Monday, May 02, 2005

Spore - a biological perspective

[Note: The following post is based on information obtained from one GameSpy article, and is intended to be a discussion of the biology relevant to Spore; this is not a review of the game.]

The Game

Back in March, Penny Arcade linked to a GameSpy article summarizing a talk by Will Wright on his newest game, Spore. The game is still in development, but it's based on an intriguing idea: instead of using premade artwork and animations to populate the game's visual realm (as in The Sims), Wright wants to have the software dynamically create all the animations, characters, and structures in the game, based on each player's input.

The premise behind Spore, as it has been described, is that each player starts out controlling a single-celled organism, probably a protist of some sort, that the player can then "evolve" into more complex organisms. The ability to "evolve" an organism occurs only after the player has helped their organism survive enough challenges (e.g., consuming enough food), and "evolution" occurs by the player choosing between specified morphological characteristics that can be added to each organism. After lots of "evolution," the player ends up with a fully 3-D, multicellular creature (an animal), which continues to "evolve" through the addition of even more structures. After the creatures have "evolved" to live on land, the creatures can "evolve" culturally, including building societies, which can "evolve" the ability to travel to and colonize new worlds, all the while interacting with organisms created by other players.

The article professes that the game editor is extremely flexible, and can build just about any organism:
"Sure enough, just about anything was possible with the editor: Wright demonstrated an upright dog whose front legs were twice as long as his back legs, a creature with an enormous floppy eggplant-shaped head that had no less than a dozen hungry beaks, a six-legged critter with two snapping heads that skittered along very fast, and finally a fully-functional Care Bear. (!)

"Regardless of what you could dream up, the game would find a way to make it work. Top-heavy characters would bobble along awkwardly, creatures with branching networks of a dozen legs would learn to walk, and animations for fighting and eating would be generated on the fly."
While the technology behind the game sounds fascinating, based on the review it appears that the game has clear biological limitations, and even a few inaccuracies. This post looks at the game from a biological perspective, and examines what the game could be like if it were more biologically realistic.


The first, and probably largest, biological problem is that evolution appears to occur solely by the player adding specified characteristics to their organisms. This is not how biological evolution works.

Evolution (by natural selection) occurs through selective pressures differentially changing the fitness of organisms with different character states; those organisms with the characteristics that confer the highest fitness in their current environment survive, reproduce, and pass their genes on to the next generation at higher frequencies than other organisms, so their genes become more prevalent in the population. While I understand the reasoning behind forcing the player to choose from a specific set of characteristics, it would be more biologically accurate if the player-controlled organisms had variation appear in their characteristics as they reproduced, which the player would then have to exploit to encourage their organisms to evolve in a specific direction. Non-player organisms could function in the same manner, and thus the ecosystem of the game world would be constantly evolving and presenting new challenges to the player. The game could even have characteristics determined by mutable genes (which could be hidden from the player), adding an even more realistic touch.

Biological diversity

It sounds as though the game is entirely focused on chemoorganotrophs (organisms that obtain their energy from reduced organic molecules, e.g., other organisms), and completely ignores phototrophs (organisms that get their energy from light) and chemolithotrophs (organisms that get their energy from reduced inorganic molecules, e.g., ferrous iron or sulfates). By removing both of these types of organisms, Wright removes two critical elements of functioning ecosystems. From a gameplay perspective, directing the evolution of phototrophs would pose its own unique challenges, such as attempting to develop defenses against the herbivores of the world, and if chemolithotrophs were allowed, players could evolve an organism that lived on the technological products of other organisms in the game world.

The review mentions that players can attempt to invade worlds inhabited by other players' organisms, and this appears to occur primarily by animal-based fighting. What if, instead of fighting with lasers and spaceships, players could evolve an invasive species of inedible plants and colonize the enemy planet with it, out-competing the planet's native plants and starving the enemy organisms of their food? Or, what if players could evolve some chemolitotrophic bacteria that would feed on the technological structures created by the enemy race, and thus reduce the enemy race's spaceships and machine guns to piles of rubble before the invading forces ever set foot on the planet? Or, even better, what if a player's primary organism was the chemolithotrophic bacteria, and they attempted to spread through the universe by feeding on other players' technology?

The game completely ignores non-animal chemoorganotrophic organisms like fungi, which still get energy from digesting other organisms, yet which grow and survive in a completely different manner (their bodies are composed of long, filamentous strands that grow through whatever structure they're living in). There are even some predatory fungi (that set traps for nematode worms), so it could be neat, and biologically realistic, to present players with the option to evolve into predatory fungi, which might even include the ability to feed on other player-created organisms in the world.

The game also seems to completely ignore parasites, which are some of the most interesting animals around. Allowing players to evolve creatures into parasites would drastically increase the number of possible interactions available in the game world. Going back to my interstellar colonization example, instead of sending down an army to invade a planet, a player could "evolve" a few species of parasites that could either weaken or kill the enemy species, or possibly even take control of them (e.g., Stargate's Goa'uld).

Vertebrates, vertebrates, vertebrates ...

The game directs players not just toward animals, but toward vertebrate animals. GameSpy lists a few of the possible evolutionary characteristics:
"Wright could give his creature extra vertebrae, he could give it fins or tails to move faster, he could add claws or extra mouths."

"More importantly, you could add functional elements, like heads, mouths, eyes, tails, fins, claws, even legs and feet."
Notably absent from this list are characteristics found primarily in invertebrate animals, such as tracheae, tentacles, shells, exoskeletons, jet propulsion systems, various invertebrate mouthparts (e.g., piercing tubes, muscular pharynxes), and antennae.

