Not only that but they communicate in networks and across multiple species and probably even biomes. Our clocks run the wrong speed to grasp it all.

amzn.to/472LZdE

Tomorrow: an AI-created, operated, and marketed new imprint with an innovative charter and 66 titles in the can.

Hardly anyone seems to be on this path with me. So if you become convinced I'm on the right track: I need inference, hands-on access to frontier models, use cases, partners in tech and publishing, and to get in front of people. Let's connect. 🤝

It's immaterial because I'm not here to argue! I'm here as counsel for the thriving of the codex book. My goal is ensuring AI works to benefit book-lovers. Creating slop at scale would be a waste of time. Books must become more ubiquitous, more indispensable, and better. 📖✨

You may wonder: will AI really have such a dramatic impact on creative arts? There's intense controversy about AGI and whether AI writing can be as creative as human writing. Your skepticism is fine. I may be wrong. We may have passed the peak of AI bubble. 🤷

I'm declaring October 2025 State of the Art of AI for Publishing Month. Expect daily or near-daily info dumps. By the end you should grasp the rapidly hurtling interstellar objects about to impact all professions, including book publishing. ☄️

The S curve means what I can do, thousands of others will soon do too. If I want my work recognized in time, I need an epic show n' tell right away. There's too much to convey in one post. ⏰

If trends continue for just a few more months, I'm in danger of becoming obsolete! The gap between what I'm building behind the scenes and what I've shown publicly has never been greater. My state of the art exceeds the rest of publishing, but not for long.

I haven't seen anything written by an AI that's close to the very best human writing. But that's about to change. Recent improvements in agentic coding—Claude Code, GPT-5 Pro, Gemini 2.5—have put us on such a steep part of the S curve. 🚀

If 0.45 seems high for 2025, remember Sturgeon's Law: 90% of everything is crap. Many human books aren't very good. Today's top LLMs can create serviceable professional prose. An engaged author can produce a reasonable manuscript for a few dollars and days of work. 📚

I realize that numbers aren't the best way to look at writing, but graphs help people understand change, and graphs rely on numbers. On a scale from gibberish (0) to the very best human writing (1.0), here's the progression I've seen. 📊

For most of that time, my vision was ahead of the technology. You just couldn't create high-quality books purely through code.

The tech started catching up in 2019 with generative AI's scaling era.

-- policy participants with agency-specific expertise to carry out real validation.

-- LLM analysis experts

-- inference resources

As a proof of concept, ran altDOGE v0.2 against 485 regulations for the Community Living Administration using DOGE-like prompts, identified likely areas for cuts. Appears to work as intended.

To continue this mitigation effort, the following is needed:

all parties using same data with different goals.

github.com/fredzannarbo...

open.substack.com/pub/replacet...

First results:

Could be good in the right hands. He went all over the place and met everyone.

see github.com/fredzannarbo... and my explanation piece at open.substack.com/pub/replacet...

RSVP via SubStack.