The Lily Pad

or, why the pond looks empty right up until the morning it isn’t

Jump to a simulation: the pond & the curve · a thousand dollars

There is a pond behind Sally’s house, and one June morning a single lily pad is floating on it — a green coin the size of your palm, turned up out of nowhere. It is a healthy pad. It does what healthy pads do: it grows, and it buds, and the patch of green doubles in size every day. One pad today, two tomorrow, four the day after. Sally, who likes the pond the way it is, does the arithmetic and finds that on the thirtieth day the lilies will have covered the water from bank to bank and choked it shut.

So here is the only question this page asks, and it is worth answering out loud, to yourself, before you read the next paragraph. The arithmetic is grade-school; the answer is not.

If a number jumped to mind, hold onto it. For a great many people — sharp people, people who balance budgets and read blueprints — the number that jumps up is fifteen. Halfway to thirty. It feels right the way a level floor feels right. And it is wrong by two weeks.

The pond is half covered on day twenty-nine. Of course it is. If the lilies double every day, then the day before the pond is full it is half full, and the day before that it is a quarter full, and you can run that backward as far as your nerve holds. On day twenty-five — with less than a week to go — the lilies cover about one part in thirty of the water. A frog sitting on the bank on day twenty-five would look out over a clean, open pond and conclude, reasonably, sanely, that the lily problem was somebody else’s problem for somebody else’s summer. The frog would be five days from losing the pond.

I want to be careful not to make this a story about how foolish the frog is, because the frog is us, and the mistake is not foolishness. It is wiring. We are built to think in sums, not products. Add one, add one, add one — that is the shape of a day’s walk, a season’s harvest, a stack of cut firewood, and our intuitions were tuned on exactly those things across a very long stretch of time when nothing in a human life doubled and then doubled again on a schedule. A straight line we can feel in the body. A doubling we have to compute, and computing it against the grain of intuition is like reading a word spelled backward — you can do it, but you feel the friction, and under pressure you reach for the answer that feels level instead.

A programmer would tell you the doubling is the simpler object. It is one line: each day, take what you have and multiply by two. There is no cleverness in it, no hidden machinery. And yet that one humble line outruns the intuition of the smartest person in the room, because the line does not get tired and our sense of scale does. Run the loop thirty times from a single pad and you have crossed from one lily to over a billion of them. Nobody’s gut feels a billion. So the gut quietly rounds the whole curve down to “not yet,” right up until “not yet” becomes “too late” overnight.

Don’t take my word for any of this. Here is Sally’s pond, and a chart of the same lilies beside it. It opens paused on day zero — press Begin and watch the two of them at once: the green creeping across the water, and the curve climbing the page. Watch where the pond still looks open. Watch when it stops.

The Experiment

Things to try:

The lily pad is the friendly version of a pattern that turns up wherever a thing’s growth is fed by how much of it there already is. And people have been getting blindsided by it for a very long time. There is an old story — first written down, as far as anyone can trace, by a thirteenth-century scholar named Ibn Khallikan — about the man who is said to have invented chess and was offered any reward he liked. He asked, modestly, for rice: one grain on the first square of the board, two on the second, four on the third, doubling across all sixty-four. The king, hearing “some rice,” waved it through. He had agreed to eighteen quintillion grains — a heap that would bury the kingdom, more rice than the planet grows in a thousand years. Sixty-four doublings is all it takes to turn a single grain into a number with nineteen digits. The inventor did not ask for much on any one square. He asked for the curve.

The same curve pays your savings account, when it pays at all: interest earns interest, and money left alone long enough does the lily trick, sitting nearly flat for years before it finally lifts off the floor, and almost nobody feels it coming in time. Money is the one version of this curve you can feel in your own life, so it has earned a small experiment of its own. A thousand dollars, left alone, at an interest rate you pick — and running alongside it, the same thousand earning simple interest, the kind that pays only on the original sum and never lets the earnings earn. Drag the rate and watch the gap between them open.

At seven percent — roughly what a patient index-fund investor has historically hoped for — the compound thousand grows to about thirty thousand over a working lifetime, while the simple version limps to forty-five hundred. Nudge the rate from seven to ten and the ending balance roughly quadruples; a few points of interest, compounded, is the difference between a modest sum and a small fortune. That is the dull, true, unwelcome advice the curve keeps trying to give: the one thing compounding cannot buy back is the years you spent off the floor. Start early beats invest a lot, because early is the only input the exponent multiplies.

