Writing software: ProWritingAid

A couple of weeks ago I savaged AutoCrit.  I was probably too harsh.  It is a useful tool, despite the limitations I found.  I still believe, though, that $30/month is too hefty a price for how much good it would do me.  What if you could have the usefulness of an algorithmic editing tool, plus Scrivener compatibility, for $60/year (or less with a coupon code)?  That sounds better to me, and that’s what I found in ProWritingAid.  I have been using it for the last week, and it’s been worthwhile.  This morning I bought a lifetime membership, which costs the same as three-and-a-half years at the yearly rate.  I may not use it that long if artificial intelligence takes major leaps forward and ProWritingAid ends up lagging behind other tools, but it has impressed me enough that I am willing to take a chance.

ProWritingAid has over twenty different reports.  I am finding the most useful to be spelling and grammar (it has caught a handful of errors that I hadn’t noticed on multiple read-throughs) checks of overused words, sentence structure and length, and style suggestions such as cutting adverbs and hidden verbs.  One of my biggest challenges is to make my writing more concise.  The myriad things ProWritingAid can flag help me reassess my writing and look for shorter, better ways to say the same thing.  It doesn’t fix my writing; it spurs me to do it.

One report that I am a bit suspicious of is the “Sticky Sentences” report.  It claims that writing that is too full of the 200 most common words is like glue; it slows the reader.  I have, however, tried this out to flag sentences, and have actually found it to be helpful as it pushes me to use more active, meaningful, or precise vocabulary.

Of course, ProWritingAid is still an algorithmic tool, and as such, it is woefully dumb.  Almost all of the grammar issues it highlights are not errors at all.  In that sense, my disappointment in AutoCrit applies equally to ProWritingAid.  It takes patience to sift through all of the non-mistakes in search of the few actual errors in the text.  At least ProWritingAid is not charging me much for all the trouble.  

In the end, ProWritingAid has one more significant advantage: it can open and save files in Scrivener format.  That convenience makes all the difference.

I would never trust the scores that ProWritingAid gives as actual measures of the quality of my writing.  Writing is art, and computers will never understand how words can move you.  Human editors are a must.  An algorithmic editing tool, however, can help you improve your writing before you send it to a real editor.

Writing software: AutoCrit

National Novel Writing Month (NaNoWriMo) concluded a month and a half ago and I have not written a blog post since.  I guess that level of daily output burnt me out.  I finished the month with 75,100 words.  Added to the 20,000 I had before NaNoWriMo started, the novel is over 95,000 words. It needs a major edit, including some fundamental changes.  All of that will come later, though.  I am letting it sit for now while I work on other things.  Chief among those is editing Neyuki, trying to cut it by 20 percent before it goes to my editor in early March.  In December I cut it from 145,000 words to 134,500.  That leaves me 18,500 words shy of my goal.  This may simply be a goal I don’t reach, although I am trying hard to cut at least another 10,000 words.

As I looked at the various sponsors of NaNoWriMo, one that intrigued me was AutoCrit.  They offer an online editing program on a subscription basis ($30/month for the Professional plan, $80/month for the Elite plan).  This week I signed up, since they offer a two-week trial for just $1.  Knowing how lame online translators are, I was skeptical that an online, algorithm-based editor could be worth the money that AutoCrit is charging.  That skepticism was well founded.  While AutoCrit’s algorithms seem to be surprisingly sophisticated, they cannot begin to match the usefulness of a human editor.  I think that might be true for even a poor human editor—and I have good ones.  I have written about them before.  The first one, my wife, is quite accomplished and does editing part time in addition to her day job as a teacher.  The second, Fran Lebowitz, whom I found on Reedsy, is a professional with truly enviable client comments.  (One of those is from me.)  Compared to either of them, AutoCrit fails miserably.

I am not saying that AutoCrit is not useful.  It told me that I start too many sentences with “and” and “but,” which is probably true.  It also says that I overuse the words “just” and “even.”  But it also told me about useless things, such as my use of clichés.  For example, it said I use “on the table” too much.  How else do you say, “She put her purse on the table”?  That’s where algorithms fail.  Language is far too complex for today’s algorithms to effectively understand or classify.  Where AutoCrit fails, though, is not just its identification of writing problems using too-simple algorithms, but then using those results to give writing an overall score.  Under Shōko’s Bed scored 74.9.  Being a professor, I can’t help but see that in letter grade terms, and I refuse to believe that my novel rates nothing more than a C.  AutoCrit needs to give up on the meta-analysis and just stick to the stuff where it’s actually useful, like counting sentences that start with “and” and “but.”  Unfortunately, the simple stuff isn’t worth $30/month.

In conclusion, don’t spend $30/month for AutoCrit.  Spend $1 instead and try it out.  Do it because it will make you all the more thankful that there are talented human beings in this world who are actually willing to edit another person’s work.  Then find one of those people.  You won’t be sorry.