Editing software will never replace the trained eye of an editor.
I'm an editor, so of course I'm going to say that. But I'm not saying it because I want to keep my job. Sure, there is a desire to always have money coming in, but it's more than that. I make this statement because I honestly believe it—as a writer.
When talking about editing software, I'm talking about the different apps like Grammarly, ProWritingAid, Hemingway, AutoCrit, and one I recently discovered, Fictionary.
I will be the first to admit that these programs have their uses. For a writer who is floundering in the dark, they might add a little direction. But for the writer who really doesn't know their craft, these programs can do more harm than good.
For any writer wanting to learn their craft and take their writing to the next level, nothing will replace the eye of another human being.
It's time to talk about all those automated programs that are floating around and point out the flaws—and why they can't replace a human editor.
Good writing doesn't always follow the rules.
When building software programs to analyze the way we write, the developers need to employ a set of rules to start from. They will build on these rules, enhancing the algorithms to match a writer's unique writing style, learning as they go. Sounds good in theory until you discover that even editors can't agree which set of rules to start from—and how rigidly to follow them.
To quote Benjamin Dreyer, copy chief at Random House
, "The English language…is not so easily ruled and regulated. It developed without codification, sucking up new constructions and vocabulary every time some foreigner set foot on the British Isles—to say nothing of the mischief we Americans have wreaked on it these last few centuries—and continues to evolve anarchically. It has…no enforceable laws, much less someone to enforce the laws it doesn't have."
That quote was from Dreyer's English (published 2019). It's a funny book about grammar. The entire book points out how crazy the English language is and frequently flaunts the so-called rules, pointing out how sometimes breaking those rules is the way to go.
The Style Guide Conundrum
The top editors in the world will happily point out that there really aren't any rules for the English language, rather a set of guidelines that have developed over time. And these guidelines are known as style guides.
When you hire a copyeditor to look at your manuscript, one of the first questions they should ask you is which style guide are you following. You have Chicago Manual of Style (often shortened to CMoS), The Associated Press Stylebook (or AP), New Hart's Rules from Oxford University Press, and let's not forget about the various university- and government-issued style guides. There will be a significant amount of overlap between these, but there will also be huge fundamental differences.
Take the Oxford comma, for example (also known as the serial comma). This is the comma that goes before the and in a list, e.g., apples, bananas, and oranges. CMoS and New Hart's Rules both use the Oxford comma, but AP does not.
I will grant you that things like whether or not an Oxford comma is used is something that could easily be programmed, creating a toggle to turn it off or on. But other so-called rules are not as clearly defined and are just begging to be broken. Many writers take great pleasure in blasting these vague, esoteric rules out of the water.
If you ask any copyeditor, they will tell you that while they follow the style guides as much as possible, sometimes they just follow their instincts as to what looks and reads best.
How do you program instinct?
Therein lies the problem for the automatic editing programs. How do you program instinct into artificial intelligence?
The concept of artificial intelligence uses neural networks and a lot of training. Examples are loaded into the system, lots of examples. The more examples they can load in, the better the detection algorithms become. But no matter how much training you provide a system, there will always be exceptions to the rules. And no matter how good those detection algorithms are, they will always make mistakes—because we humans also make mistakes.
Use these programs only with the understanding that they will get it wrong.
Most of the automated editing apps available are copyediting aids. They are designed to help you with your punctuation and grammar, and will point out when you have sentences that are perhaps a little too complex or sticky to read. The reports look at whether you have employed too many cliches or if your writing employs passive voice. But as I mentioned above, some so-called rules are begging to be broken.
I use ProWritingAid myself, and it's constantly pointing out phrases that it believes are written in passive voice. And almost every time, I just ignore the suggested changes.
In a recent piece of writing, ProWritingAid flagged the phrase her arms were tied as passive. Yes, the program was correct in saying that this was passive voice, because there is no mention of who or what is tying her arms (whoever her is). But when you look at the sentence as a whole, the passive construction is what is required.
The whole sentence: She went to hug herself to stave off the chill, but her arms were tied down.
As the writer, I needed to choose between having something that makes sense given the style of narrative being used, or something that is so-called active voice according to a program.
If I was to take on board the suggested edit from ProWritingAid without critical review, that sentence in my crime thriller would read: She went to hug herself to stave off the chill, but it tied her arms down.
*blink blink*
The suggested rework gives a horror feel to the sentence, with no real understanding of what it is. But that's not what I was going for. So, I ignored the suggestion and carried on.
Here's another example where the automated programs would have gone wrong.
Some time ago, a passage from a client came across my desk that described a character suddenly caught up in a tsunami. The writer had used a run-on sentence, something that had in the order of 60 words in it with no breaks. The effect was so powerful, because as the reader, I really did feel like I was tumbling with no idea of which way to turn—unable to breathe.
Any of the automated programs would have highlighted the sentence and told the writer to break it up and rework it. But that sentence was perfect in the context of the story.
(I would love to show you that sentence, but I can't, because of client confidentiality.)
Programs like ProWritingAid, Grammarly, and Hemingway have their uses. They are great for picking up the things I missed in my 100 read-throughs of the same passage: words that were the wrong word (a few letters jumbled up, an f when it should have been a t, or anything like that); punctuation that is not quite right; and phrases that have become repetitive without me knowing it. But the human eye is still needed to add that subjective element inherent in writing.
Developmental editing in particular is subjective.
Developmental editing encompasses things like story structure, character development, show vs tell, point of view, dialogue, and narrative voice. And that's just the beginning. Much of what I do as an editor doesn't exist in the realm of rules or guidelines. What works in one story doesn't necessarily work for another.
Overall plot structure is something that you can put numbers to, analyzing beat points and the rise and fall of tension. A breakdown of POV usage is something else that you can run statistics on. However, the determination of when a story has progressed into waaaay too much backstory is subjective. There are no concrete measures on this. It's a matter of what feels right.
You can use the algorithms found in programs like Fictionary to analyze chapter lengths, structural outlines, and consistency of point of view. But what about depth of perception, the balance between prose and dialogue… show vs tell?
Humans bring to their writing and editing accumulated life experiences. It's not something that we set out to obtain. It just happens.
Most of us have read countless books for pleasure, pondering their meaning. We've watched countless hours of movies or television, subconsciously (or consciously) analyzing the story structure and the pacing of scenes. And we have years of in-person interactions to understand how people speak in real settings—which includes body language.
This is all stuff that you can't program into a computer—not easily.
When developmental editing has fewer rules than copyediting…
Automated programs have their uses. I wouldn't be using ProWritingAid if I didn't believe that. But I will always default to the trusty eye of a developmental editor or book doctor any day. Even my critique partners and beta readers are going to pick up those inconsistencies within a scene faster than any program ever could, because they'll be able to visualize (or not) what I've written.
Use the automated editing programs with caution and know that they can never replace a human editor.