Designer's Notes

by Greg Costikyan

Elsewhere on this site, you'll be exposed to all the hype you could possibly want for Evolution. That's what sites like this are for, and I have no objection. But I wanted to talk a little bit about the design of the game -- our basic approach, the key decisions made, why we're doing what we're doing. It's a subject that, as a game designer, obviously interests me; and although I have scant evidence to support the belief, I suspect it's a subject that is likely to interest a good many gamers.

I apologize for the lack of graphics or other media on this page; but I am, of course, creating it in the medium with which I am most comfortable -- the written word -- for which there is no adequate replacement if your intention is to impart ideas.

Is There a Game in Evolution?

In 1996, Al Roireau of Discovery Multimedia approached Crossover about doing a game on evolution. Cool!, we thought. Lots of little animated dinosaurs and mammoths and things wandering about. Good idea.

Except that it isn't immediately obvious how you go about making a game out of evolution. Evolution is a process that occurs without any conscious guidance, and without any predetermined goal. People have the notion that evolution has somehow been "striving" for intelligence, that we are the "crown of creation," but this is nonsense; being smarter isn't always a pro-survival trait, and evolution never "looks ahead." It's purely a process of winnowing; what survives, survives, and breeds another generation.

But the fundament of every game is decision-making--presenting a player with a situation, and requiring him to decide what to do and how to solve the problems he faces as he struggles toward a win.

If there is no volition in evolution, if it is purely an automatic process, then any real simulation also has to be automatic--which rules out any role for a player. In other words, there ain't no game.

In fact, there have been attempts to do this before; Will Wright's Sim Earth is an attempt to show the broad-stroke history of life on the planet, while his Sim Life is an attempt to show the process of evolution on a much smaller scale. And they are admirable products, and interesting in a way -- but games they are not. All they let you do is fiddle with a few parameters, then sit back and watch things work out. This is not a lot of fun.

And not what I wanted to do, either. I like games. I liked the idea of a game with Tyrannosaurs and sabertooth tigers, but I wanted it to be a game, not a computer model.

What Decisions Do You Make?

How to make it a game?

A better way to phrase the question is: What decisions do the players make?

If it's a game about evolution, obviously they should be deciding what to evolve next. And they should be working to get to the point where they can evolve something -- and that means making their creatures successful in competition with the creatures of others. "Success" means reproducing and establishing large, stable populations; so the feeding system is obviously key.

Ergo, we need a game in which evolution is planned by the players, and in which they can determine when and how their creatures feed.

That's tricky: We wanted Evolution to be a realtime, continuous movement game, like Command & Conquer, as opposed to a more static, turn-based game like Civilization. Creatures need to be able to feed automatically, at least most of the time, without dying. But we don't want them to be too smart about it, so the player has an incentive to keep monitoring the game state, and guiding his creatures to places where they can thrive.

We solved that problem in a straightforward manner: each creature in the game has a "feeding center." It feeds from all squares within two of that center. Each feeding update, it checks to see whether it would thrive better if the feeding center moved to an adjacent square. So after time, creatures will spread out to exploit available resources. But they won't always notice if there's another region, half the continent away, where they might do well. So the player still has an incentive to guide his creatures to good locations.

The position of the feeding center changes only slowly -- but the creature wanders semi-randomly about, within its feeding radius. So we have the continuous movement we wanted.

Of course, different creatures do well at different latitude zones, and in different types of terrain, so even if one species dominates a particular region, others can still survive in places that are more hospitable to them.

162 Living, Squonking Creatures

It would be really cool if we could somehow start with the earliest land animal and allow it to evolve in any direction--somehow to smoothly morph it as it evolves, allowing all possible morphological variations given its starting position. That is, more or less, kind of how evolution works.

But it's not possible given the current state of animation and modelling technology. We're stuck with "instantial" creatures, not "algorithmic" ones--that is, animal models that are created and animated and rendered and then plopped on the CD-ROM, as opposed to ones that are somehow drawn and modelled and animated and rendered on the fly.

