Learning curves: What does it mean for a technology to follow Wright's Law?

What it means here:Here it means the total amount of solar panels ever installed, added up across every year — not how many went up in any single year, but the running grand total so far.

In general:Describing an amount that keeps growing as each new piece is added to everything that came before — the accumulated total to date, rather than a single moment's figure.

What it means here:Here it means settling the question from actual measured data — the real records of prices and production — rather than just reasoning about what ought to be true.

In general:Based on observation, experiment, and real-world evidence rather than on theory, guesswork, or logic alone.

What it means here:Here it means a countless, wide-ranging number of tiny improvements — so many separate little advances that you couldn't easily list them all.

In general:An extremely large, practically countless number of things; a vast variety.

Making the very first pizza is a lot of work: you buy the oven, find a recipe, clear the counter, learn where everything goes. But once all that is set up, the second pizza is easy — a little more dough, sauce, and cheese, and the tenth is easier still. The big costs got paid once, so the more pizzas you make, the less each one costs on average.

That is economies of scale: as you produce a larger quantity of something, the cost per item tends to fall, because fixed costs — the oven, the factory, the design work — get spread across more units, and you can buy materials in bulk. A giant factory stamping out a million phone cases pays far less per case than someone hand-making ten.

This is why big producers can usually sell more cheaply than small ones, and why "we'll make it up in volume" is a real strategy. The article points out that this kind of saving is ordinary and expected — the surprising thing about a learning curve is a separate, extra effect stacked on top of it.

Most change you picture is "add the same amount each step": save $10 a week and your money climbs in a straight, gentle line. Exponential change is different — it multiplies by the same percentage each step. Money growing 10% a year, or a rumor where every person tells two more, doesn't crawl up a ramp; it hugs the ground for a while and then rockets. Fold a sheet of paper in half about 42 times and, in theory, it would reach the Moon, because each fold doubles the thickness.

Falling numbers can be exponential too. Solar panels dropped about 20% in price every time the total ever built doubled. Twenty percent off, again and again, compounds into a 99.6% collapse over four decades.

The catch is that exponential curves are hard to read on a normal graph — they look flat, then suddenly shoot straight up. So scientists use a logarithmic scale, where each step up the axis means ten times more (1, 10, 100, 1,000…) instead of one more (1, 2, 3, 4…). On that kind of axis, steady exponential change straightens into a clean line — and a straight line is easy to spot, measure, and extend into the future. Whenever you see a chart with axes labeled 1, 10, 100, 1,000, that trick is being used.

Ice-cream sales and drowning deaths both rise every summer. Does ice cream cause drowning? Of course not — hot weather causes both. That is the classic trap: when two things move together, they correlate, and it is tempting to say one causes the other. But moving together doesn't prove it.

There are three ways to get fooled. Maybe A really does cause B. Maybe B causes A — that's called reverse causality. Or maybe some hidden third thing causes both. In the solar story, prices fall as experience grows, but which way does the arrow point? Does building more make it cheaper, or do people only build more after it gets cheaper?

The clever fix is a natural experiment: find a real situation where the cause couldn't possibly run backward. During World War II, demand for weapons was driven by the war itself, not by low prices — so when production shot up and costs fell, and then the price decline slowed once the war ended, you could be confident that experience was driving prices down, not the reverse. Whenever someone claims "X causes Y" from data, ask: could it be Y causing X, or a third thing causing both?