easier to do it on trends. He gives all his usual examples Arpanet, computer chess & Kasparov, etc… Then switches on to the democratizing effect of technology on tools of creativity, of production.

Technological trends are predictable, exponential in nature. An exponential trend is very powerful. There will be a billion fold increase in price/performance in the next 25 years. The effects go way beyond technology. It affects everything we care about: health and medicine (RNA interference allows to turn genes off), energy (solar energy efficiency is doubling every year).

He did throw into the mix a few interesting sound bites relevant to games:

- It’s unfortunate we use the name “game” because it makes it seem like it’s all pretend (oh, it’s just a game) when real things happen in game. Just like virtual reality or AI are unfortunate names.

- Play is the principal way in which we learn. We can learn real skills in games.

- Eventually “virtual reality” will be fully competitive with “real reality”.

Chris Hecker is one of those guys really worth following. He has been very active in the game community for years where he’s notoriously known as a super brilliant guy, that has never released a game. But that’s all going to change very soon as he’s gone directly from that to helping Will Wright on Spore…

He gave a profound speech today that is obviously directly influenced by his work on Spore and his life, trying to cross the bridge between art and science. I need more time to digest it fully. I’m left with the vague impression he’s wrong on his main hypothesis but I can’t exactly pin down why yet. Anyway there’s more than enough insight and deep thinking in there to make it worth anyone’s time, so here are my raw notes on the speech (italics are mine):

Chris Hecker - Structure vs Style.

Usually he tries for a speech that is

Concrete - specific - fancy demos - big name game - provided solution

Not this time.

Question : What technology has had the most profound impact on games?

He claims it’s the texture map triangle.

Because :

Powerful structure. The computer can reason about the triangle at a deep level… (morphology - topology)

AND

Expressive Style: the artist can represent subtle but rich designs.

Other examples: skinned mesh & bones, wavetable synthesis, mocap processing, html+css, typefaces

This decomposition is everywhere hard interactive problems are being solved.

What is a hard interactive problem ?

Not “wicked problems“.
Not “How to make this game fun” problems.
Not “easy” problems (nothing with a quantitative success metric).
So we’re not talking about about CPU, RAM even P or NP…

it IS problems at the intersection of technology, aesthetics, interactivity - the last one of which differentiates us from other media.

His theory: “I think solutions to hard interactive problems will aways have a deep Style vs Structure decomposition.”

Interactivity demands that the computer be in the loop but Emotion and Aesthetics demand that the human be in the loop.

Humans can create or execute algorithms
Humans can generate or illustrate or judge data
We often thinks of Humans vs computers when it should be Humans & computers.
We need to be finding out what computers are good at and what humans are good at and sort accordingly.
Because it’s not possible to bridge the gap yet. (he’s basically saying no one proposed a valid and complete top down model for AI. Although that’s true at this time, I strongly believe Jeff Hawkins is seriously closing in with his memory-prediction framework).

What technology SHOULD have hade the deepest impact?
Artificial Intelligence.
But it has not. We have not yet found a structure vs style decomposition for AI.

So what is Structure and what is Style?

Structure is the static specification of the Degrees Of Freedoms (DOFs) describing the object.
Style is the Values for the DOFs describing the object.
Static structure (This reminds me strongly of invariant representation in the brain): you can reason about it, You can write code that reasons about it.
Malleable Style : you can write code that changes it. IT’S DATA.

Choosing Style DOFs is an Art. It needs to be intuitive, expressive, frugal, blendable, efficient = it took a long time for the triangle to “win”. In physics the rigid body has not yet won.

Structure vs Style for AI ?
There will be structure vs Style decomposition for AI. Its style will NOT be based on code. There will be a “Photoshop of AI”. (That’s mostly where I have the feeling he’s wrong)

Does behavior require code? He used to think yes, but he’s not so sure anymore.

Who’s never heard of Sid Meier? Having designed Civilization is enough to make him a legend in game circles.
But that’s far, far from all he did. Here are my slightly reworked notes on his interview at GDC today (Italics are my added thoughts).

Standing the test of time: a Q&A with Sid Meier, questions by Noah Falstein.

On the development and design process:
Sid, born before video games, played board games as a kid (Risk and Monopoly - blech!). He feels game design has been more about evolutions than revolutions. So we should be cooperative on ideas and not be too protective. “Games are a series of interesting decisions” as he famously said. The development philosophy he follow is to make the historical research after the game is done. Fun comes first and build the game around that and only when that’s done, research the theme and make it fit the gameplay. Keep the cycle of play/improve during all development. It’s all about iteration and collaboration. “The more people you can have playing and giving you feedback, the better. I’m just the gatekeeper.” The audience today is much broader, not just hardcore gamers, so it’s important to also widen the scope of your design accordingly. The word “casual” is a tough one to pin down : it can be about either the cost or the complexity of the game. He’s personally not interested in “simple” games he’s interested in “complex” games (duh! He made Civ!), but he’s attracted to working with smaller teams. “Bring the player’s imagination in the game” : working with so much limitations in older games forced them to do that. And that imagination is still around but not being tapped as much now that we have fancy graphics. Extra stuff he said when replying to audience questions at the end:
Reward system : “people need some sort of validation at the end of the game”.
The prototyping process is not pretty but it’s important to try out as many ideas as possible in order to find what works. Having a working prototype is essential.
He draws a solid wall behind the “game logic” and the “presentation”.

