Spore: My View of the Elephant

A few weeks ago and with little fanfare, Spore turned five-years-old. The game was announced at GDC 2005 during Will Wright’s annual mind-blowing speech on whatever floats through his head. The initial concept – of a game in which the player evolves a species from cellular development to galactic dominion – generated an immense amount of hype, which the game struggled to fulfill upon its 2008 release. Spore received middling reviews from the gaming press, who found the gameplay weak and unfocused, and harsh criticism from the scientific press, who felt tricked by the promise of a game built from real science.

For myself, the time is now right to put down my own thoughts on Spore’s development – my memories of the project are still fresh, yet enough time has passed to ensure that criticism doesn’t impact active teams. I joined Spore in May 2007 for what ended up as the final 15 months of the project; however, the team started the game in 2000, which meant that I saw just 20% of the complete story.

Thus, my view of the game’s development is inevitably incomplete – bringing to mind the parable of the blind men and the elephant – and needs to be viewed from that perspective. I would welcome – indeed, encourage – other members of the Spore team to speak up on their own experiences with the project, especially if their perspectives differ from my own. Nonetheless, here are four lessons from my time with Spore.

1 – Don’t be afraid to challenge the initial vision

Ultimately, Spore was about two big ideas – powers of ten and procedural content. The first idea refers to the classic short film by American designers Charles and Ray Eames, which zooms in, by powers of ten, on a man and a woman until reaching quarks and then zooms out to the entire universe. This film inspired Will to create a game with similar radical shifts in scale, jumping from a cell to a creature, then to a tribe, then to a civilization, and finally to a space-faring empire.

The other idea – procedural content – was that all content in the game (such as creatures, vehicles, and buildings) could be represented with just a few kilobytes of data – which was, in Will’s words, “the DNA template of a creature while the game, like a womb, builds the ‘phenotypes‘ of the animal, which represent a few megabytes of texturing, animation, etc.” From this seed grew the powerful editors (which enabled some subversive creativity), procedural animation (which could truly handle anything), and content pollination (which shared the community’s best works).

When Will started developing the game, the core idea was powers of ten, reflected in the game’s original title, SimEverything, which promised a game at every zoom level. While prototyping that game, procedural content emerged as a way to fill the player’s universe, and that concept kept growing and expanding until it wasn’t clear anymore which concept was Spore’s big idea. A game can have two big ideas, of course, but the problem was that only one of these ideas was any good.

Spore’s biggest issue was that the play at each stage was fairly shallow because the team was making five games at once. (At one point, Will described each of the game’s five stages as light versions of classics – cell is like Pac-Man, creature is Diablo, tribe isPopulous, civilization is Civilization, and space is Masters of Orion.) However, making five different games at once is a bad idea; making one good game is usually hard enough.

Each of the five stages had different controls, different interfaces, different nouns, different verbs, different goals, and so on. Some effort was made, of course, to share ideas and elements across stages; however, the compromises involved often watered down what was supposed to make each stage distinct in the first place. For example, each stage required a friendly means of engaging with other entities; in the creature stage, this mechanic became dancing for other creatures to make friends while, in the civilization stage, this mechanic translated into attacking other cities with music instead of bullets. Neither mechanic was the best idea for its own individual stage, and the justification was high-level consistency. Thus, the powers of ten idea put the team in a state of perpetual compromise where every major decision had to be considered according to its effect across all five stages.

On the other hand, procedural content was a genuinely interesting and fertile idea – one which was novel for the time, appropriate for a game about evolution, and rich with gameplay possibilities. The tragedy of Spore is that the team never re-evaluated its first big idea in comparison to its second one. Indeed, one of the problems with traditional, siloed game development is that initial assumptions are rarely challenged as the game is never exposed to the oxygen of actual player feedback.

Focus is an important asset for a team; if the game’s scope could have been reduced to just the biological stages (cell and creature), the team could have focused on fully exploring the intersection of procedural content and evolutionary gameplay. In the later, social stages, the editors served a mostly cosmetic role anyway, which pushed them to the background. Unfortunately, the best thing about powers of ten was that it soundedlike a great idea, generating a huge amount of hype and press, so the die was cast at the 2005 reveal. What makes players buy a game, however, is often not the same thing as what actually makes them play it.

2 – Gameplay must support the theme

I have written before on the importance of a game’s theme matching its actual gameplay, and that the mechanics can easily subvert the intended meaning of a game, regardless of the designer’s stated goals. Indeed, I wrote about how this dissonance affected Spore:

The reception of Spore, a game sold with an evolutionary theme, provides a recent example. In the October 2008 issue of Science magazine, John Bohannon wrote the following about how the game delivered on the theme’s promise:

I’ve been playing Spore with a team of scientists, grading the game on each of its scientific themes. When it comes to biology, and particularly evolution,Spore failed miserably. According to the scientists, the problem isn’t just thatSpore dumbs down the science or gets a few things wrong – it’s meant to be a game, after all – but rather, it gets most of biology badly, needlessly, and often bizarrely wrong.

