LATEST ARCHWINGER BOOK UPDATE - The editing process is hard!

a.k.a. kill your darlings

A lot of people out there — more than you’d think — write books using AI.

AI actually sucks at writing, but not for the reasons most people assume. It’s shockingly good at English. If English isn’t your first language and you haven’t been steeped in it for decades, an LLM can probably produce more natural-sounding prose than you. The book itself might not be better, but the English will feel more “normal” to a native speaker.

If I fed an LLM the entire manuscript of my book and asked it to write the next chapter in my exact style, telling it exactly what needs to happen, it would nail my voice so well that you guys—and even my wife—would be completely fooled. I’d spot the handful of non-Archwinger tells, but you probably wouldn’t.

The problem isn’t style or creativity. AI does English just fine. The problem is content. Tell it to write a chapter and the things it chooses to include feel weird and random. It can’t land a joke, its metaphors are… off, and the points it makes are like when it draws you a picture of a girl with three thumbs on her left hand. It just reads janky and wrong.

One area where I’ve found AI to be really helpful is line editing. AI is really good at spotting typos, clear grammatical errors, and catching inconsistencies like when you say one fact in chapter 3 of your book then contradict yourself later in chapter 8.

This comes with a huge asterisk, though. AI still sucks at writing. It’s just really, really good at spotting issues. So if I give it a chapter, it can tell me dozen things that read unclear or fall flat. But 90% of its suggested changes are stupid, and the other 10% aren’t insightful, just what I would have done anyway. However, even if I don’t use its suggestions, it’s really useful for spotting places I could have written something better and getting me to think about whether I want to change anything.

This turned out to be a helpful tool in the writing process because when I sent my finished draft to a real human editor, the language was already pretty good. People tell me I’m a good writer already. So polish and sand a few more rough edges, and the first draft was darn fine, English-wise. That let the human editor focus his notes on actual structural issues rather than language ones.

That’s where I’m at in the process currently.

• A little about the book:

The current working title is: “Wait. What Was the Question?”

It’s written in an autofiction/confessional/memoir-style. It’s not actually autobiographical, but they say to write what you know, so the main character shares a lot of traits with me so I can write him pretty darn confidently, and a lot of the book events are loosely inspired by my life or the lives of people I know. But what actually happens in the book is pretty much nothing like my life. I’m not going to do this trick again though, because everyone just thinks it’s autobiographical anyway.

The biggest issue with this style is there is a LOT of internal monologue. That’s intentional. Not lazy writing. A lot of authors write internal monologue and exposition as a way to tell the reader things they’re not smart enough to implicitly tell through dialogue and action. In this book, the internal monologue isn’t an omniscient spirit giving the reader secret knowledge. He’s more like a third character in the room. And he’s wrong 90% of the time — that’s the fun part. When the guy’s takes are completely different from what the reader is actually reading, you learn more about the character and the story than anything the acutal page is telling you.

The main character isn’t on a hero’s journey. He doesn’t overcome adversity, resolve conflict, or achieve victory. He just grows chapter by chapter until by the end, he’s a slightly more honest man and the gap between what you’re reading and what he’s telling you is smaller. But there’s still a fun, heartfelt story taking place in the background.

• The structural issues:

Most chapters follow the same pattern: the main character is talking to someone in the present. He finds an excuse to launch into a story from his past, an anecdote, or some annoying thing he wants to monologue about. While that scene is happening, his internal thoughts are firing—but they’re not always about the scene in front of him. Sometimes he’s talking directly to you, the reader.

Picture an episode of The Wonder Years, or if you’re too young to know what that is, picture a YouTube reaction video. There’s a line of dialogue. Then the recording is paused and the narrator talks to you for anywhere from one snarky remark to up to two or three paragraphs about something related. Then it’s unpaused and the next line of dialogue happens.

This is a pretty big structural issue because it requires reader attention and for the reader to be pretty darn invested in the book. It’s a lot of mental work for the reader to be trying to follow the scene on the page, but there’s a three-paragraph sidebar before the next line of the conversation.

That said, it’s not always important. In many parts of the book, the internal anecdotes, stories, and monologues ARE the story. The scene on the page isn’t nearly as big of a deal and is the secondary plotline.

It’s not actually hard to understand what’s going on. It’s just a bit of a disjointed reading experience, which many readers will not like. If you’re really enjoying the story, or if you’re a huge fan of me for some dumb reason and insist on pushing through, it’s a good book. But the structure will turn off a lot of casual readers.

