The following comes out of several years' experience in teaching science fiction and fantasy writing at George Brown College in Toronto. I've talked to a lot of writers about how they envision the worlds they write about. My ideas below are based on what has worked for me and others, and what hasn't.

What a world is

The most important thing to realize is that fictional worlds do not exist on paper. They exist in the mind of the reader, as a result of cues and guidelines provided by the writer. After all, no matter how exhaustive your description of a world might be on paper, it will not "come to life" until the reader takes it as his own, and visits it in his imagination. It is the reader who will ultimately see the colour of the sky, and smell the scent of a new earth. Your task as the writer is not to do this for him (except in a preliminary manner) but to provide the means for him to do it himself.

In other words, your experience of the world will not be the same as the reader's. It is the reader's task to visit the world. It is your task to point to it.

In order to point to a new world, you will have visited it yourself. But most of the SF writers I know do not experience this in quite the same way as their readers.

The "backless maiden"

There is a story in the Arthurian cycle of a knight who was tempted by a beautiful maiden who, however, would never turn her back on him. When she brought him to her house she cooked him a fine meal. But as she stood before the fireplace, he saw the light of the fire behind her eyes, and realized she would not turn away from him because she had no back: she was the animate mask of a woman, and he was in great danger.

A science-fictional text is a "backless maiden", and one of our chief dangers as writers lies in not realizing it. There is no more to any fictional world than the words on the page. The world comes to life because of the inferences and associations we make based on those words. So, the task of the SF writer is not to exhaustively describe, it is to find that set of images, metaphors and scenarios that will inspire the reader to see more than the words themselves denote.

If you think about it, this kind of lets us off the hook. The reader is our active collaborator. Provided we give the right hints, the reader will create the world in all its stunning detail. So, it is not necessary for us to imagine everything that the reader might want to experience. The reader will fill in the blanks. It is the writer's responsibility to make sure that those dots are lined up properly, so that the reader doesn't end up imagining a world that doesn't go with the story the writer wanted to tell.

The problem of technical expertise

Many beginning SF writers assume that they have to know all the sciences, in depth, in order to write SF. My reply has usually been, "your only obligation is not to be wrong in your science," and even this requirement can be relaxed if you find that bending science is the only way to get at the theme you want to express. If your story relies on some feature of quantum mechanics then you should know your QM. If your story uses modern physics you should endeavour at least not to use it wrongly. But you are not a scientist; if you think your job is to educate the reader in science, then write essays. Fiction is ultimately about people.

I always get arguments about this point. A lot of SF writers will disagree with it. What, though, is the one element in SF stories that has proven the most ephemeral over the years? What has changed the most? It is the so-called "eternal truths" of scientific theory. Early stories used the existence of the ether, or N-rays, or broadcast power, which were all seen as solidly possible at the time. We can still read these stories and enjoy them, though, and even be moved by them (take, for instance, The War of the Worlds) even when their science has been disproven. This is simply because they work as stories; that is, they are about people.

The reality you describe ultimately has to be at the service of the story. Starting from the opposite premise is the single biggest reason I've seen for stories failing at the level of world building.

What beginning writers tend to do

For my second attempt at a novel, I took notes. Lots of notes. I ended up with more pages of notes, in fact, than there were pages of novel. I had everything worked out about the milieu--its economy, climate, history. The story itself never really worked, and it was only much later that I realized I had been putting all my creative energy into world building, nearly none into story telling.

I know a lot of writers who have spent years developing a universe for their stories. The game is great fun, after all. These writers have written encyclopedias about their imaginary worlds, and the process of thinking the place through produces countless story ideas. This is good. All too often, though, the writer finds all this detail crippling when it comes down to actually writing a story or novel. The facts about the place have been decided, and everything the writer produces has to conform to them. Half way into a story, the writer finds that some crucial event can't happen because that would violate the history, or physical structure or social mores of the invented world.

When this happens with novels, the writer often ends up creating highly convoluted (and unsatisfying) plot structures because he wants to explain everything to his own satisfaction.

There is such a thing as satisfying yourself. There is also such a thing as satisfying the reader. The only way to satisfy the reader is to ensure the internal consistency of your story. By requiring that the story conform to your world-plan, you place it on the Procrustean bed of external consistency. This is fine for realistic novels set within historical time, because in those the reader will feel internal consistency is violated if you violate an external consistency (historical fact) that the reader is familiar with. SF supports itself entirely by means of its internal logic; there are very few SF paradigms so familiar that the reader will be guaranteed to know them (hyperdrive being one such exception).

The reader doesn't know the full background of your world, and is unlikely to care until after you have engaged his emotions with the strength of your story. The Lord of the Rings has a very intricate and detailed background, but the background is interesting because the story itself is solid.

For this reason, when I work, I start with an imagined world, which inspires story ideas. When it comes to writing, I demand that the world change to suit the needs of the story. No matter how much effort I may have put into imagining the world, it can all be discarded if it doesn't serve the needs of the plot, characters, and theme.

If you'd like to study this further, take two examples: the first being Larry Niven's Known Space series, the second the universe of John Varley's Invasion stories. Niven required that each of his stories be consistent with his previous work; the final result of this was that he ran out of options for new stories in this universe. In contrast, Varley's novel Steel Beach belongs partially to the same universe as some of his other stories, but diverges wherever it needs to for internal consistency.

Obviously, of the two strategies, I advocate Varley's.

Creating dramatic milieus

The best imaginary worlds (for us) are those that are dramatic in their very foundation. When the world itself is designed to produce dramatic situations, you can mine it creatively for years.

The TV series Star Trek is an example of a milieu designed on dramatic principles. The transporter, for instance, exists because you can't waste screen time with your characters flying everywhere. It allows smooth scene transitions. The holodeck permits adventure outside the normal environment of the ship. The very fact that the series was set on an exploratory vessel meant stories would arise naturally.

Not all interesting ideas are dramatic, and not all imaginary worlds naturally generate conflict. I have thought out in detail a parallel Earth where Tesla's science was accurate; this world has broadcast power, and wars are fought by remote control. Unfortunately, the ideas, while interesting, don't of themselves generate conflicts. I could set a story in this world, but the motivating force behind the story would (at least if I wrote it now) have to come from something other than Tesla's science. So why include it, why not just write a mainstream story in the first place? If the ideas are decoupled from the story, they will just get in its way.

I work best by imagining constraints. For instance, imagine that the aliens arrive and discover to their horror that we are omnivores, unlike everybody else in the galaxy. They place the solar system in quarantine. Now, if we look at any time from that moment forward, all kinds of potential stories appear: the eugenics campaign to genetically reengineer humans as vegetarians, and the reaction against it; attempts to make do with the resources of the solar system as the human population climbs into the trillions; encounters with the aliens at the edge of the system; attempts to break the quarantine, and resistance to this at home and from outside. It's a naturally dramatic world.

The story makes the world

Full-fledged world building is a fun, but not necessarily useful activity for a writer. I imagine the worlds I write about in detail, but it's usually after I've begun a story, and the story constrains the possibilities of the world. As a writer, I feel my creative energies are best directed at the story itself, because if I tell that well enough, the readers will create their own world, more completely and intensely realized than I could tell it to them.

Ultimately, the only place your literary world will thrive is in the imagination of your reader.