The Split

Trimmed to bullet-point summary 2026-05-09 — original prose archived at _archive/blog-articles-pre-trim-2026-05-09/the-split-high-agency-or-irrelevant.md. Pending rewrite in voice.

Two groups: tiny cadre of operators running 5-7 terminals in parallel + large majority opening ChatGPT once a month and concluding “AI hallucinates.”
Miessler estimate: fluent group is 1-10% of workforce.
Gap is agency, not IQ or resources.

The Bifurcation In Numbers

Miessler UL 487: runs Claude Code across multiple terminals “like a CEO handing work to 100 workers across 100 companies.”
Use cases: image migration off S3, normalizing two decades of blog posts, fixing thousands of broken links.
Pays $200/mo for Max; would be cheap at $2,000 — recouped 5x in a single week.
Miessler “Why AI Will Replace Knowledge Workers” episode: median knowledge worker producing ~23% of a real week’s work; rest = meetings, Jira theater, cube-farm politics.
Bar to beat: $50T global payroll.

High Agency Without the Cringe

“High agency” framing in circulation: George Mack, Chris Williamson, Modern Wisdom circuit.
Failure mode: NPC-framing + Bugatti thumbnails — identity worn for superiority.
Miessler UL 482 reframe (load-bearing): High agency is a tool, not a value. Values are what you want humanity to be. Tools are how you get there. Capitalism, socialism, high agency, stoicism — sliders to dial up/down per service to values.
Miessler’s values: “Star Trek liberal” — elevate humanity, equal opportunity, humans foreground.
Author shares the shape, not the identity politics.

High Agency Is Learnable

Load-bearing claim: not a personality trait. A set of habits that compound.
Author’s path: electrical engineer → decentralized infrastructure → orchestrating Isidore (49 skills, 33 hooks) on PAI.
What changed = sequence of small choices. Open the tool instead of dismissing. Ship broken instead of polishing forever. Read the post. Build the pattern. Write the skill file. Delete what doesn’t work. Repeat tomorrow.
Miessler’s three filters for surviving 2026 (“Starting 2026” piece): drive, creativity, AI tooling — in that order.
Drive is learnable: enough good input → system starts wanting to output something.
Miessler’s cheat code: read 100-1000 great books. Author’s: extract wisdom from every UL episode and let it recompose.

The Moat That Isn’t There

Most dangerous story: “AI can’t replace expertise.”
Miessler (“Great Transition” + knowledge-workers episode): expertise = knowledge + understanding + intelligence + creativity.
- Knowledge → AI has read every book. Check.
- Understanding → cross-references Soviet economics with Idaho chicken farming. Check.
- Intelligence → modern agent systems navigate shifting requirements. Check.
- Creativity → moat draining via Anthropic Skills. Expertise packaged as markdown files leaking from Cliff/Ravi/Suzy into open source. Miessler’s term: “peeing in the pool.”
“I know things others don’t” career plan = update.

Practical Moves

Pick an assistant and name it. Cognitive shift from “I use a tool” to “I orchestrate an entity with persistent memory.”
Build the scaffolding, not the prompt. Miessler UL 482: “Scaffolding matters more than the models. Context matters more than the models.”
Define the canonical shape, then delegate enforcement. Miessler normalized 20 years of blog posts by writing the canonical shape once, then pointing Claude Code at broken posts. Generalizable: every manual cleanup in your life has a canonical form describable in a paragraph.
Refuse both elitism and apathy. Miessler’s third rail: build, teach, share the tools.

Honest Part

Nobody stays on high-agency side forever. Bar moves; people moving it are also being replaced by next wave.
Question is not “am I safe?” but “am I still learning fast enough that the gap between me and the frontier isn’t growing?”
The loop: read → extract wisdom → feed back into the system → ship something small.