Hockey Stick vs. Linear Projection Gap
Humans forecast linearly; technology compounds exponentially — bet on the gap
People project technology adoption in a straight line because time feels linear to humans. But when a new technology is genuinely superior, and operates in the virtual world with no friction, adoption follows an S-curve with an exponential leading edge. Most market participants miss this and underprice the transition. The framework says: find the gap between what the consensus projects (linear) and what the technology will actually do (exponential) — that gap is where asymmetric opportunity lives.
Khosrowshahi developed this pattern recognition through M&A work at Barry Diller's IAC, where he justified paying above-market prices for companies like Hotels.com, Ticketmaster, and match.com. The insight was that apparent overpayment at the time was actually correct pricing when the exponential trajectory played out. He extended it at Uber, where the original TAM was estimated as taxi + black car market but expanded multiples beyond that through Jevons Paradox dynamics.
The Jevons Paradox extension is critical: when a technology makes something radically cheaper or easier, usage expands non-linearly beyond the original market size. Uber was not a 1:1 replacement for taxis — it created new trip demand that never existed. The same dynamic applies to AI replacing intellectual labor: the capability creates new categories of work that weren't economically viable before.
- Human time perception is linear; technology adoption curves are exponential — never confuse the map for the territory.
- Apparent overpayment is correct pricing if you have conviction on the exponential trajectory the market is discounting.
- Jevons Paradox applies to every major technology transition: cheaper access creates demand that did not previously exist.
- Virtual-world transitions face near-zero friction — once a better technology exists, the transition accelerates faster than any linear model predicts.
- Find companies executing fastest on the transition, not necessarily the largest incumbents.
- Identify the transition vectorName specifically which behavior is moving from one platform to another — physical to digital, human labor to AI labor, heuristic rules to ML models. The transition must be real and irreversible, not hype.Pro tipLook for transitions where the new technology has a 10x advantage in at least one dimension (cost, speed, reliability) — incremental advantages don't generate hockey sticks.
- Map the current consensus projectionExplicitly write down what the market currently believes about the TAM and adoption timeline. This is your baseline linear projection — the thing you are betting against.Pro tipAnalyst reports and incumbent investor calls are the best source for the consensus linear projection.WarningDon't confuse 'the market is wrong' with 'I am right.' The gap must be grounded in a specific mechanism, not contrarianism for its own sake.
- Locate the leading edge operatorsFind companies already executing on the exponential trajectory — not at scale yet, but with compounding adoption metrics. These are the acquisition or investment targets.WarningSpeed of execution matters more than size. The fastest operator on the right transition beats the largest incumbent almost every time.
- Apply Jevons Paradox to resize the TAMAsk: if this technology makes X 10x cheaper or easier, what new categories of demand become viable that didn't exist before? Restate the TAM including this latent demand.Pro tipThe original taxi + black car TAM for Uber was ~$30B. The actual ride-share TAM became $300B+ because cheaper, easier access created demand that never existed.
- Calibrate the overpayment ceilingUsing the revised TAM and exponential trajectory, calculate the price at which you are still buying cheaply relative to the realistic outcome — not the consensus outcome. This is your conviction ceiling.WarningThis only works if the technology is genuinely superior with no friction holding it back. Do not apply to technologies that face real regulatory, physical, or network moats.
At Barry Diller's IAC in the late 1990s and early 2000s, Khosrowshahi applied this framework to justify paying above-market prices for internet properties that analysts considered expensive relative to current revenues. Each company was dismissed by incumbents as limited TAM. Each transition (physical to online travel, physical to online ticketing, physical to online dating) followed an exponential curve that made the apparent overpayment correct.
Uber's founders strongly opposed integrating the taxi segment, viewing it as a declining legacy business in direct conflict with the Uber brand. Khosrowshahi overrode this and launched the integration anyway. The founders' linear projection: taxis are shrinking, so adding them shrinks Uber's growth narrative. The exponential reality: adding taxis to Uber's tech layer expanded supply and demand non-linearly.
Uber now runs 40 million trips per day with pricing, routing, matching, and batching driven by small AI models — not heuristic rules. The transition from rules-based to ML-based operations was not visible in early Uber metrics. The exponential trajectory of ML capability compounded into a full operational replacement of human judgment at scale.
Khosrowshahi built this framework while executing M&A for Barry Diller at IAC in the early 2000s, buying early internet properties that incumbent analysts considered overvalued relative to current revenue. Every major deal — Hotels.com, Ticketmaster, match.com — was paid above consensus at the time. His retrospective insight: 'We never completed a successful deal because we got the company cheap. We actually overpaid for every single great company that we bought, but we overpaid based on what the market thought at the time, not what the reality turned out to be.'
He brought this directly to Uber's strategic decisions, including the counterintuitive acquisition of the taxi segment that Uber's founders strongly opposed. That segment became Uber's fastest-growing business — a direct validation of the framework applied in reverse (founders discounted taxi as declining linear; the reality was exponential growth driven by Uber's tech layer on top).