The characteristic
What actually broke in the dot-com bust?
You can tell the growth story in your sleep. So could every CEO in March 2000.
The anchor between price and cash flow broke. Companies with no revenue and no path to profit were valued entirely on a growth narrative. The characteristic failure was not greed; it was a story so coherent that checking the math felt like missing the point.
The mitigation available then was almost nonexistent — no editable model of the downside, only conviction. Founders sat in IPO lock-ups they could not exit, holding paper that repriced around them.
The cost
What did believing the story cost?
It cost roughly 80% of the index. The NASDAQ shed nearly 80% of its value from March 2000 to October 2002 — trillions in paper wealth, much of it concentrated in single positions that could not be sold during restricted windows.
The implication for a concentrated holder is geometric, not arithmetic: an 80% drawdown requires a 400% gain merely to return to flat. The same arithmetic trapped holders in the 2008 global financial crisis, where leverage replaced narrative as the distortion. Robert Shiller named the mechanism — narrative-driven prices detached from fundamentals — in Irrational Exuberance (2000).
What would a known failure surface look like?
It looks like reading the position instead of believing it. You see the drawdown that would trap your equity, the lock-up window that removes your exit, and the valuation premium your net worth carries on faith. The story stays — but it sits next to the math, not on top of it.
The influence today
How does today’s influence factor show up?
The System Diagnostic
Today’s AI valuations rhyme with 2000: serious people repeating “this time is different” while revenue multiples run ahead of free cash flow. The mechanism is identical — narrative substituting for numerics. The diagnostic maps where your concentration breaks when the story is priced, the same engine we run against an AI sector re-rating in 2026. The defensive logic behind it — convexity bought cheaply against the tail — is set out in our note on tail-risk options.
Spec: diffusion sampling · fat-tailed paths · concentration-aware · stated 4.2% false-comfort rate
We print the rate at which the engine reports a tail smaller than the realized one, because a model that hides its error rate is the failure mode (see Research). Francis Bacon’s Painting (1946) is the pairing: a figure of authority rendered as distorted flesh — the moment the compelling story is exposed as a carcass, which is what the numbers do to a narrative-priced position.
The objection
“Isn’t ‘this time is different’ actually true for AI?”
Maybe in part — and the engine does not argue the story. It prices the position if the story is wrong. You edit the inputs, re-run, and read the exposed equations: not to deny the upside, but to know the drawdown you underwrite to hold it. The quantitative method behind the path simulation is documented, error rate first.
3 fields · 48-hour document · no call, no sequence.
Frequently asked questions
What caused the dot-com crash of 2000?
The link between price and cash flow broke. Companies with no revenue and no path to profit were valued entirely on a growth narrative, so when the story stopped attracting new capital the prices had nothing underneath them. The trigger was not fraud but a story coherent enough that checking the math felt like missing the point.
Could the dot-com pattern repeat with today’s AI valuations?
The same failure mode is already priced into parts of the AI complex: revenue multiples running ahead of free cash flow on a “this time is different” narrative. The story may even be partly right, but a position priced on the story still has to survive the math if the story is wrong. We model that explicitly in our AI disruption 2026 scenario.
What was the structural failure for individual holders?
Concentration plus illiquidity. Much of the paper wealth sat in single positions inside IPO lock-ups that could not be sold during restricted windows, so holders watched the repricing without an exit. An 80% drawdown then requires a 400% gain just to return to flat — the recovery is geometric, not arithmetic.
How would a tail-risk hedge have behaved through the bust?
A convex hedge — out-of-the-money protection bought cheaply while the index was euphoric — pays asymmetrically as the underlying collapses, offsetting losses on the concentrated position rather than tracking them. The cost is a steady premium drag in calm years; the payoff is solvency through the drawdown that traps unhedged equity.
What should investors watch for today?
Watch the gap between price and free cash flow, not the strength of the narrative. When multiples expand while cash generation stalls and the consensus explanation is that fundamentals no longer apply, the position is narrative-priced — the precise condition the diagnostic is built to measure.