Lessons · The Honest Machine

How to Build a System That Tells the Truth.

Eight short lessons, each teaching one idea. Work through them and you can look at any build system — yours or someone else's — and ask the honest questions: where does change get checked? Who can't be trusted to grade themselves? Which claims were never verified? Then fix what you find. No prior coding assumed.

How to Take Them

Every lesson has four doors — pick the one that fits how you learn, and you'll reach the same place:

Builder — do the hands-on exercise first, read after.
Architect — see the whole model as one diagram, then the detail.
Translator — start from an everyday analogy and cross over to the real thing.
Validator — work the check-yourself gates at the bottom, one at a time.

Not sure which is yours? Take the 5-minute quiz → Nothing saved, nothing sent.

Take them in order — later lessons lean on earlier words. After each one, pass its gates, then come back to the flashcards two or three days later, when forgetting has had a chance to start. The tier ends with a Graduation Build: give one real thing you look after — a website, a budget, a classroom system — the same honest-machine treatment.

Tier One · The Honest Machine — The Principles

The Graduation Build

Reading isn't the tier. Pick one real thing you look after — a website, a budget, a curriculum folder, a classroom system — and give it the honest-machine treatment:

One official home (Lesson 1): a single home with history, and a stated way changes get in.
Honesty audit (Lesson 2): find one place it performs confidence without evidence. Rebuild that verdict as a flag with its proof attached.
One named law (Lesson 3): write the anti-false-alarm rule for its most trust-critical check, with one tricky case it must survive.
Enforcement (Lesson 4): move one rule from "I remember to" into something that runs without you.
Witness run (Lesson 5): define agreement in advance, then check your thing against one independent outside reference. Report the result — no explaining it away.
Layer sort (Lesson 6): name what in your system is habitat, project, mesh, and harness. Find the hub.
Claim audit (Lesson 7): find your oldest unverified safety claim and run the check today. Date it.
Queue (Lesson 8): stand up your reflection queue with three real pins and a drain date on the calendar.

The deliverable is a one-page write-up: what you found, what you fixed, and the one flag you'd show a stranger as proof your system now tells the truth. Present it to someone who will ask hard questions.

Tier Two · Building the Machine — The Hands-On Track

Once the principles are yours, this tier gets concrete: actually wiring an AI to do real work on your projects — safely. It's more technical than tier one and builds on it, so take tier one first.

The n8n Track · The Tool the Principles Were Learned On

Tiers one and two are tool-agnostic on purpose, and they stay the front door. This track is the specific tool I learned it on — labelled, not hidden, because stripping the tool out removes the evidence the principle was ever learned rather than asserted. Every page here names the universal invariant underneath it, so the tool is the vehicle and the invariant is what you keep. You do not need n8n to read them.

More of this track is queued and not yet written — Glossary II, and an advanced tier drawn from the same notes. They arrive here as they're written and reviewed, not before.