On creating Civilization:
Like in Will Wright’s Sim City, the idea of creating and building was one of the driver behind Civ, as opposed to just blowing stuff up as was the norm in strategy games. Adding different systems, each clearly understandable on its own, increased complexity in a way that made the sum greater than the parts.

On the subject of narration: how do you feel about story ?
One of our key role as a game designers is to pay attention to “who’s having the fun”. The player should be having the fun and the game designer not so much. He feels the player is more interested in his own stories than the ones the game designer can come up with so he designs in this way (sooooo true).

On balance:
Balancing has a lot to do with iteration. It’s about allowing rich choices but keep the complexity at a level that’s manageable to the player. The computer can manage a very high level of complexity but the player can’t necessarily follow that. Later on, to an audience question: “Our approach is really to play the game and keep playing. And if something is imbalanced it will show up.” But balance is not the end all: a positive experience is more important overall.

On addiction: did you expect it would be that addictive and how do you feel about it?
He did not expect it and was scared about it at first when he heard stories of severe addiction. It was a revelation and at the time they felt that this gave a glimpse of the future and how people would be willing to spend a lot a time in games. As this was pre-internet, people came to convention with the need to speak about their “Civ experience”: they did not have any message board were they could express how they felt. “It’s always scary the effect your game has on people”. To the audience, later on: “I’m more confused by why people are not addicted to games! (…) What do we do with this industry/art form? We are already influencing people but do we take responsibility for that?”

On Dinosaurs: (a game project that was canceled one year in development because he couldn’t find the fun)
An idea that he has not given up on. “We all have failures and it’s important to recognize them early and let go as soon as possible.”

On Civilization Revolution: (the console port of Civ he is currently working on)
The approach with Civ revolution is to put you back in that “king seat”. Giving you the feeling that you are making important, high level decisions. Finding what is “fun managment” and was is “micro managment” and get rid of the latter so that “every decision is interesting”.

More opinions on games: (mostly extracted from audience questions)
- It’s kind of a golden age for gaming. Games he liked recently : Grand Tourismo, Mercenaries, Halo “of course” (Unbelievable! I would never have thought Sid Meier enjoyed twitch games…), “when you do something all day, you want something else” which is why he’s not playing so many strategy games.
- What motivates you to make games: “The fundamental desire to play a game that hasn’t been done before.”
- Would he collaborate with American McGee on “Trains in Hell”?: No, each designer has his own approach so two designer working together you’d probably get the worst of both as opposed to the best of both.

Last question:
Q: With Spore, Will Wright is attempting a combination of everything he’s done before. Do you have any ambition of making some kind of monumental game that would sum up all the mechanics you’ve build along the time.
A: “NO!”.

Apparently, the good Dr Kawashima - who inspired the Brain Age games that sold over 17 million copies worldwide - does not want a single yen from the 22 million dollars in royalties he is due:

“Everyone in my family is mad at me but I tell them that if they want money, go out and earn it.”

he says, because

“to hear this may put you off, but my hobby is work.”

Score one for passion, zero for money!

Also interesting, his comment on work versus play:

“Having fun is not studying. Making them study is not to entertain children but to pressure them to make efforts. People fall to lower and lower places unless they are driven to go higher.”

This correlates with my own observation that strictly favoring fun over anything else prevents one from learning those skills that demand hard dedication and self-discipline. Unfortunately, there are some things in life we don’t (yet) know how to teach in a fun way but that still need to be learned…

Garrett Lisi simply rocks. He might very well turn out to be the guy who cracked the Theory of Everything with his - tongue in cheek - Exceptionally Simple Theory. But whether he’s right or not, the simple fact that he managed to have a shot at this after having spent the last 10 years surfing around is nothing short of amazing. Not everyone has been zombified by teevee after all!

As if that wasn’t cool enough, the slashdot reading, self-declared “contemplative hedonist”, says stuff like:

“I want intense pleasures from life. This means not spending too much time in a lab but finding balance between thinking and having fun”.

(loosely translated back to english from an interview in Le Monde)

and

“The best thing there is to do on this planet is surf. There’s just nothing else that even comes close to generating the feeling you experience as you paddle like crazy to catch a wave, drop down the face, crank a hard bottom turn in the pit, and pull into the barrel as the lip throws out over your head. The desire for those perfect waves wreaks holy hell with the rest of your life. It makes you spend all your dough on plane tickets to crazy places just so you can risk your butt paddling over jagged reef and camping out on lonely beaches.”

All that and he doesn’t believe in string theory. And he hangs out at Burning Man. And he still finds time to teach. So yeah, I’m a total fanboi but seriously: how much cooler can you be and what more do you need to revive your faith in humanity? Now if only kids would pick him up as a role model instead of the usual cretinous MTV celebs…

Live long, Garrett Lisi and keep catching those waves!

EDIT: another nice interview he gave, complete with a Douglas Adams reference and a picture of E8.