Yes, it's time for another outing of Free Advice for Struggling Writers, our semi-regular writing advice column, which I'm hoping to make more regular.

You remember the Topic Sentence — when you were an impressionable young scribble-ninja, your schoolteachers pounded it into you. It's the thing where you start (or occasionally end) a paragraph with a single pithy sentence that sums up the rest of the paragraph. It's the pastry shell into which the paragraph's meat is stuffed. This paragraph does not begin with a topic sentence, as such.

It's very tempting to use topic sentences in your fiction, and sometimes it may actually make sense. When you're doing a chunk of exposition, for example, the topic sentence may actually make the infoshovel go down more easily. You can start a paragraph by saying, "The Elysian Confederacy grew quickly in the outer rim of the galaxy, but then ran into trouble." And the rest of the paragraph explains how quickly the Confederacy grew, and what sort of trouble it encountered.

But when you get farther away from the essay format, the topic sentence becomes more and more of a treacherous friend. Take this paragraph:

Captain Samson was in a terrible mood. He stormed around the flight deck, barking semi-intelligible instructions at Navigator Ingalls and Second Gunnery Master Jordan. Samson nearly maimed Chief Petty Officer Jorgenson during weapons trials. And then, during a tactical session with the ship's five strategic telepaths, the Captain broadcast thoughts of such astounding belligerence, Senior Telepath Vonox actually fainted.

So what's wrong with that paragraph? Well, it's pretty badly written in general — sorry about that. Whenever I try to write "example" paragraphs from non-existent fiction, my writing gets much worse for some reason. But the main thing that's wrong with it is that the first sentence is completely redundant. If the rest of the paragraph is even remotely well written, you will already know that Captain Samson is in a terrible mood without needing to be told.

But the reverse is just as likely to happen. You can have a paragraph where the first sentence says everything that you need to say, and then the rest of the paragraph elaborates needlessly:

Every stomp of Jordan's boots on the crumbling shale announced that he wasn't going to let some ancient book tell him what to do with his life. He strode ahead of the rest of us, and his heavy gait was full of defiance in the face of prophecy. He kicked rocks out of his way and muttered under his breath about long-dead gods who should keep their opinions to themselves.

That first sentence can stand alone — or can be combined with other sentences which actually convey some new information. But trying to turn it into a topic sentence results in a dull, repetitive paragraph.

Another huge problem with topic sentences I've noticed, in my own writing and other people's, is the tendency to separate out "theme" and "action." Let's say you have a paragraph where the topic sentence contains meaningful context, and the following sentences all describe actions, or circumstances, that we need to know about, and which only make sense if you know the context. It still may make more sense to intersperse the "context" stuff with the "action" stuff a bit more, rather than maintaining some kind of artificial, ninth-grade-essay separation from "topic" and substance."

But for sure, there are some good uses of the topic sentence as well. Even at the best of times, it smacks of "telling" rather than "showing," but there's nothing wrong with a little telling if you do it with style. It can be fun to set up a theme with a clever opening sentence, and then provide variations and amplifications on it in the sentences that follow. And if you're especially sly, your topic sentence can set up an expectation, which the rest of the paragraph then subverts as much as it bears out.

My personal favorite, among topic sentences that actually do some heavy lifting, is the quote followed by the narrator's own observations that bear it out. Like so:

"The ghosts of Delta Prime do not need your pity, Earthling," Allura told me, and it was true. The ghosts were too busy going to their ghostly shopping malls, getting spectral manicures and pedicures, and playing undead bingo, to wonder what I thought of them. I saw them everywhere: holding remembrance services for the living, dancing on the planet's shifting air currents, and huffing the exhaust from my land cruiser, which seemed to affect them like really good sake.

I guess my main point about topic sentences is: Don't overuse them. It can be tempting, in prose fiction, to try and sound pithy, or to signpost for the reader what's going on at any given time. But too much of this sort of tour-guiding in your stories can bore and annoy the reader, even if he/she isn't consciously aware of the reason. A little bit of topic sentencing, here and there, can feel delightfully retro and whimsical. (Chosen at random, here's a lovely topic sentence from P.G. Wodehouse: "George Mackintosh did not, I am glad to say, carry out his mad project to the letter.") But too often, they come across as "telling you what I'm going to tell you, and then telling you what I'm telling you." Best to get the news out there straight away, without any harrumphing and prefacing, most of the time.

So that's this week's Free Advice for Struggling Writers. Next week, if I can bestir myself to finish it, you'll get my dissertation on "denoument versus picking at a scab." So what do you think about topic sentences?

Awesome pulp cover scans by Matthew Kirscht on Flickr.

Welcome to Free Advice for Struggling Writers (as if there's any other kind), a new semi-regular column here at io9. We've taken a few stabs doing a writing advice column for ages, but I always felt a bit presumptuous giving anybody else advice about how to write. But now that I'm in the middle of revising a fantasy novel, I'm having all sorts of thoughts about novel-writing that might actually be helpful. And which might, at least, start some interesting conversations.

