1500ish words today. Today was a fun data day.
I created the bones of my editing engine. I was pleasantly surprised to find out that someone had already built one of the tools that I needed. And it was free. So I was able to integrate that into the engine with pretty remarkable success.
I’ll keep this simple, but let me know in the comments if this is hard to follow so I can tweak my messaging moving forward because this will very, very helpful to many of you and it will save you editing costs and catch typos.
BACKGROUND, IN PLAIN ENGLISH
Prior to today, when I sent my book to an editor, they made recommendations.
- Some edits were spelling/grammar errors
- Some edits were story issues.
When the editor makes spelling/grammar edits, I accept them and go on with my life. I TRY to remember some of the lessons the editor taught me, but I don’t always catch them when I’m writing the next book. Therefore, my editor has to correct the same errors again, which frankly is not a good use of her time. Also, having to correct the same errors over and over again may make her miss other errors nearby that are hiding in plain sight.
The question: how can I catch as many of the editor’s recommendations BEFORE my next book goes to her again so that she won’t have to make the same edits twice? Put another way, how can I eliminate errors that I’m certain the editor will catch?
PREP WORK
I keep all of my editing documents, so I reviewed my last 5 novels and categorized the edits. I used macros to count how many edits I received, which gave me:
- The average number of total edits (spelling+story). My average is 290 edits per novel.
- Approximate number of total edits per chapter. My average is 8.5 edits per chapter (based on 50K).
- This is approximately 1 edit per 177 words for an average of novels that are 50,000 words. Ideally, you want this number to be as high as possible.
And yes, I’m sharing these numbers out of transparency. No idea how I stack up to other writers, but you can certainly compare your numbers to mine.
Anyway, these numbers became my baseline.
CREATING THE ENGINE WORKFLOW
I used a mix of Word macros (some created by the amazing Paul Beverly) to identify particular edits that my editor(s) caught in the last 5 novels. Word macros are limited in what they can catch, but they can catch a lot more than you think. And, they can catch things that Word’s spell checker cannot.
Specifically, you can teach a macro to catch:
- Wrong word/phrase choice
- Improper Hyphenation
- Spelling/grammar errors
- Other usage issues
I am also working with a developer to build a separate macro that catches frequently repeated words on the same page. It cycles through a page and highlights words/phrases that appear within a 100-word radius, such as the word “scream” being used four times in one page, for example—something I did once. It can also exclude the most common words in the English language, honing in only the words that count.
TESTING THE ENGINE
Anyway, with all of these elements together, I started teaching the engine what to look for and testing it. This took about 2 hours.
Then I ran my most recent manuscript through my workflow to test it with a real novel.
THE NUMBERS
Hopefully all this math checks out. If not, I know y’all will let me know. LOL
In 2 minutes…
The workflow caught 178 errors.
I accepted 166 of those errors. Sometimes the engine isn’t always right. Still 166/178 is a pretty good success rate.
(The errors that I accepted were all, without exception, items that my editor would have corrected. This isn’t theory. It’s practical.)
Since Word duplicates its numbers when it counts tracked changes, I divided this number by half to get the closest approximate number of “actual” edits, which was 83. Just trust me on this number.
The engine also caught 7 additional duplicate words that did not show up as tracked changes.
The engine also caught 7 errors that Word likely would have captured if I had spell-checked the document prior. I did not, mainly because I was curious how much Word’s spell-checker would contribute to the errors caught.
I also caught 2 errors by eye as I was glancing through the manuscript.
That brings the total number of errors to 99.
99 errors / 38 chapters = approx 2.36 errors per chapter
Word caught 7 out of 99 errors, which is 7% of the overall total. So if you ever wanted to quantify how much Word actually contributes to a book’s manuscript quality, there you go.
PRETEND WITH ME FOR A MOMENT
If my editor usually catches an average of 290 total edits per novel, and I captured 99 of those edits before I sent my novel to her, that means the engine probably reduced my total edits by 34%.
If my editor usually catches an average of 8.5 edits per chapter, and the engine catches an average of 2.36 (spelling and grammar) edits per chapter, I reduced the number of edits per chapter by 25%.
The reduction numbers get even better when you consider that the numbers I tracked include both spelling and story errors. If you look at just the spelling/grammar edits by themselves (which I didn’t do due to time), it’s an even more drastic reduction.
Also, the repeated word part of the workflow isn’t turned on yet, so that will account for more errors caught next time.
And Grammarly and ProWritingAid both would each catch somewhere between 6-12 errors if I had used them. So I could have been looking at 120-130 actual errors caught, and that’s on the conservative side.
The more errors I take off my editor’s plate, the more she focus on other things that may be hiding in plain sight. In other words, the more errors I find ahead of time, the more errors she can find after the fact because they’ll “stick out” more.
As you can see, this starts adding up in a very big way. All with free tools, just a little bit of programming costs (my choice), and an understanding how to use the full horsepower of the technology that’s already on your computer.
AND THAT’S NOT ALL
I can capture all of the edits and do additional analysis, too.
In the next phase of the build, I should be able to drill down to the chapter level and see a breakdown of spelling vs. story edits.
I developed a “chapter scoring engine” to grade how I did and where my editor spent most of her time. Based on a few parameters, I can score each chapter to determine which ones caused her the most trouble for both spelling and grammar and story. Again, this will all be automated—no data entry.
Might it be true that chapters I dictated with Dragon Anywhere drove the most errors? Or, does writer’s block have an adverse impact on the number of errors I create? Do longer chapters have more errors than shorter ones? What about chapters told by a certain POV character?
I tracked all this stuff while I wrote. So when I run all of it through the scoring, I will get a chapter ranking. I can dig into the worst-ranked chapters to figure out what the drivers of the errors were. I have no idea what to expect so that will be an interesting experiment.
With the next book, I can start looking for drivers of errors ahead of time, and the scoring may help me further predict where the most errors will appear.
Again, we’ll see. This is just a pilot and it’s a little messy right now, honestly. There are a lot of bugs to work out with the chapter scoring, and it may not even work. Sometimes ideas fail. But if this does work, it will open lots of interesting doors for collaboration with my editor.
If the scoring is accurate, I can score novels ahead of time and flag the potentially problematic chapters proactively so my editor can spend more time on them.
All of this took me the better part of a weekend, so you don’t have to spend very much time to gain insights into your editing skills.
ANYWAY, A PRETTY PRODUCTIVE DAY
I continued working on Indie Author Confidential, Vol. 4. It’s coming along just fine. I’m posting a little early today as I have some studying to do since law classes start tomorrow.
Have a good night.
