The most damaging thing about reading one wealth book is that it sounds settled. Read five and you discover the field is not a body of knowledge but an argument, and that the people at the top of it flatly contradict each other on load-bearing questions. That is not a weakness in the field — it is the most useful thing about it, because each contradiction has a condition attached, and finding your condition is how you pick a road.
One · Raise the price, or cut it?
Restrict supply and raise price. Never discount, never run an outlet, own every shop it is sold in. Margin is the business, and scarcity is what protects it.
Cut price relentlessly and flood supply. Give the margin away to buy volume, then own the layer everyone must pass through. Scale is the business, and margin is the bait.
Both built giants, so price is not the variable — the moat is. If what you sell derives value from meaning (who owns it, what it says, how hard it is to get), scarcity is your protection and discounting destroys it permanently. If it derives value from function, scale economics protect you and any margin you hold onto is an invitation to a competitor. The fatal position is the middle: a functional product priced as though it were a meaningful one. Pick a side deliberately, then behave consistently for years.
Two · Never sell, or sell well?
Selling is the skill. "Any time you don't sell, you buy." Reprice your holdings continuously and sell when the market will pay more than the asset is worth to you.
Never sell. One dividend in sixty years; thirty-three years without a liquidity event. Selling interrupts the compounding and triggers the tax, and the exponent is where the money is.
The split is clean, and it turns on one question: does your asset compound on its own, or does it need a cycle? Buffett's businesses generate cash and reinvest it whether he acts or not, so time is his ally and selling is pure loss. Zell's buildings do not compound — they ride credit cycles — so time is his enemy and exiting is the entire craft. Ask honestly which kind you own. And note the third case: Rihanna owns a compounding-looking asset that is actually depreciating, held it like Buffett, and is watching the mark fall.
Three · Which leverage is the good one?
Code and media. They replicate at zero cost and need nobody's permission, which is why a person with a laptop can start today and a person waiting for capital cannot.
Capital — and the strongest evidence ever assembled is that permissioned capital, obtained cheaply enough, beats everything. Float at negative cost is the most powerful leverage that has ever existed.
They are both right and the disagreement is about starting position, not about physics. Capital is the higher-return leverage and it is closed to you at zero. Code and media are lower-return and genuinely open this afternoon. The practical reading is sequential rather than competitive: permissionless leverage is how you get to the point where permissioned leverage becomes available to you. Four of the ten in this book made their fortune on capital — and every one of them had access to it before they started.
Four · Conviction, or feedback?
Definite optimism. Have a specific plan and hold it with conviction. Iteration, testing and incrementalism are what people do when they have no secret and no vision.
In fat-tailed domains you cannot forecast, so never take a position that cannot survive being wrong. Many small bounded bets, no single unbounded one.
The evidence is unkind to conviction. Thiel's own hedge fund ran the exact posture his book prescribes — one high-conviction thesis, no feedback loop — and collapsed. Meanwhile Huang, the great conviction story, quietly funded his decade-long bet out of a boring profitable business rather than betting the company. The synthesis is: be definite about direction and bounded about exposure. Conviction is free; irreversibility is what costs you. If you cannot survive being wrong for three more years, you are not being convicted — you are being reckless, and the two feel identical from the inside.
Five · One thing, or many?
One thing, past the point where it is boring. Thirty-three years in one company. One offer, one price, one script, sold thousands of times identically. Variance is what kills.
Many at once. Mobile networks in five countries simultaneously. Dozens of brands acquired across four decades. Breadth is the strategy.
This one only looks like a contradiction, and seeing through it is worth more than any other page here. Sawiris was not running five businesses — he was running one playbook in five geographies: buy the licence nobody will price, launch fast, own prepaid distribution. Arnault has performed the identical act perhaps thirty times: buy a heritage name, plug it into owned retail, raise price. Neither man diversified his method. They replicated one method across many instances.
So the real rule is: one playbook, as many instances as you can run. The trap that catches capable people is the opposite — many playbooks, one instance each. Five businesses that each require a different skill, a different buyer and a different form of leverage is not a portfolio. It is five apprenticeships running concurrently, none of which will ever reach mastery, and it is the most respectable-looking way to stay unwealthy.