Don’t count on it

Last night I wound up shaving some words off a chunky short story in order to meet the 5k word limit for submission to a magazine. Obviously, when you’re operating with a tiny margin of error, precision becomes more important than when you simply need a ballpark figure for your own edification.  While checking the manuscript in several programs and webapps, I discovered something curious that led me to conduct a little experiment. As a test, I copied and pasted the same manuscript into a half dozen applications that determine word count, among other statistics.

Here’s the word count total as determined by Microsoft Word 2010: wordcount_word

Here are the same words pasted into Google Docs:wordcount_docs

OK, not exactly the same but close. How about a few webapps?

The word count from

And from


And, finally, from

So, we’re talking about a variation of 224 words across six different tools.  This file was in standard manuscript format, with no diacritical marks or other funky special characters. One explanation I can think of is that, because a few italicized passages are underlined, some parsers do not consider an underline as a word break.  However, this is the standard manuscript specification, so you’d think a word count tool would take that into account.  And I still find it curious that no two applications returned the same word count.  Clearly not all parsers are created equal; in fact, it would appear that none of them are!

Caveat scriptorus!  Especially if you’re skating on the edge of a word count limit.