the second hundred · metaphor 198
You make something quicker, cheaper, easier to do — and instead of doing less of it, you find yourself doing more. Every hour the efficiency saved gets poured straight back into the very thing it was meant to spare you.
You buy the faster laptop to save time, and end up doing more on it. The city widens the motorway to ease the jam, and within a year the jam is back, longer. The lightbulb that sips a tenth of the power doesn't cut the electricity bill — it just means the whole street stays lit all night. There's a quiet expectation buried in the word efficiency: that using less per use means using less overall. Often it means the opposite.
The reason is almost too simple. Efficiency doesn't just save resources — it lowers the price of the thing itself, and cheaper things get used more. Whether the extra use swamps the saving is not a matter of virtue or willpower. It's a matter of how hungrily demand answers a lower price. Below a certain hunger you really do save; above it, the saving is not just erased but reversed. And there is an exact tipping point where the sign flips.
The mechanism
Follow the chain. Doubling efficiency means each unit of service — each mile driven, each lit room, each computation — now takes half the resource. But the resource costs money, so the effective price of the service halves too. And a cheaper thing gets used more. How much more depends on elasticity ε: the percent rise in use for each percent fall in price.
The instrument draws this exactly. On the demand curve, improving efficiency slides the equilibrium down and to the right: lower price, larger quantity. Total resource consumption is price × quantity — the area of the shaded box. Two forces pull on that box: the price side shrinks it, the quantity side stretches it. When demand is timid (ε < 1), the shrink wins and you genuinely save. When demand is eager (ε > 1), the stretch wins and the box grows — you burn more than before you got efficient. At ε = 1 the two forces exactly cancel: the box holds its area, the saving is a perfect wash. That single value is where the sign of the whole enterprise flips.
What to try
Set elasticity low — the inelastic preset — and drive efficiency up. The equilibrium slides down a steep demand curve, the box shrinks, and the lower panel shows real resource used falling below the baseline: efficiency delivered. Now raise ε past 1. The demand curve tilts flat, the box begins to swell as you improve efficiency, and the lower curve bends upward through the baseline into the red zone — the backfire. Same efficiency gain, opposite result, decided entirely by how hungry demand was.
Hold at ε = 1 exactly and slide efficiency across its whole range: the box keeps a constant area and the lower curve runs dead flat along the baseline. That's the knife-edge. The rebound readout puts a number on it — the fraction of the engineering saving that gets eaten by extra use. At ε below 1 it's under 100% (some saving survives); at ε = 1 it's exactly 100% (all eaten); above 1 it tops 100% (you're worse off than when you started). The marginal rebound simply equals the elasticity — the cleanest statement of the paradox there is.
The mapping
Jevons noticed it in 1865 with coal: better steam engines used less coal per unit of work, and Britain promptly burned far more coal, because cheap power opened a thousand new uses. The shape recurs. Traffic: add lanes, driving gets cheaper in time, more trips appear until congestion returns — “induced demand.” Lighting: centuries of ever-cheaper light haven't lowered what we spend on it; we just light everything, always. Your own hours: the tools that make a task ten minutes instead of an hour rarely give you the fifty minutes back — you do six times as many tasks, and the inbox that empties faster fills faster.
The human sting is in that last one. We adopt the efficient tool as a way to do less — less driving, less spending, less toil — and discover we've only made the thing cheaper to want, so we want more of it. The rebound isn't a moral failure; it's what a lower price does. Which is why efficiency alone rarely shrinks a total: if you actually want to use less, you usually have to hold the price up — a tax, a cap, a rule — so the saving can't be spent straight back on more of the same.
Read as life lessons
Making something efficient lowers its effective price, and lower prices summon use. Saving per use and saving overall are two different things, and they often point opposite ways.
Whether efficiency saves or backfires isn't willpower — it's ε. Below one you save, above one you backfire, and at exactly one the gain is a perfect wash. The number decides.
If the goal is a smaller total, efficiency needs a price floor beside it — a tax or cap — or the freed resource is simply spent again on more of the very thing.
The mapping, exactly
| Mathematics | Life |
|---|---|
| efficiency e | How much service you wring from each unit of the resource — miles per gallon, tasks per hour. |
| effective price p = 1/e | What the thing costs you to do once it's efficient — the price that quietly drops as e rises. |
| elasticity ε | How hungrily you answer a lower price — the appetite that decides whether saving survives. |
| quantity Q = eᵉ̂ | How much of the service you end up wanting once it's cheap — the trips, the lights, the tasks. |
| resource R = p·Q | The total actually burned — the shaded box, and the only figure the planet or the paycheck feels. |
| ε = 1, the sign flip | The knife-edge between efficiency that saves and efficiency that swells the very use it promised to spare. |
The honest model
One constant-elasticity demand curve, scaled so the baseline sits at 1. Efficiency e sets the effective price p = 1/e (resource price fixed at 1). Demand answers with Q = p^(−ε) = eᵋ, and the resource actually consumed is service divided by efficiency, R = Q/e = e^(ε−1) — which is also exactly p·Q, the area of the box on the demand diagram. The instrument places the equilibrium and shades that rectangle live; the lower panel plots the real R(e) against the naive promise 1/e.
Everything hinges on the exponent ε−1. Negative (ε < 1): R falls with efficiency — saving. Zero (ε = 1): R is flat — a wash. Positive (ε > 1): R climbs — Jevons backfire. The rebound shown is the share of the engineering saving eaten by extra use, 1 − (1−R)/(1−1/e), whose limit as the improvement shrinks is simply ε. No number is typed in; each is read off the same two closed forms as you move the sliders.
Where the metaphor tears
The tidy model makes elasticity constant, but real appetites saturate: you can only drive so far, light so many rooms, run so many errands. Most measured rebounds are partial — well under 100% — so efficiency usually does save something, just less than the brochure claimed. Treating every efficiency gain as a certain backfire is as wrong as ignoring rebound entirely. The paradox is a warning, not a law of nature.
The sharpest Jevons effects are economy-wide and indirect: cheaper energy doesn't just grow its own use, it makes everything built with energy cheaper, rippling into uses the single curve never sees. That macro rebound is real but murky — hard to measure, easy to over- or under-claim — and no one-market diagram captures it. The instrument shows the mechanism cleanly by leaving most of the economy off the page.
The framing treats rising consumption as failure, but the extra use is often the point — light, mobility, and computation reaching people who had none. When efficiency lets the poor heat a home or a student reach the internet, the “rebound” is welfare, not waste. Whether backfire is bad depends on what the resource is and who now gets to use it — a judgment the arithmetic can't make.