So — a lot of people seem to regard the "elevator pitch" as a necessary evil. And I've certainly thought of it that way myself. OMG, my future society has 29 social classes, and there's a war between ninth-dimensional nano-roaches and sentient clouds that only live between galaxies! How can I possibly scrunch all that greatness down into two sentences?

But I'm going to try and explain, in about 800 words, why the elevator pitch is a really good thing to think about — and why it should be woven into the fabric of your novel as you're writing it.

People often think of the "elevator pitch" as being some kind of crass Hollywood thing: like "It's Rain Man – With Time Travel! It's 28 Days Later meets 9 1/2 Weeks!" But really, the "elevator pitch" just summarizes what your story is about. And if you can't explain what your story is about in a couple sentences, maybe that's actually a bad sign.

Here's how I've gotten to think about the elevator pitch lately: It's a contract with your readers.

Forget the agent you're going to pitch your book to, forget the editor, forget all the people between you and the reader. Eventually, if you're lucky, readers are going to be picking up your book based on some version of the same pitch. (You can't really control your back cover copy, but you'll do a lot of the marketing of your book anyway.)

And when you describe your book in a few pithy sentences, you're making a promise to your potential readers, about what they'll get if they invest time and money in your book. You warrant that you'll deliver a story about a guy who gets migraines that let him rewrite the past. Or a dragon who disguises herself as a racehorse so she can take part in the Kentucky Derby. If the reader buys your book based on that pitch, you'd better deliver on it.

And that means that your "dragon disguised as racehorse" book really has to be about the dragon disguised as the racehorse. Your reader is going to expect the dragon to be the main character, and there should be some compelling reason the dragon wants to be a race horse, and then we should see the dragon's quest for a horse disguise, and how the dragon learns to live among horses. The reader is going to be wondering if any of the horses, or jockeys or owners, see through the dragon's glamour (or plastic surgery?). The reader will be counting on there being some kind of arc where the dragon changes, and accomplishes something, in the course of becoming a thoroughbred. And most people who've consumed sporting narratives in the past will expect the book to end with a huge, crucial, suspenseful race, which the dragon/horse either wins, or loses in some poignant fashion. And you'd have at least a sneaking suspicion that some people will learn the dragon's true identity at the end.

But wait – in the course of writing this novel, you get fascinated by the culture of the jockeys. And you invent a marvelous backstory for the dragon's jockey, but also a whole subplot involving a different jockey, who's part gnome. And also – there's a bookie who's using magic to figure out the odds. And once you think about the backstory of your dragon main character, you invent a whole dragon culture, and a dragon civil war, and you decide that dragons are facing extinction because people are trafficking in dragon eggs. And so on. These subplots, and digressions, and possibly philosophical lines of inquiry, are marvelous and should absolutely be in your novel. It's part of the joy of speculative fiction that you can draw such a broad canvas – but as you're writing your novel, remember that your readers are expecting it to be about the dragon who becomes a racehorse.

Look at it this way – a decent writer can deliver the dragon/racehorse story in its barest outlines, carrying the reader from A to B to C like a dependable mount. A good writer can give you the dragon/racehorse story, but throw in tons of complexity and subplots and exploding ideas and supporting characters and world-building, until it's really the story of a whole world. But it takes a really brilliant writer to give you that whole tapestry of complexity, but still make the story about the dragon who becomes a racehorse.

And I'm by no means saying you have to follow your readers' expectations for how your dragon/racehorse story will go – in fact, part of trying to be the brilliant writer, instead of the decent writer or the good writer, is surprising the reader. But here's the key: You have to know what the reader's expectations are, and then make a conscious decision to subvert them. If you ignore the reader's expectations, and then your story just goes off the track (so to speak), you risk losing the reader. Say halfway through, your dragon/racehorse breaks a leg and gets shot in the head. If you haven't set up that development at all, the reader may just feel cheated, not surprised by your clever twist.

So the reader is expecting your dragon to face challenges learning to carry a small human around a racetrack, but eventually surmount them and build up to the big race at the end. There are a million ways you could have something different happen, but you still need to be following the story beats that lead to the Big Race, while laying the seeds of your surprising twist.

So how do you weave in all the backstory and subplots and thoughts about the humaneness of horse-racing, without having them overwhelm your "A" plot? That's a huge topic for another column (and possibly for the comments on this one, if you like) but in a nutshell: if you stick in that stuff artfully, it can help build suspense in your main plot. Just as your suspense reaches fever pitch – OMG! The dragon has a sprained ankle! – you cut away to one of your subplots, or go off into a digression about the fastest dragon who ever lived, and how that dragon failed to outpace the egg-traffickers. But also, have a certain amount of faith in your reader, that as long as you're hitting the story beats about your dragon protagonist and putting that stuff front and center, the reader will be able to recognize that as the main plot, even if there's 10 pages of other stuff sandwiched between scenes with your main dragon.

