Let me explain.
Whenever writers prepare manuscripts for submission, they are usually asked to provide a word count. Historically, the reason for this was to help editors figure out how much space a story or article would require when set into type. Consequently, the editor was less concerned with the actual number of words, and more concerned with the number of lines those words occupied on the page.
Chuck Rothman's essay What Is a Word? on the SFWA webpage explains how this would work. He posits the following five word exchange of dialogue, noting that a magazine editor would consider this to be twenty-five words:
"I'm pregnant," he said.
Even though a computer's word processor would only count this as five words, the editor knew that it would take up much more space in the magazine. So writers had various methods to estimate their word count. Rothman discusses one method in his article, but one good rule of thumb is that a page of double-spaced Courier 12 point text with one-inch margins all around would equal about 250 words.
Now here's where things get interesting: writers frequently get paid by the word. Since writers aren't fools, they would usually choose a method to create the largest possible word count that they could reasonably claim for a manuscript. Publishers would generally accept the writer's word count unless things looked a little off, in which case they might apply their own algorithm and come up with a smaller number of words. (Interestingly enough, I've never heard of a publisher who recalculated a word count and ended up with a larger number of words. I leave the obvious conclusion to the reader.)
But more and more, the fact that a computer's word processor can provide an exact word count is changing the publishing definition of a word.
I first noticed this in 2004, when my novelette "Paying It Forward" (Analog, September 2003) was nominated for a Hugo in the Short Story category. Although Analog magazine categorized it as a Novelette due to the amount of space it took up via a traditional word count, the Hugo administrators had apparently chosen to count the exact number of words in the story, which placed it below the Novelette limit and into Short Story. Since the story's length was on the cusp of 7500 words, it seemed a reasonable thing to do.
But now things are changing another way. I've recently spoken to a few anthology editors who have told me that when it comes to paying for a story, publishers are now choosing to go with the computer's word count instead of a traditional word count. And since many stories for anthologies are now submitted electronically, without a paper manuscript ever being produced, publishers can run the word count tool themselves and come up with the smaller number in an instant.
Obviously this benefits the publishers more than the writers, as they can pay less for a story than they used to. But that does seem to signal a sea change, similar to the way that editors are no longer necessarily requiring manuscripts to be in 12 point Courier as long as the text is readable.
Which leads me to wonder: have other writers started to notice this change as well? And what, if anything, should we do about it?