Marlowe, A.I. for Novels

by
This content has been archived. It may no longer be accurate or relevant.

From The Book Designer:

I recently had the opportunity to evaluate Marlowe, Authors A.I.’s analytical software for novels. Created by Matthew L. Jockers, Ph.D., and his data team, Marlowe is an artificial intelligence that serves to help authors improve their novel before sending it off for professional editing. The goal of this software is to “help authors refine their manuscripts and identify new market opportunities for their works.” (Note: I searched their website but did not find any reference to helping authors “identify new market opportunities for their works.”)

Marlowe is relatively new—first released in January 2020—and so I was a little skeptical about the reliability of the algorithms, fearing its creators could still be working out the kinks. Also new is the Authors A.I. organization itself, a June 2019 venture.

The manuscript I submitted to Marlowe was the final manuscript for my latest novel Nineteen Hundred Days, a book in the literary fiction genre that had gone through three levels of professional editing:

  • manuscript critique
  • line editing
  • copy editing

The 24-page report I received for this manuscript includes 15 areas of analyses.

Plot Structure

This section of the Marlowe report is about narrative arc and major turning points in the story line. Below is the visual representation provided for my novel.

The dotted line running across the graph indicates emotional neutrality.

The purple line represents conflict and conflict resolution. Upward slopes mark instances of conflict resolution where the story takes a positive turn, and downward slopes indicate the story taking a darker turn or some level of complication.

The more peaks and valleys, the more of an emotional roller coaster the character(s) is going on, which could be an indication of how successful the novel is in engaging readers. If the purple line doesn’t veer away much from the dotted line, the story is likely flat and uninteresting.

According to Marlowe, a good story will result in this line vacillating between above and below the dotted line with highs and lows throughout. And it’s not just about the number of spikes, it’s also about the depth of each one.

The green line represents the narrative arc. Marlowe claims there is no optimal shape for the narrative arc, and they are working on obtaining some comps from bestsellers for future reference. But my experience is that fictional stories are best structured if they include these narrative arc components:

  • exposition
  • conflict
  • rising action
  • climax
  • falling action
  • resolution

When optimally included in the story line, these components form a definitive narrative arc.

My interpretation of this analysis for Nineteen Hundred Days is that it has a typical narrative arc (at least according to my knowledge and experience). With respect to the level and depth of conflict in the story line, I can only compare it to the sample report that Authors A. I. has on its website for The Da Vinci Code which has five peaks (compared to my six) with approximately the same depth and occurrences above and below the dotted line. While my novel is not in the same league as The Da Vinci Code, I can only feel good about this comparison.

. . . .

Pacing

Marlowe analyzes the story’s pacing by plotting where it thinks readers will turn the pages more quickly (peaks on the graph) and the slower moments (valleys) where there is likely scene setting and background information given, claiming that the most successful writers vary the pace of their story to provide variety.

When I saw this analysis for my novel, I immediately wondered what was going on at the 57% mark to cause such a prominent “valley.” So I looked at chapters 20 through 25 and found three significant events:

  • the protagonist’s best friend dies of cancer
  • he is asked by police if he can identify someone they are looking for
  • he learns of his mother’s jail sentence

I would not call these “slow moments,” so I am at a loss as to why the graph dips so low at the 57% mark.

Link to the rest at The Book Designer

Click to Tweet/Email/Share This Post

10 thoughts on “Marlowe, A.I. for Novels”

  1. I got up this morning and started cascading a lecture on all aspects of writing, from Free Writing to start pages, all the way up through structure: Three Act Structure(Rule of Three, Rule of Four, The Paradigm, extended Paradigm), Fichtean Curve(Classic, Right-Hand Rule, Left-Hand Rule), Flashback(Simple, Cascading), etc…, then how to start the story small to see it as a whole, then uncompress the story to final size.

    There is one simple fact:

    – Story grows with the telling.

    Think of growing zucchini rather than going to the store to buy zucchini.

    Look at YouTube videos for growing zucchini to see what I mean. All those leaves and stalks are essential. Too many people want to start writing, “Once upon a time, blah, blah, blah, The End.” and complain when most of the pages they write can’t be bundled up, published as is, and make money, but I digress.

    This lecture takes me at about an hour as I work my way through the morning routine. This happens to me many times a week while I’m trying to wake up, so you can imagine my shock when I fire up TPV and there is an article about software to analyze structure.

    Yikes!

  2. I asked the OP author (on The Book Designer) how Marlowe compares to Grammarly. She said she’d never tried Grammarly, which I found interesting.

    Anyway… has anyone here tried both, and how do they compare?

      • I am the author of the Marlowe post on The Book Designer, and you are right–I am not an expert on artificial intelligence for critiquing novels (never claimed to be). But I do know a thing or two about what constitutes good fiction, and that is why I agreed to run one of my novels through the Marlowe software and provide feedback. I hope you found the assessment from an established author useful.

        • Thanks for coming over to explain, F.

          I enjoyed your article, which is why I excerpted it here.

          I would invite you to visit on a regular basis. I believe you will find a good number of other established authors also stop by to comment on a regular basis.

        • Thanks for jumping in here, Florence Osmund. I found your post very interesting, but since Marlowe seems to duplicate some of what Grammarly (which I already use) does, I was hoping for a comparison to a more-established AI. In any case, it’s all helpful to authors in the self-editing phase.

          • Thank you, Harald. It was an interesting experience for me since I have traditionally relied on my own self-editing and then an editor to catch my mistakes. AI can add a useful editing layer if the software is dependable.

              • Felix – I was asked to state the genre of my manuscript in the initial submission, but I do not know how they used this information, if at all, in their analysis.

Comments are closed.