The animals described in the article are all vertebrate animals; however, vertebrate animals are not the most successful animals on Earth. In both number of named species and current population levels, arthropods, with their exoskeletons and jointed appendages, are the animal rulers of the planet (though bacteria outstrip any animal in population levels), and arthropods are arguably the best adapted organisms to terrestrial life. It's distressing that players might be prevented from evolving their organisms into one of the most successful lineages on the planet.

Ecology, ethology, and spaceflight

The game also seems to link specific morphological characteristics to ecological niche changes:
"Wright proceeded to add not two, but three legs to his creature. Then he let it loose. Now, suddenly, his creature could walk. And he did so -- he walked right out of the sea and onto the land."
As a minor note, this quote implies that two legs are the default for living on land; this is not the case. 99+% of terrestrial animal species with limbs walk on all four, six, eight, or more, of them, and even humans have four limbs. There are no completely terrestrial organisms extant (that I can think of) that have only two limbs.

More importantly, this quote seems to imply that legs are a prerequisite for terrestriality; they are not. Terrestrial mollusks (snails and slugs) and annelids (worms) do just fine without any limbs, and thus it would be fascinating to give players multiple options through which they could become terrestrial. Terrestrial mollusk bodies are highly mutable and expandable, and one can imagine that if mollusks were given the right "evolutionary" pushes, players could develop a wide array of functional terrestrial body plans. Also, there are many aquatic organisms that have legs (e.g. crustaceans), so organisms that evolve legs shouldn't necessarily switch to living on land. Why not allow civilizations of organisms to evolve in the oceans?

Once players have evolved a terrestrial creature, the game appears to take a very human-focused viewpoint, implying that evolving sentience and sociality is the ultimate step in evolution (since without it you can't advance to the games next stages):
"We thought Wright had made his point but were surprised when he went back into the game and showed us how it was about to take a weird turn. See, along with adding extra legs and mouths and tails and claws to your creature, you could also slowly invest in the creature's brain power. When you maxed out the creature's brain, it suddenly developed ... sentience!

"And here, the game shifted focus. Instead of managing a single creature, you were suddenly in charge of a whole tribe of your creatures. Wright's odd three-legged little critters danced around a hut in the center of a village. You could still evolve your creatures, but now, instead of buying new appendages or physical features, you bought them things like fire or weapons. Wright built a fire pit and his critters danced around it (making up a three-legged dance as they went along.) He built a drum and they started playing with it."
While evolving sociality has certainly helped a few specific lineages become more successful (e.g., Hymenoptera [bees, ants, and wasps], Isoptera [termites], Mammalia), the vast majority of organisms on Earth are highly successful with neither complex social behaviors nor high intelligence. Thus, if you have a game that lets players design whatever organism they want, there should be no preference given to organisms that have intelligence or social behavior.

If the game designers wanted players to be able to travel in space and colonize new worlds (so they could advance to the next game stages), why not present players with the option of "evolving" characteristics that would allow survival in space without technology? One could imagine space-faring dragonflies (ala Lexx), hymenopterans (ala Starcraft), or any other number of fantastic creatures, all of which would significantly increase the strategic components involved in interspecific relations (e.g., how does a dragonfly communicate with, or fight, a race of large mammals?)

A wrapup

Wright has used a grounding in reality to create some of his earlier successes; one of the reasons I enjoyed playing The Sims was because it was based on research investigating how humans actually spend their time, not how humans want to spend their time. If Wright were to provide players with a game strongly grounded in biological theory, and allowed players to evolve organisms with the true range of biological diversity, the game would have a tremendous array of possible trajectories for players to follow. All of these trajectories could be successful in their own way, making the game much more enjoyable and replayable.

1 comment:

Radagast said...

The SIMS programs are commercial games that intentionally place the human user at the center as the intelligent decision maker and (in Spore) designer. Even though they use the terms of evolution, they do NOT pretend to simulate Darwinian evolution. At best, they show how one or two sub-processes of evolution might work. A true evolution simulation must create a starting set of living organisms, let it 'simmer' for (say) 500,000 years, then see what has 'evolved'.

The comment that "free evolution without external interference is next to impossible" is a tautology (and hence correct but trivial). Darwinian evolution means adaptation of life in its various forms to its immediate environment. Therefore, a correct simulation of evolution will always end up adapting the species to its environment, and so the environment ALWAYS INTERFERES.
May 7, 2005, 7:48:35 AM PDT – Like – Reply

PZ Myers
I'm also pessimistic. They've been down this road before, with Sim Earth and Sim Life, which were just awful. Whoever they've got writing these things doesn't know much biology, and hammering them into their misconceptions made them failures as games, as well.

Sim Ant was entertaining for a little while, but ultimately was so limited in what you could do with it that it palled.
May 3, 2005, 8:39:23 AM PDT – Like – Reply

The game sounds fascinating, and I completely agree that it has huge potential to expand down the avenues you've outlined.

I have a problem with using "evolution" to describe what the game is doing though. Even if the game were modified to cover all the possibilities you suggest, the creatures' development would still be under the control of an "intelligent designer", so it's more like breeding than evolution.

Many years ago I tinkered with artificial life programming and read up on what AL researchers were doing. Producing an algorithm which allows free evolution without external interference is next to impossible.

I developed a program which had critters, as we called them, exploring a landscape, competing for food and resources, and breeding. It was an elegant program, and a lot of fun to watch, but the critters were heavily constrained by the data structures we gave them.
May 3, 2005, 6:13:55 AM PDT