How to Estimate Almost Anything Without a Calculator

The question that seems impossible

How many piano tuners work in Chicago? The physicist Enrico Fermi was famous for asking students questions like this, questions that look unanswerable because nobody has the figure memorised. His point was that you do not need the figure. You need a handful of numbers you roughly know, and a way to chain them together.

That chain is now called a Fermi estimate, and it is one of the most useful habits you can build. It turns "no idea" into "somewhere around this", and it is almost always closer than a gut guess. It is also exactly the skill our How Close game tests: no calculator, no lifeline, just your sense of scale.

Step one: break it into pieces you can guess

The trick is to replace one hard question with several easy ones. For the piano tuners, it goes something like this, with every number a rough assumption rather than a fact:

• Chicago has about 3 million people, so roughly 1 million households.
• Perhaps one household in twenty has a piano: 50,000 pianos.
• A piano gets tuned about once a year: 50,000 tunings a year.
• A tuner can do about four tunings a day, 250 days a year: 1,000 tunings per tuner.
• So Chicago needs about 50 piano tuners.

Why the errors cancel out

Every one of those numbers could be off. Maybe one household in ten has a piano; maybe tuners manage only three a day. The surprising thing is that the final answer still tends to land within a factor of two or three of reality. Some guesses run high and others run low, and across several steps the errors partly cancel each other.

That is why breaking the question down beats guessing the whole thing at once. A single guess has one big chance to be wildly off. A chain of five modest guesses has five small chances, pointing in different directions.

Think in powers of ten

For most estimates the goal is the right order of magnitude: is the answer in the tens, the thousands or the millions? Getting that right is worth far more than the exact first digit. A guess of 300 when the answer is 500 is good. A guess of 300 when the answer is 30,000 is a different kind of wrong.

So when you are unsure, ask first which power of ten the answer lives in. Is the number of books in a big city library more like 10,000, 100,000 or 1,000,000? Choosing between those three is usually easy, and once you have the bracket, refining inside it is quick.

Average your bounds the right way

When you have a range in mind, say "somewhere between 100 and 10,000", the obvious move is to take the middle: about 5,000. That is usually the wrong middle. For quantities that span powers of ten, the better midpoint is the geometric mean, which here is 1,000, because 100 is ten times smaller than 1,000 and 10,000 is ten times bigger.

A quick way to do it in your head: average the number of zeros. One hundred has two zeros and ten thousand has four, so the middle has three: 1,000. When the bounds are not neat powers of ten, pick the round number that feels equally far from both in terms of "times bigger", not "how much bigger".

Sanity checks and anchors

Once you have an answer, test it against something you already know. If your estimate says every person in the country drinks forty cups of coffee a day, a step has gone wrong. Good anchors are the numbers you carry around anyway: the population of your country and city, the length of a day in seconds (about 86,000), a year in days, the height of a floor in a building (about 3 metres).

Beware the opposite problem too. Psychologists call it anchoring: a number you saw recently, even an irrelevant one, drags your guess towards it. In a famous 1974 experiment by Amos Tversky and Daniel Kahneman, people who watched a rigged wheel land on 65 guessed a much higher share of African countries in the United Nations than people who saw it land on 10. Build your estimate from your own pieces rather than adjusting from whatever number is in the air.

How to practise

Estimation improves quickly with feedback. Guess first, then look the number up, and notice which direction you tend to miss in. Most people underestimate very large numbers and overestimate small risks, and simply knowing your own lean is half the correction.

A few rounds of How Close a day is a good drill, because it tells you straight away how far off you were. If the arithmetic in the middle slows you down, the shortcuts in mental math that actually saves time help, and Math Sprint keeps them sharp.