The thesis

The asset is human capital itself: the revenue that people generate.

A power law

Ranked largest to smallest, that revenue follows a power law,

r_{(k)} \;\propto\; k^{-s}, \qquad s > 0,

where $r_{(k)}$ is the revenue of the $k$ -th name by rank.

A few names stand far above the rest; the rest run low and close together.

The shape is the point: the top of the curve is not a little larger than the middle — it is larger by orders.

The single name is a gamble

Held one name at a time, exposure is to a single draw from that distribution.

Most draws land in the low, crowded tail; the surplus sits in a few places no one can mark in advance.

A single position is therefore a bet on identifying the outlier — and the outlier cannot be identified in advance.

The book is an asset

Held as a book of many names, the same distribution becomes an asset.

The few high outcomes are captured because the book holds them all; the many low outcomes are absorbed because no single one carries the result.

What was variance in one name becomes structure across the book.

The tail index governs the regime

A power law is not one distribution but a family, indexed by how heavy its tail is. Let $\alpha$ be the tail index of the revenue distribution,

\Pr(R > r) \;\sim\; r^{-\alpha}, \qquad r \to \infty.

The index decides which moments exist, and therefore what diversification can and cannot do:¹

$\alpha > 2$ — the mean and the variance are both finite. The book's average behaves classically: it concentrates on the true mean, and its dispersion falls with scale.
$1 < \alpha \le 2$ — the mean is finite but the variance is not. The average still settles on the mean, but more slowly, and a single name can dominate a cycle.
$\alpha \le 1$ — the mean itself is undefined. No amount of diversification stabilizes the average; the largest name in the book carries it.

A heavier tail is the source of the surplus and the source of the fragility at once. The structure is built to hold the asset across the realistic span of $\alpha$ , not to assume the benign end of it.

Diversification is necessary, not sufficient

In the finite-variance regime, the book's realized mean tightens with scale — the formal statement is in The waterfall.

Outside it, diversification still helps but does not close the gap: dispersion falls more slowly than the square root of the count, and a single outsized name can still swing a cycle. Holding more names lowers the odds that any one breaks the book; it does not make the book self-insuring.

The structure therefore does not rest the senior tier on convergence alone. The baseline it pays is held first by realized revenue, then by the monetization partner's contractual support, then by the Protocol Reserve — a defined order of support that holds whatever the realized tail turns out to be.

Scale narrows the band; the support order holds the floor. The protection is structural, not statistical — set out in The waterfall and Economics.

Captured by scale, not selection

The outlier cannot be named in advance, so it is captured by scale, not selection.

Each name funded raises the odds of holding one, and steadies the whole.

As the book grows, the spread of realized return narrows where the tail permits, while the expected return holds; where the tail is heavy, the support order behind the senior tier carries what the statistics do not.

Capital is the lever: it is what converts the distribution from a gamble into a held position.

The rank exponent $s$ and the tail index $\alpha$ describe the same law from two angles; for a rank-size law $r_{(k)} \propto k^{-s}$ the distributional tail index is $\alpha = 1/s$ . Finite mean requires $\alpha > 1$ ; finite variance requires $\alpha > 2$ . ↩