First principles and semantic tree learning, the operating system I actually use
For years I used these two ideas as if they were different disciplines.
First principles thinking was how I questioned premises before a decision. Semantic tree learning was how I organized what I was reading for the long term. Two tools, two use cases. I had written about each of them and thought about each of them separately.
A few years into running multiple ventures, I noticed they were actually the same practice. You cannot do one well without the other. This post is the synthesis, which I think is more useful than either piece in isolation. If you are familiar with the two ideas already, this is where the real operating leverage lives.
Two halves of one practice
First principles thinking is the method. Strip a problem to what is actually true. Rebuild only from verified premises. Accept nothing on the authority of repetition.
Semantic tree learning is the architecture. Build a tree of knowledge with a trunk of fundamentals, branches of applications, and leaves of details. Let the trunk anchor everything you learn. Skim the leaves. Read the trunk slowly.
Notice what happens if you only use one.
First principles without an architecture is a perpetual questioning machine. You interrogate premises all day and never consolidate what you have learned into something durable. You are always ready to challenge. You never accumulate. Eventually the questioning becomes its own goal, disconnected from any practice it could improve.
Semantic tree without first principles is a beautifully organized library of conventional wisdom. You have trunks and branches neatly arranged. You have also accepted the conventional trunks of every field you have studied, any of which might be wrong in ways you never check. Your tree is internally consistent and externally mistaken.
Either failure mode is familiar. The first is the person who reads everything, doubts everything, and produces no operating result. The second is the credentialed expert who cannot see around the premises of their own field.
The combined practice avoids both.
What the combined practice actually looks like
When I enter a new category, the first question is not “what do I need to learn” but “what is the trunk of this field.” That is a semantic tree question.
Once I have a candidate trunk, the next question is not “what does this trunk say” but “is this trunk actually true.” That is a first principles question.
I interrogate the trunk from first principles before I commit to learning on top of it. If the trunk survives scrutiny, I grow branches from it and pick up leaves as they become useful. If the trunk fails, I find a different trunk and repeat.
The result is a set of trees where each trunk has been personally verified rather than inherited on faith. The trees are fewer and shallower than a specialist’s. They are also more robust because the foundation is something I actually believe, not something I have been told.
An example from aviation
When I started working in private aviation, the conventional trunk of the field was something like this. Aviation is a capital intensive business where the primary economic unit is the aircraft, and the primary economic question is how much it costs to put a specific aircraft in the sky for a given hour.
That trunk produces a specific set of branches. Ownership models. Fractional programs. Charter economics. Utilization analysis. Maintenance schedules. Crew certification. It is the trunk of every textbook and every industry briefing.
First principles questioning of that trunk: is the primary economic unit actually the aircraft, or is it the flight? An aircraft is an inventory item. A flight is the unit of value to a passenger. Aircraft based economics is producer thinking. Flight based economics is consumer thinking. The conventional trunk is built from the producer’s side of the table, and it stops seeing the market the moment the market starts pricing itself from the consumer side.
Once I pushed on that premise, several things became visible that were not visible inside the conventional trunk.
Empty leg flights are not a quirk. They are a structural consequence of pricing inventory by aircraft hour instead of pricing service by flight. The industry’s “surplus inventory problem” is an economics framing problem.
Brokerage models are not a compromise. They are a more honest representation of the unit of value. You do not own a specific aircraft. You sell a specific flight, and you source the aircraft that makes the flight economics work.
On demand charter is not an edge case. It is the natural conclusion of flight unit economics applied systematically.
None of this was invisible to the industry. Plenty of operators knew all of it. What was different was that inside the conventional trunk these ideas were peripheral. Inside a re examined trunk they moved to the center. That re framing is what made Vanbert a different kind of operator than a pure ownership play. More on the specifics here.
An example from scent marketing
The conventional trunk of scent marketing is that it is a discretionary luxury amenity, adjacent to interior design, sold as a premium service to properties that have already optimized the rest of their experience.
That trunk produces branches like: which hotels are candidates, which design firms to partner with, which fragrance houses to buy from.
First principles questioning: is scent marketing actually a luxury amenity, or is it an ambient memory encoding mechanism that any commercial environment can use?
The neuroscience is the same for a luxury hotel and a mid range supermarket. The olfactory bulb does not know the room rate. The question of where scent marketing produces real value is not determined by luxury positioning. It is determined by dwell time, attention context, and the value of memory for the category.
Once the trunk shifted, new branches appeared. Healthcare waiting rooms. Retail where customers linger. Hospitality adjacent spaces like airport FBOs. Categories previously treated as peripheral became strategic. More here.
The pattern is consistent. A conventional trunk was imported from one category and constrained where the practice could be applied. First principles questioning of that trunk opened the aperture.
How I apply the combined practice, in practice
Four working habits.
One. I do not accept a field’s trunk on authority. Before I invest time into learning a field’s branches, I spend a weekend or a week questioning the trunk. Often the trunk survives and I move on. Occasionally it does not, and the entry point to the field changes.
Two. I re examine trunks periodically. A trunk that was true ten years ago may not be true today. The schedule of re examination depends on how fast the field is moving. For aviation, about every two years. For scent marketing, every five. For capital allocation, on an ongoing basis.
Three. I separate leaves from trunks deliberately when I read. A news article is a leaf. A quarterly report is a leaf. A twenty year old book on a durable topic is more likely to be trunk material. I budget my reading time inversely to the time horizon of the material: leaves get minutes, trunks get days.
Four. I am comfortable with small trees. For a category outside my core work, a shallow tree rooted in a verified trunk is more useful than a deep tree rooted in an unexamined trunk. The shallow tree lets me make decent operator level decisions. The deep tree on a bad trunk can lead me systematically wrong across hundreds of decisions.
Why this matters for operators specifically
Founders and specialists can afford to go deep in one field. They have the time. Their success depends on depth.
Operators running across categories cannot. Their success depends on making decent decisions across several fields without becoming an expert in any single one. For that, the combined practice is the operating system. First principles keeps the trunks honest. Semantic tree keeps the learning efficient. Together they let you enter a new category at operator speed, which is the speed of making good decisions with incomplete information.
The compounding effect
After ten years of applying the combined practice, something unexpected happens.
The trees you have grown start to connect. A verified trunk in one field turns out to share structure with a verified trunk in another. Insights about attention economics from media work show up in scent marketing. Insights about allocation discipline from capital work show up in operating a portfolio of ventures. The trees become a forest, and the forest is navigable in ways a collection of separate trees would not be.
That is the compounding. It only happens if the trunks are honest. It is why the slow work of first principles on the foundation pays back over a long time. You cannot connect trees that rest on inherited wisdom, because the inheritance is specific to each field. You can connect trees that rest on something true, because the truth is portable.
That is the learning practice I have ended up relying on most. If you want the two pieces separately, the first principles and semantic tree posts go deeper on each. The synthesis, if you have stayed this long, is here.
Further reading: First principles thinking · Semantic tree learning · Contrarian thinking from the Caribbean