Failing Hard at Experimental Design

When you’re setting up a scientific experiment or exploring an alternative engineering solution, it’s a good idea to fix as many variables as you can so that you can better judge the ones you’re studying. I have utterly failed to exhibit this prudence in the course of writing The Grey God. Many of the writing process variables are in play. Few are fixed. This is not methodical.

  Raether’s Enzyme The Gray God
Incubation Screenplay NaNoWriMo
Drafting Planning Pantsing
Word Processor Movie Magic Screenwriter 2000 and Microsoft Word Scrivener
Notebook OneNote Scrivener
Workshopping Helium Exchange (screenplay) Scribophile
Self-editing Tools Word and Grammarly Grammarly, HemingwayApp, AutoCrit

This post will attempt to tease out the results of my ongoing experimental use of AutoCrit in the self-editing phase of the project.

Gamifying the Editing Process

AutoCrit is a suite of editing and text analysis tools that lives in the Cloud. Its advanced features are accessed via subscription. The free tier includes a text editor that is adequate for composing a novel’s manuscript. Where it gets more interesting is in the ‘Professional’ mode. For $30/month, AutoCrit will dice and slice your text with a variety of statistical and heuristic analyzers and assign your work corresponding quality scores. It will also give you a score relative to your genre and major authors. AutoCrit turns editing into a game.

Gamification is not a new phenomenon. Reframing tasks as games has a long history in education, sales, and other endeavors which include elements of competition or self-improvement. Rewarding people with scores, badges, smiley faces, and thumbs-up add game-like perks to fitness trackers and social media. AutoCrit rewards you with better scores for ‘improving’ your writing.

I put improving in quotes not to disparage what AutoCrit has to offer, but point out that optimizing your score in AutoCrit has notable limitations. The software doesn’t read your story. It can’t judge the depth of characterization, find plot holes, or critique your pacing and structure. What it can do is apply fixed rules and identify patterns that deviate from an analysis published works. It can flag word usage that is considered poor form in commercial fiction.

Working with AutoCrit is like having a precocious six-year-old niece who has taken an interest in your work. You explain to her a few of the rules you’re working with and she takes them to heart. She sits down beside you and starts pointing to the text. “What about that?” “And that?” “Right here, you have an adverb. Adverbs are bad, right?” She really gets into it. You say, “That’s okay, it’s where my character is talking, so it doesn’t need to be right.” She ignores that because she’s on the hunt. You’re just about to politely suggest she go outside and play when she points out something you missed. You realize that what she lacks in nuance, she makes up for in tenacity. Playing the game with her is productive, but it is not without risks.

Applying AutoCrit to The Gray God required exporting the manuscript to a Word .docx file and uploading it into AutoCrit. I parked Scrivener on one side of the screen and the AutoCrit browser tab on the other and worked my way down through the manuscript. The biggest initial finding was that I had an adverb problem. Sure, I didn’t overuse words ending in -ly, but I dropped in ‘just’ all over the place. Things were just outside. Just before. Just it. I used the word in just about every adverbial form it could take. I just didn’t see it until AutoCrit called me out on it and suggested I remove just about a couple of hundred uses. Cleaning that up tightened the writing and improved my score. I got into the game and was happy with the results.

I was so happy that I cracked open Raether’s Enzyme, which had been in lockdown since the proofreader worked on it. The AutoCrit pass took longer than I expected. I corrected passive voice issues. In pursuit of a higher score, I shaved over 3,000 words off the text. The manuscript was leaner and smoother than before. This was good. It was also broken.

The damage resulted from my ignorance conspiring with my greed for a better score. AutoCrit flagged ‘that’ as a (potentially) problematic filler word and recommended an aggressive target for reduction. In the course of lowering my ‘that’ count, I substituted ‘which’ for ‘that’ in numerous places. My score improved. As you might guess, the two words are not entirely interchangeable. I suspected as much and researched ‘that’ vs. ‘which’. Sigh. Yes, there is latitude for using ‘which’ in place of ‘that’, but to be right you should use ‘that’ for restrictive clauses. And you precede ‘which’ with a comma in nonrestrictive clauses. I shouldn’t have made those changes. Now I wonder what else I broke playing the AutoCrit game.

I’ll find out over the next month when the Raether manuscript comes back from the copyeditor. I may reactivate my AutoCrit Pro subscription to see how it scores the changes.