Questions
Open questions
What I'm trying to find out. If you're working on any of these, or have an answer I haven't found, write to me.
Where does value go when software becomes a commodity?
Tool → outcome → prediction. Each step moves the money from what just became abundant to what stays scarce.
Why does a firm that serves firms never get paid for the view it has?
The multiplier never pays. The one who sees every company in a vertical sits one level below the one who gets the margin — until someone builds the level.
What does a rollup actually buy?
Not revenue. A seat in a vertical and the history that comes with it. And in most small firms, a person who's about to leave.
What makes behavioral data worth anything?
Not volume. Length, comparability, consent. One vertical, one system, watched long enough that the seasons show. The tape.
Can you sell the signal and use the signal at the same time?
No. Two games, and they're incompatible. Keep it from anyone who competes with you, sell it to everyone who doesn't.
How do you value a customer base when nobody's sold one yet?
V = N × r × CAC + P. The buyer's CAC, not yours. The seat priced at what it would cost anyone else to take it.
Does Jevons kill the thesis or prove it?
Price falls, demand rises, margin per unit collapses. What survives is what intelligence can't make more of: the tape, the switch, the seat.
What is an incumbent?
Whoever holds an advantage denominated in time, not money. Every vertical has two — one holds the product, one holds the relationship. AI is the moment you find out which one you were.
What keeps a customer when everyone has the same AI?
Not the product. The same model on a different history gives a different answer — and the one with everyone else's history gives the forecast. The customer stays with whoever sees their market first.