How to Get Better at Size It Up: Judging Size by Eye

The first few rounds of Size It Up feel like a test of general knowledge: either you know how long a giraffe is or you do not. After a few dozen runs it becomes clear that knowledge is the smaller half. Most rounds are lost to the same handful of habits, and habits can be changed. The numbers below are the rules of Size It Up Unlimited, and the habits carry over to any version of the game.

Size It Up does not care how many metres you missed by. It cares how many times too big or too small you were, which is why a guess at a whale is marked exactly like a guess at a beetle. Within 5% of the true size is a perfect 100. Fifteen percent off still earns 90, twenty percent earns 85, and fifty percent too big earns 63. Twice the true size scores 37, and so does half of it, because twice too big and half the size are the same mistake measured as a factor. Three times off earns 10, and four times off in either direction earns nothing.

The shape of that curve is most of the strategy. It is gentle near the answer and falls away hard further out, so the gap between a careful guess and a perfect one is a handful of points, while the gap between a careful guess and a wild one is most of the round. The goal is not to be exact. It is never to be twice off.

A run is five rounds, and they are not worth the same. The first two count once, the third counts double and the last two count triple, so a perfect run is 1,000 points and 600 of them sit in rounds four and five. A careless first round can cost you a tenth of the run. A careless fifth round can cost three times that.

The late rounds are harder on purpose, too. Their targets come from the less familiar end of the catalogue, and the size gap between the two shapes is allowed to stretch further. Slow down where the multiplier is, not where a round happens to feel hard.

The reference on the left is the most useful thing on the board, and it is easy to stop looking at once a number has arrived in your head. Its real size is written across it, but the better use of it is as a ruler: how many of these would fit along the other thing? "Two buses end to end" is a far more reliable judgement than "about twenty-five metres", because you are comparing two shapes you can see rather than recalling a figure you half remember.

Then check the other direction. Once the length looks right, look at the height against the reference as well. If one edge looks right and the other looks absurd, the drawing has fooled you, and it is usually the absurd edge that is telling the truth.

The label on your guess says what is being measured, tall, long, wide, across or in area, and that word matters more than it looks. Animals are a typical adult, usually measured nose to tail. Statues are measured by height, and a few include their pedestal. Stadiums are measured across their full width. Bridges are the classic trap: they are measured by their main span, the stretch between the towers, not their full length end to end. The Brooklyn Bridge's main span is under 500 metres; the whole bridge is well over a kilometre and a half. Guess the wrong one and you are nearly four times off while feeling sure of yourself.

Countries, states and lakes are measured by area, and the answer is written in square kilometres and square miles. The score, though, is judged on the size of the shape on the board, not on the ground it covers. A guess with double the true area is only about 41% too wide, and it still earns around 70 points. Half the true area is about 29% too narrow, which scores the same. You have to be four times off in area before you are twice off in width and down at 37.

That changes how to play places. Being close is easier than it looks, so spend your care on getting the width right against the reference and stop worrying about the exact number of square kilometres. The number that comes back at the reveal will often look alarming next to your guess and still be a good score.

People judging a quantity drift toward the middle of whatever range they have been shown. Harry Hollingworth measured this central tendency of judgement in 1910, and it has turned up in almost every estimation study since: small things are guessed too big and large things too small. Area makes it worse. In S. S. Stevens's classic measurements, perceived area grows more slowly than real area, so a shape twice as big does not look twice as big.

In practice, when a guess for something enormous already feels generous, it is probably still short, and when a guess for something tiny already feels minuscule, it can go smaller. The reveal leaves a dashed outline where you guessed. Watch which side of the true shape it lands on for a few runs. Most people miss the same way round after round, and once you know your direction you can correct for it before you lock in.

After each reveal, Size It Up turns the answer into something familiar: "about 9.5 dollar bills end to end" for an aardvark, "as tall as 3.5 Statues of Liberty stacked up" for the Eiffel Tower. Those comparisons come from a short list of yardsticks, and knowing their sizes turns every reveal into practice for the next round.

• A credit card is 8.6 cm long and a dollar bill 15.6 cm.
• A baseball bat is just over a metre, an adult man 1.75 m, and a basketball hoop 3.05 m off the floor.
• A school bus is 12.2 m long, and a football field 91.4 m from goal line to goal line.
• The Statue of Liberty is 93 m from the ground to the tip of the torch, and the Empire State Building 443 m to the tip of its antenna.
• For places, New Jersey is about 22,600 km², Florida about 170,000 km², and the lower 48 states about 8.1 million km².

A perfect run is 1,000 and effectively nobody gets one, because it takes five guesses in a row within 5%. Staying within 15% on every round is at least 900, which is an excellent run. Keeping every guess inside half again of the true size puts you past 600. If your runs sit below 500, the fix is almost never more knowledge. It is one round that went twice off, and that round is usually a place or a very large animal pulled back toward the reference.

Calibration only comes from volume, which is the argument for an unlimited size comparison game while you are learning: five new pairs a run, drawn from nearly two thousand things, as often as you like. If one kind of thing keeps catching you out, it is worth reading the sizes on their own. We have written up how big famous monuments really are and how big countries are once you take away the Mercator map. If places are the weak spot, MapTap asks the neighbouring question of where in the world they are.

Hollingworth, H. L., The Central Tendency of Judgment (Journal of Philosophy, Psychology and Scientific Methods, 1910).

Stevens, S. S., Psychophysics: Introduction to Its Perceptual, Neural, and Social Prospects (Wiley, 1975).