And if we're pre-rendering creatures, we obviously want to hit the high points: the dinosaurs everyone knows, the creatures that are key to the evolution of life on land, creatures that people don't know very well (like Indricotherium) that just look really cool. And there's a limit to how many we can supply; in Evolution, there are over 100 frames of animation for each species in the game. Keeping all the animations for one species in memory takes up close to a meg. It doesn't take a genius to realize that we can't have too many.

Not to mention the fact, of course, that 3D modelling and animation is expensive. And while models for Tyrannosaurus and Smilodon are commercially available, we have to make our own for Acanthostega or Pareaisaurus or Megazostrodon. In fact, fewer than ten creatures are animated from unmodified commercial models, although some creatures are similar enough that modifying an existing model is not a great deal of work (e.g., Kentrosaurus is not that different from Stegosaurus).

So we have 162 species in the game. We'd love to have more, but the amount of work involved in creating this many is simply staggering. But it's worth it: you'll see creatures moving that you've never seen before--walking, resting, fighting, dying.

The Tree of Life

Evolution has a "Tree of Life," which you can think of as something like the technology development diagram in Civilization. Each species on the Tree can evolve into one or more descendant species. Certain species are key not just because they can feed well in a particular environment, or are fierce predators with which you can attack your enemies, but also because they lead to other cool creatures.

But Evolution's Tree of Life, unlike the real world's, is polyphyletic.

Say what?

Here's the thing. You and I and every human being on the planet descend from Australopithecus. That means we're monophyletic: our phylum (the species Homo sapiens) derives from a single (mono) other phylum.

If you descended from Australopithecus, but I descended from the chimps, and she descended from giant lemurs from Madagascar, we'd be polyphyletic. Of course, things don't work that way.

But in Evolution sometimes species Z can have several potential ancestors -- call them W, X and Y. In any particular game, only one of those potential ancestors will evolve into Z -- W this time around, maybe X next time we play -- but any of W, X and Y could wind up being Momma, depending on what the players do.

Why? Consider this. At the end of the Permian, 96% of all the species on the planet became extinct. At the end of the Cretaceous, 60% of them died.

If our "Tree of Life" followed that of the real world 100%, then all but one of the species available in the early part of the game would be dead ends, incapable of leading to anything in the future, because they died in the real world, and we never knew what kinds of descendants they might have led to.

Would you, as a gamer, bother to develop a species that leads no where?

So instead, everything can lead to something, until you get to the modern day. That's not implausible, by the way; evolution tends to recreate the same basic forms over and over. If the diapsids hadn't led to the dinosaurs, quite possibly an anapsid or synapsid would have evolved into something remarkably dinosaurian.

But the polyphyletic Tree of Life solves another problem as well. In Evolution, each species is controlled by a single player. If I evolve Tyrannosaurus first, only I can breed 'em. If you want a fearsome dinosaurian predator, you'll have to go for something else.

In the real world, of course, all of the Mammalia are monophyletic; all the mammals in the world descend, ultimately, from a single species of early Mesozoic therapsid.

If Evolution worked the same way, then one player would develop the earliest mammal first -- and control every mammal in the game thereafter. This would, to put it mildly, make the end game rather boring.

So in Evolution, there are four "early mammals", each of which can derive from different ancestors, and each of which can lead to later mammals. So up to four players can wind up with mammals, preserving competition into the Cenozoic.

The World, She Is A Changin'

One of the things that irked me about Sim Earth, which I admire in many other ways, is the fact that the continents look the way they do in the modern day, and never change.

I understand why Maxis did it that way, of course: they had their hands full simulating the biosphere, and didn't want the headaches of trying to simulate continental drift, too.

But of course we know that the world today looks drastically different from the way it looked in the Paleozoic. And Evolution begins with the Carboniferous -- 360 million years ago. Can you say "Gondwanaland?"

So I knew we needed continents to move about, mountains to rise and fall, terrain to change, glaciers to advance and retreat...

