What Is an Order of Magnitude? Thinking in Factors of Ten

Quantities in the world span ranges that ordinary arithmetic handles badly, from the size of an atom to the size of a galaxy, and comparing them by subtraction is useless while comparing them by ratio is informative. Counting orders of magnitude means counting factors of ten, so two quantities differing by three orders of magnitude differ by a factor of a thousand. That way of thinking underlies scientific notation, logarithmic scales including those for sound, earthquakes and acidity, and the standard prefixes from nano to giga. It also matches how the difference between quantities actually matters in practice, since whether a cost is ten pounds or a hundred is a more meaningful distinction than whether it is a hundred or a hundred and ninety. The practical skill is knowing which order of magnitude an answer should land in before calculating it, which catches the errors that matter, since a slip of a factor of ten is far more consequential than a small arithmetic mistake and far easier to miss.

Order of magnitude estimation, sometimes named after Enrico Fermi, follows a recognisable method:

• •Break an unknown quantity into factors that can each be guessed within a factor of ten
• •Estimate each factor deliberately, stating the assumption rather than pretending to knowledge
• •Multiply them together, accepting that individual errors partly cancel since some guesses are high and others low
• •Check the result against anything independently known, which is where errors of scale reveal themselves
• •State the answer as a range or as a power of ten rather than as a figure with false precision
• •The classic exercise asks how many piano tuners work in a city, answered by estimating population, households, piano ownership, tuning frequency and tuner workload, and it lands close to the real figure remarkably often

The technique is not a party trick. Physics uses it constantly to check whether an effect is large enough to matter before any detailed calculation, and the practice of estimating first is taught explicitly. Engineering uses it for feasibility, since a design whose requirement is a thousand times beyond available materials does not need refining. Policy analysis uses it to check whether a proposed measure could plausibly close the gap it addresses, and many proposals fail that test immediately, which is a useful filter. Medicine uses it in dosing checks, where an error of a factor of ten is a recognised and lethal category of mistake that prompting systems are designed to catch. Business uses it for market sizing. In each case the value is the same, which is that a rough answer available in two minutes frequently settles a question that a precise answer would take weeks to produce, and settles it well enough.

People are consistently poor at large numbers and the failures are systematic rather than random. Scope insensitivity means that willingness to pay to save two thousand birds is barely different from willingness to pay to save two hundred thousand, which has been demonstrated repeatedly and which undermines reasoning about problems of very different sizes. Very large and very small numbers are compressed in perception, so a million and a billion feel similar despite differing by three orders of magnitude, and a useful corrective is that a million seconds is about eleven days while a billion seconds is about thirty-two years. Exponential growth is consistently underestimated, which matters for compound interest, epidemics and resource use. Probabilities at the extremes are handled badly, with very small risks either ignored or overweighted depending on how they are presented. Practising order of magnitude reasoning partly corrects these, and the reason it helps is that it forces a person to state the factors explicitly rather than responding to an impression.

Counting factors of ten compares quantities that subtraction cannot handle, and knowing which power of ten an answer belongs in catches the errors that matter most. Breaking an unknown into factors each guessed within a factor of ten, then multiplying, lands surprisingly close because individual errors cancel. A million seconds is eleven days and a billion is thirty-two years, which is the gap intuition keeps missing.