Why being 10× off is actually good

In most quizzes, close doesn't count. In estimation it is the entire game — and the scale that decides what counts as close is not the one you learned in school.

Zoggle · a guide for people who like numbers

Suppose the answer is 100,000 and you guess 50,000. How wrong were you?

The instinctive reply is "by 50,000" — you subtract. That is how we are taught to measure error, and for most of daily life it works fine. But it falls apart the moment quantities get large. Being 50,000 off matters enormously if the answer was 60,000 and not at all if the answer was fifty billion.

Estimation uses a different ruler. Instead of asking how far apart two numbers are, it asks how many times apart they are. Guess 50,000 against 100,000 and you were off by a factor of two. Guess 200,000 and you were also off by a factor of two. Both misses are equally wrong, which matches how it feels — both are "about right, in the wrong direction."

What an order of magnitude is

An order of magnitude is a factor of ten. Numbers that differ by one order of magnitude differ by a single digit's worth of size:

Getting the order of magnitude right means landing on the correct row of that table. It is a genuinely useful level of knowledge — knowing that something is "in the millions" tells you almost everything you need for a decision, even if you're fuzzy on whether it's two million or eight.

Your brain is bad at big numbers, and that's not your fault

Ask someone how long a million seconds is and most people shrug. About eleven and a half days. Now ask about a billion seconds: almost 32 years.

Nearly everyone finds that gap shocking, which reveals something important: we process "million" and "billion" as adjacent words rather than as quantities a thousand times apart. Human intuition evolved for counting things in front of us — a handful of berries, a dozen people, maybe a few hundred animals. It has no native machinery for anything past that, so it compresses everything large into a single blurry category labelled "lots."

This is why headline numbers so often fail to land, and why practising with orders of magnitude is worth doing. You are installing a sense that did not come standard.

So what counts as a good guess?

On a multiplying scale, here is a fair reading of performance:

• Within 2× — excellent. You didn't just find the right scale, you found the right neighbourhood inside it. On hard questions this is closer to luck than skill, and it should feel good.
• Within 10× — the real target. You identified the correct order of magnitude. In practical terms your answer is useful: it would let you make a sensible decision.
• Within 100× — a miss, but an informative one. Usually it means one link in your reasoning chain was badly off rather than that your whole approach was wrong. Worth finding which link.
• Beyond 100× — a structural error. You misread the units, misunderstood the question, or made an assumption that was wrong by a mile. These are the most instructive misses, because the lesson is large and obvious.

Averaging on a multiplying scale

There is a subtlety in scoring several estimates together. If you were 10× too high on one question and 10× too low on another, what was your average error?

Ordinary averaging gives a nonsense answer, because 10× too high looks like "+900%" while 10× too low looks like "−90%" — the same quality of miss described by wildly different numbers. The fix is the geometric mean: multiply the factors together and take the root, rather than adding and dividing.

This is the correct tool whenever you're averaging ratios, growth rates, or anything else that multiplies rather than adds. It is also why one catastrophic miss can't be hidden behind two good guesses: on a multiplying scale, a disaster stays a disaster.

The practical payoff

Once you think in orders of magnitude, a certain kind of nonsense becomes visible instantly. A claim that a policy will "save the average family $50,000 a year" collides with the fact that median household income is a few times that. A statistic about a disease affecting "millions of Americans" invites the immediate question: out of 350 million, is that one in a hundred or one in three?

You don't need the exact figure to sense that something is off. You need the right row of the table — and that is a skill you can build.