It was a lot to bite off, more so than I had anticipated. But the upshot is cool; not only will there be a "Real Earth" scenario showing how our planet has changed, but on randomly-generated worlds, continental drift will produce plausible results, too.

And, of course, there are the disasters. The asteroid-impact hypothesis may still be controversial, but the fossil record clearly indicates that enormous catastrophes have reshaped the biosphere pretty regularly. If Evolution was plausibly to claim to represent the planet's history, we had to have disasters: asteroid and cometary impacts, enormous volcanic events like the Deccan and Siberian traps, explosions of nearby supernovae...

They have a major impact on the game, of course, but in addition, we're doing 3D animated "cut scenes" for each. Work on them has only just begun, but what I've seen looks spectacular. There's nothing quite like seeing a comet wipe out most of North America...

Research

A game like Evolution is a constant balancing act: of course, we want to represent the creatures in the game and the nature of evolution as accurately as possible. Contrariwise, we want a compelling, immersive game. And sometimes -- certainly not always -- those goals are in conflict. I've talked a little bit about the compromises we've had to make -- the fact that players drive evolution, while real evolution is an automatic process; that our "Tree of Life" is polyphyletic. Even the game's ultimate goal -- to be the first to develop an intelligent species -- is, in a sense, defying reality, because evolution has no goal other than sheer survival.

But I think it important to emphasize that we aren't simply taking funny looking beasts from the past and putting them in a game willy nilly. This isn't Monopoly with scales and fangs.

All of the creatures in the game -- except, of course, the non-human intelligent species and their immediate precursors -- are real creatures, who actually existed in the past. Each is modelled -- both physically and in terms of behavior -- on what we know about those beings. The history of the planet, in terms of plate tectonics and geological change, is modelled on that of the real world.

No one at Crossover is an expert in paleontology. But I, at least, have a degree in geology, and we've spent a lot of time on research.

And, along with Discovery, we take the game's educational goals seriously: we hope players will come away with a better understanding of our past and the planet's evolution and the nature of paleontological descent. And we hope, of course, that they'll have a hell of a good time in the process.

Alpha

One of the frustrating elements of Evolution's development, for me, has been the sheer difficulty of working with software. Only now (4/15/97), more than six months since the project's inception, do we have an "alpha." That means: software that contains all the elements of the game in at least usable form, stable enough that you can actually play.

Getting to this state has meant more than a man-year of programming time, and at least as much work by the artists. By contrast, when I design a boardgame, I have a functional prototype within a few weeks of the time I begin.

So we now have a program that works. The next step is to make the game itself work. Inevitably, what you envision at the start turns out to have problems in practice. At the moment, for instance, the little creatures are feeding and breeding at far too high a rate; they're crawling around like bugs, and evolving into mammals by the Late Permian. This is solvable, of course; a matter, mainly, of fiddling with parameters.

We've got our work cut out for us, making Evolution sing -- but everyone on the project believes we've got a possible winner here. We hope you'll agree when you see the final game.

A Collaborative Form

Another contrast with paper gaming is the sheer number of people involved. A boardgame, typically, has a designer, a developer, a graphic artist -- maybe a few others, like playtesters and illustrators.

The programming for Evolution alone has involved four people. We have two full-time animators doing the creatures; more doing the cut scenes. A composer for the music, a sound designer for the sound effects, the art director, storyboard artists... and, inevitably, producers to manage all these disparate talents.

Computer gaming isn't at the level of the movies, where credits scroll on forever; but they are true collaborations, involving a staggering number of people. You begin to understand why many computer games have budgets in the millions, these days.

I'm putting my name on this document -- and I'll be listed in the credits as its designer -- but the designer is no more the auteur of the whole than is the director of a movie. The programmers and artists and producers play at least as important a role. I hope when you play the game, you'll check out its credits, and appreciate the contributions of so many folks.

-- Greg Costikyan
New York
April 15th, 1997

Return to Games page.