Addition is commutative: 5 + 3 and 3 + 5 both land on 8, so swapping the order of the inputs changes nothing. This is so automatic you probably never noticed it was a rule. Then you meet subtraction, and the same swap gives you two different answers — 5 - 3 = 2 and 3 - 5 = -2. Not just different in magnitude. Different in which half of the number line they live on. This satellite is a counter-example machine: pick any two numbers, swap them under subtraction, and see the two results sit on opposite sides of zero.

The rule and its failure

For addition and multiplication of real numbers,

a + b = b + a \qquad a \times b = b \times a.

For subtraction and division, the analogous statement is false:

a - b \neq b - a \text{ in general} \qquad a \div b \neq b \div a \text{ in general}.

"In general" because there is exactly one edge case where both orders agree — when a = b, you get 0 or 1 no matter which way you read it. Every other pair produces two distinct answers.

The counter-example machine

Drag the two points a and b on the number line below. The two readouts compute a - b and b - a live. Watch what happens:

Interactive demonstration that subtraction is not commutativeAn interactive number line from negative ten to ten with two draggable red points labelled a and b, plus live readouts above showing the values of a, b, a minus b, and b minus a. As the reader drags the points, the two differences update and always mirror each other across zero.the two sums below differ by a sign flip — subtraction does not commute−10−50510↔ drag either point
Drag the red point ($a$) and the blue point ($b$). The readouts show $a$, $b$, $a - b$, and $b - a$. Notice that $a - b$ and $b - a$ are always mirror images — if one is $+7$, the other is $-7$. Only when $a = b$ do they agree, both being $0$.

Reading the numbers

Try these specific configurations on the slider and check that the numbers match what you would expect on paper.

a b a - b b - a land on the same side of zero?
5 3 2 -2 no
3 5 -2 2 no
7 7 0 0 yes (tied)
-4 6 -10 10 no
0 -8 8 -8 no

The pattern is unmissable. The only row where the two results match is the trivial a = b case. In every other row, the results are negatives of each other.

Why the mirror?

Read the algebra. You have a - b and b - a. Factor a -1 from the second:

b - a = -(a - b).

Why: distributing the minus inside gives -a - (-b) = -a + b = b - a, which is what you started with. So b - a is literally -1 times a - b. Multiplying by -1 on the number line is a reflection around zero.

So subtraction's non-commutativity isn't a random quirk — it is the statement that reversing the order flips the sign. The two answers always live on opposite sides of zero, equidistant from it, because reflection preserves distance.

Division behaves the same way

Exactly the analogous thing happens with division. 8 \div 2 = 4, but 2 \div 8 = 0.25. The two results are reciprocals of each other — not mirror images, but also not equal. Swapping a and b in a / b gives b / a = 1 / (a / b), the reciprocal.

Addition and multiplication let you swap inputs freely. Subtraction and division punish the swap — with a sign flip for subtraction, with a reciprocal for division. These are two sides of the same non-commutativity coin.

A place it bites on exams

In physics class you get a problem like "car A is at position x_A = 12 m and car B is at position x_B = 5 m; find the distance between them." The distance is |x_A - x_B| = 7 m, which is the same either way. But the displacement from A to B is x_B - x_A = -7 m, while the displacement from B to A is x_A - x_B = +7 m. The sign tells you which direction you moved. If you swap the two subtraction arguments without swapping the direction of the question, you give the right magnitude and the wrong direction — and in mechanics problems, that wrong sign propagates all the way through the momentum equation to a final answer off by a full factor.

This is the practical consequence of "subtraction doesn't commute": order carries direction. If you change the order, you change the direction, and physics problems are graded on direction.

A place it helps on exams

The flip is also a free algebra move. If you ever need to rewrite b - a as -(a - b), that's the reflection in action. It is sometimes useful for factoring: x - y and y - x differ only by sign, so (x - y)^2 = (y - x)^2 and you can always pick the order that makes the rest of the simplification cleaner. JEE factoring problems regularly hide this identity; noticing that b - a is "just" -(a - b) lets you combine fractions that would otherwise look different.

The reflex, in one line

If swapping the order of an operation's inputs changes the answer, the operation is not commutative. Subtraction and division fail the test; addition and multiplication pass. When you see subtraction in a problem, write the order the question gave you — don't reorder it to look more natural, because the order was carrying meaning.

Related: Operations and Properties · Number Systems · Is Subtraction Just "Adding a Negative"? · BODMAS — The Convention, Not a Theorem