The same shape draws other things, and not all of them are friendly. It is the curve a colony of bacteria draws in a warm flask, each cell splitting into two on a clock. It is the curve a rumor draws, and a contagion — one of the harder lessons the world keeps trying to teach, that a case count which looks like background noise on Monday can be a crisis by the end of the month, growing in exactly the silent, floor-hugging way the lily pad does. (If that particular curve interests you, it has its own page over in Six Degrees of Contagion, where the doubling runs along a network of who-touches-whom.)

And it is the curve under the machine you are reading this on. In 1965 an engineer named Gordon Moore noticed that the number of components his industry could cram onto a chip had been doubling about once a year, and ventured that it might keep on. Ten years later he eased the estimate to a doubling roughly every two years — and that revised line, christened Moore’s Law by other people, held for the better part of half a century. A doubling every two years does not sound dramatic. Run it for fifty years and it is a factor of tens of millions, which is the whole distance from a chip with a few thousand transistors to one with tens of billions, from a room-sized calculator to the slab in your pocket. Ray Kurzweil took that one curve and argued it was a special case of a larger one — his Law of Accelerating Returns — claiming that information technologies as a class grow exponentially, each generation handing the next a better tool to build the one after that. You do not have to buy his timeline for the Singularity to grant the narrower point: a lot of what humans build now rides curves their makers’ intuitions are no better equipped to feel than Sally’s is to feel her pond.

This site takes its name, with a wink, from Kevin Kelly’s What Technology Wants, and the borrowing is exact here. The lily does not want the pond in any sense that involves wanting. It has no plan to take the water and no idea the water ends. “Wants,” on this whole site, is shorthand for what a thing reliably tends to do when you leave it alone — and a process fed by its own size tends, reliably, to double, and to fool every linear mind watching it. The wanting is in the arithmetic, not in the lily.

Now the turn, because a page that only told you exponentials are bigger than they look would be selling you the same hype as everyone with a hockey-stick slide to pitch. Here is the part the slide leaves out: there is no such thing in the physical world as a true exponential. Not one. Every runaway curve anyone has ever measured is the bottom half of an S — a doubling that has not yet met the thing that stops it.

The bacteria in the flask double on schedule until they drink the sugar dry and choke on their own waste, and then the curve rolls over and flattens, and a fair number of them die. The pond fills and the lilies stop, because there is no thirty-first day’s worth of pond. Even Moore’s Law, the most quoted exponential of the age, has been visibly bending for years now as the features on a chip close in on the width of a handful of atoms, where the physics stops cooperating and the doubling stalls. Run the experiment above with the real ceiling switched on and you have watched this happen: the rocket becomes an S, the curve leans into its limit, the explosion turns out to have been a phase. That was Malthus’s whole nervous argument about people and food, two centuries back — he had the exponential right and the ceiling’s height wrong, which is its own lesson about how hard the top of one of these curves is to call from the bottom.

And the reason this belongs on this site, rather than in a brochure, is the thing the experiment showed you in Try 6: early on, the runaway and the S-curve are the same line. While the pond is still a tenth full, no measurement you could take from the bank would tell you whether you are riding a curve that bends gently into a comfortable plateau or one that is going to slam a wall you cannot yet see. The data look identical. This is the cousin of a problem that has its own page on this site — over in Tomorrow’s Weather, the very same S-curve equation, pushed too hard against its ceiling, stops settling down at all and tips into genuine chaos. The smooth saturating S and the unpredictable storm are two faces of one little rule about growth meeting a limit.

So this is not a hymn to exponentials, the way the conference circuit sometimes sings one. A doubling is not a verdict. The same curve, the very same arithmetic, describes compound interest and a tumor’s cell count; a Renaissance in chips and a panic in a crowd; the spread of a good idea and the spread of a bad disease. The lily pad does not know it is beautiful, and the cancer does not know it is monstrous. The curve has no opinion about which it is drawing — that part, as ever on this site, is left to us. What the math reliably tends toward is the doubling and the blind spot; whether a given runaway is a thing to feed or a thing to stop is a judgment the arithmetic cannot make and will not make for you.

Which loops us back to the frog on the bank, on day twenty-five, looking out at his open water. He is not wrong about what he sees. The pond really is nearly empty; the lilies really are a smudge. He is only wrong about what it means, and he is wrong in the one direction the human mind reliably leans — toward the level floor, away from the wall. There is a cheap fix and an expensive one. The expensive one is to grow a new intuition, which takes most of a lifetime and does not fully take. The cheap one is sitting right there in the experiment: when you suspect you are on one of these curves — a savings balance, a case count, a chip, a pond — stop reading the world on a linear axis and switch to the log. On that axis the doubling is a straight line, and a straight line, even a frog can extend.