The man behind countless classics at Bullfrog and Lionhead Studios--including Dungeon Keeper, Populous, Syndicate and Fable 2--explained how Lionhead turned its love of dreaming and experimentation from a vulnerability to a great strength.

Black & White: a chimera born of mad science
Molyneux began by describing the development of Black & White as "utter chaos."

Many ideas from this chaos were turned into game features--"that incorporated such things as the creature--okay, that was a plus. But also..."

I had one idea actually at GDC, and this was back in 1998 or 9, I can't really remember. And I was talking to a journalist, and for some for some reason, maybe because I was in sunny California...

I said 'Hey, why don't we make the weather inside Black and White the same as the weather outside in your garden in the real world.' And so I went back to Lionhead, and went back to all the programmers and said look, I have this really good idea.

They went off, went to 10,000 weather sites, translated all the data, did all that stuff. While they were doing that, they weren't working on the game itself, the game mechanics weren't working.

It was just a dumb and stupid idea, and it really derailed the game for a little while. And that chaotic place, where this game Black & White was made, is where these experiments came from.

Lionhead formalise experimentation
The rampant experimentation had to be reigned in, but this was risky in itself:

Lots of people have cool ideas. It's not just designers. It's certainly not just me. And because we just focus more and more we're doing "the game," there's no forum for these people to experiment.

And if someone has a really bright idea, what do they do? Well a lot of times I'll tell you, they leave your company and they go somewhere else, because the idea is so smart.

The solution was to formalise dreams and experiments at Lionhead.

Turn the page for more. _PAGE_BREAK_
"Anyone can propose an experiment. And the way they do that, they get a sponsor from a senior member of the staff... these sponsors then sell the idea to this creator board."

Once an experiment has been approved, a small team of 1-5 people is formed to work on the idea for 1-12 weeks--most lasting 6. During this period they are scheduled like any other project, with check points and milestones to meet.

"If you can have a very small experimental team that costs a man-month to do an experiment, it's much more likely you can do more experiments and throw them away."

"When they come back to that board, they then ask some very simple questions, depending on what this experiment is.. We have to sit as a board and measure the success of that experiment."

Success is judged by what it adds to the game, how much it will cost, how often it will be seen in the game and how well it can be taught to the player. The ideas to survive will then make it into the game and the rest are dumped, with no great loss.

Fable 2, the fact that we did put those innovations in there... we didn't have to do that, but we only did that because we experimented in smaller test-beds first.

If me as a designer and us as a design team, if we didn't have these experimental groups here, then we just wouldn't have been able to experiment with what we did in Fable 2.

Tools of the trade
Molyneux also spoke on the technical side of Lionhead's experimentation process.

On the low end is a very simple prototyping engine which "enables even layman coders to play around with an experiment" but it can be difficult to get a feel for an idea with so little context.

"A lot of the times, we kind of give ourselves the excuses. Oh well, it's not fun because the graphics aren't right, or the sound isn't right--and we can use existing tech to get around that."

This is when Lionhead build on their previous work to see the idea properly.

"The other way is to use existing game engines--use Fable 1, use Black & White, really to use existing technology we've got to get to the core of the idea.

The grandest dream of them all
Molyneux went on to explain his current grand dream and prototyping hope for the future--Concrete. In essence, Concrete is video game design LEGO, where existing assets and elements can be pieced together to rapidly create complex prototypes.

The dream of Concrete... this industry talks so much about sharing code, sharing engines, sharing AI... but maybe there's some fundamental things we should do before we can do that. And so we created this framework called Concrete, and the idea with Concrete is if you create any asset, and by an asset I mean anything from a piece of graphic to an animation, if there is a framework in it that allows that asset to essentially talk to other objects, and be compatible with other objects, then this dream starts to become a reality. And this dream was, imagine if you had a game like Fable, and a game like Black & White, why shouldn't you be able to literally physically pick up a tree from Fable and put it in Black & White?

The big dream is that me as a designer would then be able to pick the landscape from one game, the trees from another game, the villages and their AI from another game, combine them all together and make a game in 48 hours. Obviously that is still an absolute dream, and nowhere near a reality...