• Fixing stuff:

So I started going through, chapter by chapter, and reorganizing things a little bit. Breaking up the dialogue less. Grouping scenes and beats together better, then putting any related exposition or internal monologue before or after. This let me trim a bunch of redundancies and streamline things. But it wasn’t reading well. Something felt clunky every time I went back through and reread the first few chapters.

So I fed them into AI to see if there were any weird spots I wasn’t catching that I should consider editing. I even asked it “which version of this chapter is better” then pasted the original version and the edited version — without telling it which was newer. It said (I’m paraphrasing) “The first version. And it’s not even close. The voice is sharper, the ideas flow better, though the pacing drags in all of the internal monologue parts and you should trim those by at least 20% if not more. Frankly, the second version sounds very amateur and reads like a first draft you refined to get the first version. Is that what happened?”

So I tell AI it’s actually the opposite. I was concerned that breaking up the scene-on-page dialogue with other stuff was a big readability issue. AI said something like “Yes, that’s a valid concern and your largest structural issue because I’m AI and I always agree with whatever you tell me. But instead of fixing the problem, you made it worse. Instead of having small sections of scene interleaved with internal monologue, you now have big sections of scene side by side with competing big chunks of internal monologue. That’s not the way to fix this. The fix is rewriting the book so the internal thoughts are short and only react to what’s happening on the page instead of doing YouTube-reaction-video commentary to the reader.”

That would help the readability problem, but it would also change the entire voice and feel of the book. And the voice is the book. So I’m reverting to the draft I sent the editor. I’m going to step away for a week or two, do my day job, work on the next book, and come back to this one with fresh eyes. Then I’ll do a lighter, more honest pass, still based on his advice—reading it like a casual reader and fixing what actually feels hard to follow. Rather than trying to structurally reorganize the whole thing.

If it stays a little janky in places, it’s my first book. It’ll be fine.

• Next steps:

So one more read-through in about a week to address the issues noted above. Then probably one more clean-up pass to catch what I mess up when I make changes the first time.

The day job is kicking my ass again, and we’re making fairly regular trips to Houston to see my dad, whose health isn’t great. So progress may be slower than I’d like.

Per my last update post, I’ll still probably ping a few literary agents who have worked on similar types of books and see what kind of (non) traction I get. I want to get a feel for what that process is like and the kinds of bullshit people throw at you. (The most common line is “I’d be interested in seeing your next book” which is code for “you write okay, maybe even well, but this shit you sent me sucks and won’t sell. In fact, you probably shouldn’t even publish it, or if you do, use a pen name, because having this non-selling title tied to your identity will hurt any future approaches you make to publishers.”)

The book’s actually pretty good. It just doesn’t slot neatly into any marketable category, and nobody wants a voice/character-driven confessional debut from a total unknown. Those are dicey even for established authors. So I’m realistic: this is probably heading to Amazon, which is fine.

• The future!

I’m about 10k words into my next book. It’s a bit of a modern riff on Kafka’s The Metamorphosis, which is one of my favorite reads of all time. I spent a lot of time between wives at my favorite wine bar reading. Any time a cougar bothered me, she’d usually ask what I was reading. If you want to get rid of girls fast, start telling them about your favorite book, Metamorphosis.

Kafka’s novella is short. Maybe 25k words. Worth the read. The original is in German, so if you read a translated copy, you’re not reading it for the prose and writing style. Just the story. It follows the tale of Gregor Samsa, some dude who wakes up one morning and has turned into some kind of large beetle-like creature. His first thought is “omg I’m going to be late to work”.

It turns out he’s the family workhorse and despite doing this for ages and never having taken a sick day for 15 years, he has built up zero goodwill with his family or with his employer, and the very second he’s a burden, everyone hates him. You can probably imagine why I like the book.

I’d like my next book to be full-sized, rather than novella-length, but it’s looking like the story will be told by about 20k words. There’s a secondary story arc I want to add and interleave, though, that I think will be fun. But I still have to get this first book published.

• Conclusion

I’m still hoping to have my first book published June/July of this year. Unless something goes horribly wrong. Or horribly right and some big publisher wants to give me a huge advance to go through their process while I spend their money buying a condo in Costa Rica.

Thanks for riding along. Uh... buy, like, subscribe, whatever social media things there are to make me look big and popular. Do those things.