The bottom line is, stop thinking of the "elevator pitch" as a necessary evil that you're going to use to hoodwink some agent into thinking your book is Seabiscuit meets The Last Dragon, when it's really so grand and multilayered, you need a whole book just to describe it. The "elevator pitch" isn't just how you market your book – it can also be a way to think of what your book is really actually about.

Pulp magazine covers from Toyranch and Terry McCombs on Flickr.

I wouldn't claim to be an expert on short fiction writing, but I have written over a hundred of the little fuckers, a large proportion of which have been science fiction-y. Here are a bunch of do's and don'ts, that I discovered the hardest way possible.

World-building should be quick and merciless. In a novel, you can spend ten pages explaining how the 29th Galactic Congress established a Peacekeeping Force to regulate the use of interstitial jumpgates, and this Peacekeeping Force evolved over the course of a century to include A.I.s in its command structure, etc. etc. In a short story, you really need to hang your scenery as fast as possible. My friend and mentor d.g.k. goldberg always cited the Heinlein line: "The door dilated," which tells you a lot about the surroundings in three words. Little oblique references to stuff your characters take for granted can go a long way.

Make us believe there's a world beyond your characters' surroundings. Even though you can't spend tons of time on world-building, you have to include enough little touches to make us believe there's stuff we're not seeing. It's like the difference between the fake house-fronts in a cowboy movie and actual houses. We should glimpse little bits of your universe, that don't necessarily relate to your characters' obsessions.

Fuck your characters up. A little. Just like with worldbuilding, you can't necessarily devote pages to your characters' childhoods and what kind of underwear they wear under their boiler suits. Unless your story is really a character study with a bit of a science fiction plot. I used to have a worksheet that included spaces to fill in in info about each character's favorite music, hatiest color, etc. etc. Never filled those out. If I'd tried to force myself to come up with a favorite color for every character, I would have given up writing. But do try to spend a bit of time giving all of your characters some baggage, just enough to make them interesting. Most science fiction readers are interested in characters who solve problems and think positively, but that doesn't mean they can't have some damage.

Dive right in — but don't sign-post your plot in big letters. When I started writing stories, my early efforts meandered around for pages before something happened to one of the characters to make him/her freak out. And then the rest of the story would be the character(s) dealing with that problem. And then, as I got more practiced, I found the foolproof map to awesome storytelling: introduce whatever it was that was freaking out my characters in the very first sentence of the story! And then the story could be about them dealing with that problem, until they solved it in the very end. It was so perfect, how could it fail? It took me another year or two to realize that plunging the characters into the story's main conflict right away was just as boring, in its own way, as the ten pages of wandering in circles. The best short stories I've read are ones which start in the thick of things, but still keep you guessing and let you get to know the characters before you fully comprehend the trouble they're in.

Experiment with form. Short fiction isn't one form, it's a whole bunch of forms jammed together according to their length. Short stories include your standard 3,000 word mini-odyssey thru the psyche. But they also include flash fiction (sometimes defined as under 100 words, sometimes under 500 or even under 1,000.) And those wacky list things that McSweeney's runs sometimes. In fact, for a while there, postmodern short fiction was all about the list, or the footnotes, or the krazy monologue, or the story told in office memos. Try writing super-short stories of only 10 words, or mutant essay-stories written by a fictional person. Also, if you always write third person, try first person. Or if you're always doing first person, try third.

Think beyond genre. Often the best genre fiction is the stuff that cross-germinates. Pretend you're actually writing your story for the New Yorker, and try to channel George Saunders or even Alice Munro. See how far you can go towards writing a pure lit piece while still including some elements of speculation. Or try writing your story as a romance. Or a mystery. Imagine it as a Sundancey indy movie.

Don't confuse your gimmick with your plot. You may have a great idea for a piece of future technology, or some amazing mutation that turns a whole bunch of people into musicvores who survive by eating your memories of rock concerts. Maybe you have the most original basic premise evar — but that's not your plot. Your plot is how your new widget changes the people in your story, and how it affects their lives. Or what decisions your people make as a result of this new technological breakthrough.

Don't fall into the character-based/plot-based dichotomy. People, especially in writing groups and workshops, will try to categorize stories as based on either plot or character. This is a poisonous idea that will turn you into a cannibalistic freak wearing a belt made out of human spinal cords. There's no such thing as a character-based story or a plot-based story, because every story has both. Even the most incident-free Ploughshares romp or the most twisty thumpy space opera tale. If you start thinking that stories can be categorized into either pile, you'll end up writing either eventless character studies or plot-hammer symphonies starring one-dimensional nothings.