The source of this dissonance is that, even though it was sold as such, Spore is not really a game about evolution. Spore is actually a game about creativity – the reason to play the game was to behold the wonder of other players’ imaginations as they used (and misused) the editors to create objects not imagined by the game’s designers – from musical instruments to fantastical creatures to dramatic scenes.

Spore didn’t need to be marketed or sold as a game about evolution, but since it was, players’ expectations had to be anticipated. Although one might not be surprised that the game was a disappointment to actual scientists, the crucial decision to limit the impact of the editor on gameplay ensured that players would not be able to experience the fantasy of evolution – that the editor would enable the creation of an infinite number of unique creatures, with behavior and performance dependent on player choice.

Of course, the editor enables an infinite number of visually distinct creatures, but the gameplay effects of the creature parts are unfortunately quite discrete. The feet components each carry with them a canned set of attributes – for example, Stubbtoe gives “Sprint 2,” “Dance 1,” and “Speed 2″ – regardless of the position of the foot, the length of the attached limb, or the shape of the body. Thus, the attributes of each creature is simply a summation of all the named body parts, and although the procedural animation guarantees that a many-limbed creature will walk convincingly, the player’s creativity in designing the creature’s shape has no impact on actual gameplay.

This disconnect stand in sharp contrast to the cell stage, which does deliver an editor with consequence. The exact position of each Proboscis, Flagella, and Cilia matters as the player-designed cell swims along, chomping prey while avoiding predators. Thus, the cell editor delivers actual gameplay that the creature editor does not, a key expectation for players.

How did the creature editor lose its bite? Obviously, the 3D world of the creature stage posed a greater challenge than the 2D world of the cell stage, but reducing the game’s scope to just the biological stages could have helped considerably. However, the root issue was a philosophical debate about the role of the editor in the game. Should the editor enable unparalleled aesthetic customization, at the expense of gameplay consequence, or should the game mechanics support every choice made by the player, even if that meant limiting the flexibility of the editor? Should Spore be an interactive art museum or a customizable video game?

The question is akin to asking if Spore is a game or a toy, which is, in fact, one question that got asked a lot during the game’s development, often without a clear answer. For players able to put imagination before gameplay, Spore is a magical experience; indeed, the game is at its best when played by ten-year-olds. However, for core gamers expecting a game about evolution (or, perhaps more accurately, a God-game about intelligent design), Spore fell short.

3 – The only prototype which matters is the game

One distinctive element of Spore’s development was a focus on prototypes to test out design ideas quickly and efficiently. Chris Hecker and Chaim Gingold gave a very well-received talk at GDC 2006 on this topic; they demonstrated one important prototype which proved that the creature editor could work, that editing a creature in 3D by pulling, prodding, and stretching various body parts was fun.

Generally speaking, prototyping is a great idea; the process saves time and money, focuses the developers on tangible problems, and suggests ideas that would never emerge from a design document. However, unless a prototype is meant to answer a very specific and relevant question – such as whether the creature editor will feel right – an over-abundance of prototypes can lead to a false sense of progression. 100 compelling yet limited prototypes are less likely to lead to a great game than a single playable one.

The team eventually posted fourteen prototypes for players to try out, and what is notable about the selection is how few of them had a meaningful impact on the final game. I do remember Gonzaga/SPUG being used by the creature team, but much of the game’s core design was still up in the air when I joined, with most of the old prototypes long forgotten. (The civilization stage, for example, was just a tech demo of a spherical world.) A game must be greater than the sum of its parts, and the gameplay systems can only be understood when they exist within one complete experience, regardless of the shortcomings of the current technology.

Sid describes this process as “finding the fun” as he is pulling and prodding his prototype into a playable game; however, he is working on a single prototype, which will eventually morph into the final game. For Civilization 4, I built the game on top of a cancelled RTS project, which allowed us to have a playable game working, using 2D billboards for art, within months. Jon Shafer prototyped Civilization 5 within the Civ4codebase, only switching over to the new graphics engine when it was ready. Visitors to Blizzard are regularly shocked by how their games appear to be shippable years before release, which is how they balance the demands of AAA production with the importance of iterative game design. These fully playable “prototypes” signal the beginning of the most crucial part of game development – when the designers can change the game rapidly based on feedback from actual playthroughs, not disparate, standalone experiments.

Spore was not fully playable until, at best, the final year of the project. Shipping the game earlier was never an option, and shipping the game latter was politically and emotionally impossible because of the time and resources already invested in the preproduction process, which did not result in a playable game. Of course, many successful games are not fully playable until months before release, but they tend to be part of established genres or franchises; the fun was already found in previous versions, and the team simply needs to improve the core gameplay without ruining anything. Having an unplayable game until shortly before shipping is not ideal, of course, but some projects have very demanding time constraints.

Spore was given an eternity of development time, and – more importantly – was new, new, new, new, new. No game had ever been made like it, with such immense scope and without a familiar template, so the project had an immense amount of risk, which only grew greater as more and more money was spent without a fully playable prototype. Creating such a prototype for a project with as much technical innovation as Sporewould have been no small feat (much of the procedural animation, for example, was not finished until the final stretch), but no other method exists to make a fun game from scratch.