The original law, and the Bitcoin adaptation
Metcalfe's Law originally proposed that a telecommunications network's value scales with the square of its number of connected users — each additional user adds value not just for themselves, but by creating new potential connections with everyone already on the network. Applied to Bitcoin, the standard adaptation regresses market cap against the square of daily active addresses:
Active addresses is a genuinely on-chain, measured quantity (not modeled or estimated), which is why this model sits somewhat differently from Rainbow Chart or Power Law: one of its two inputs is a real network-usage metric, even though the relationship itself (market cap scaling with the square of that metric) is a fitted assumption rather than a proven law for a decentralized monetary network.
How far off the fit has run
Actual market cap has traded both well above and well below the fitted Metcalfe line for extended stretches — in recent years the ratio of actual to predicted has ranged from roughly 1.4x to over 21x, an enormous spread that undercuts any claim of tight predictive power, even while the broad correlation between address growth and market cap growth has some real basis.
Why active addresses is an imperfect proxy
One wallet doesn't equal one user — a single person can control many addresses (for privacy, for different wallets, for exchange-generated deposit addresses), and a single address (an exchange's own hot wallet) can represent thousands of underlying users. Address count is the best on-chain proxy available for "how many people are using the network," but a proxy is all it is — never a direct headcount.
Today's market cap plotted against the fitted line, with the real 1.4x–21x historical deviation range, is on the Metcalfe's Law chart.