Most people think getting better at something means collecting more.
More books.
More frameworks.
More prompts.
More systems.
More rules.
For a long time, I thought that too.
Then something strange happened.
The deeper I went into building systems alongside AI, the fewer ideas I actually needed.
Entire collections of principles started collapsing into one another.
Five rules became one.
Ten workflows became three.
Dozens of “best practices” turned out to be different expressions of the same underlying idea.
I started calling it The Great Compression.
Not because information disappeared.
Because understanding did.
Compression Is What Understanding Feels Like
When you’re new to a field, everything looks different.
Every situation appears unique.
Every problem deserves its own solution.
Experience changes that.
The longer you spend with something, whether it’s plants, businesses, software or people, the more your brain starts seeing patterns instead of events.
Chess grandmasters offer a fascinating parallel.
Research has shown they don’t outperform beginners because they memorize more individual positions. They outperform them because they compress complexity into recognizable patterns.
A novice sees thirty-two separate pieces. A grandmaster sees meaningful structures. Patterns.
Patterns that immediately suggest strengths, weaknesses, opportunities, and dangers.
They aren’t calculating every possible move from scratch; they’re recognizing entire clusters of relationships at once.
I think something similar happens as we mature in any complex domain.
We stop collecting isolated facts and begin recognizing stable patterns of behavior.
The real leap isn’t from ignorance to knowledge.
It’s from knowledge to chunking, from seeing hundreds of disconnected ideas to seeing a handful of deeply interconnected ones that reliably predict what happens next.
Spidey sense says that it happens as a step change as well. Not linear.
Working with plants taught me this long before AI did.
At first you notice individual characteristics.
This plant grows faster.
That one stretches more.
This one likes more water.
Another hates being overfed.
Eventually something shifts.
You stop seeing individual observations.
You start seeing relationships.
You realize the plant isn’t one characteristic.
it's not just its name, or how it makes you feel.
It’s an expression of interacting traits, characteristics, and predispositions.
Nothing exists in isolation, and the whole always matters more than the parts.
That observation never left me.
It followed me into every other system I’ve ever worked with.
Then AI Did Something Unexpected
I assumed becoming better at working with AI meant discovering more prompting techniques.
Instead, something almost backwards happened.
The prompts became less important.
The behaviors became more important.
Over months of building systems, I noticed certain behaviors repeatedly appeared together.
Not because I asked for them individually.
Because together they produced better outcomes.
Things like:
- Evidence Before Explanation.
- Reward Deletion.
- Force Hierarchy.
- Challenge Your Strongest Conclusion.
- Runtime Before Documentation.
- Stop When Evidence Ends.
Individually, they’re useful.
Together, they behave differently.
The combination consistently outperformed any individual principle.
That surprised me. It shouldn't have, but it did. 1+1 =3. Duh.
These bullets are the titles of independent governance level frameworks that don’t overlap in terms of doctrine or scope.
For being written independently, without ‘the other’ in mind, they certainly bump up against one another in reality often enough to shed evidence.
Behavioral Chemotypes
I don’t know if that’s the final name, however it’s the best working name I have today.
A chemotype isn’t defined by one characteristic.
It’s defined by the interaction of many things that all roll up to a single trait.
The whole is more than just the sum of it’s parts.
I think reasoning behaves similarly.
Not as isolated habit but stable clusters of behavior.
One habit doesn’t explain a great decision.
Neither do two.
But certain combinations seem to produce entirely different reasoning.
Not incrementally better.
Qualitatively different.
That’s the important distinction.
The Goal Isn’t More Habits
It’s tempting to build an endless checklist.
Always verify.
Always challenge assumptions.
Always compress.
Always delete.
Always execute.
Pretty soon you’ve created another system nobody can actually operate, including you (See “The Human API”)
The really interesting thing is that the best question isn’t:
“What habit(s) should I add?”
It’s:
“What combinations consistently produce better thinking?”
That’s a much harder question. It’s also a much more useful one.
Ideas Inspire. Behaviors Compound.
One realization from this journey has stayed with me.
Knowledge accumulates.
Wisdom compresses.
Behavior compounds.
Reading a great idea changes your afternoon.
Practicing better behaviors changes your next thousand decisions.
Hence, the general curiosity and fascination with the physical environment around us.
The XYZ. More fun to describe reality with an acronym.
More specifically, things we repeatedly see.
The objects we repeatedly touch.
The cues that quietly shape how we think without demanding our attention.
We all have them.
Something in our physical environment that helps bring us back to center.
An item in the room that we can glance at that reminds us of our next move.
A physical object which acts as a trigger to recalibrate towards true North
I’ve started calling them Real World Anchors.
Not because a coffee mug changes your life.
Because repeated exposure to a worthwhile principle changes the way you approach hundreds of tiny moments over time.
The object isn’t the product.
The principle, and the principle put into action (right now) is.
The object is simply the carrier.
The environment becomes the runtime.
Be the architect, not the API.
A Different Kind of Progress
Most people measure progress by accumulation.
More features.
Documents.
Meetings.
Frameworks.
More is better.
At a certain point though, frustration channels into a seek-and-destroy red team flurry:
What that we’ve built deserves to survive?
How many things can disappear without losing capability?
How many rules collapse into one?
How much complexity can be removed while preserving understanding?
That’s The Great Compression.
Not becoming simpler by knowing less.
Becoming simpler because you finally understand more.
Maybe that’s what maturity looks like. Life schools you as you get older. Hard.
Enough to pause before.
Before everything.
Not carrying more ideas around in your head.
Carrying fewer ideas that explain far more of the world.
Continue reading: It’s Easy to Get Lost in the Building