Except the example he gave was also wrong. 3 + 4.2 is 7.2, using exact arithmetic, which should be the default assumption.
If you're measuring values under uncertainty, then his explanation makes some sense. But that's a property of measurement and uncertainty, not of numbers or arithmetic.
It's 7.2. Without context (and he gives none for the expression) everyone (including me as a scientist) must assume abstract numbers. What's more, we know he's not using measurements because there are no units...
If you're measuring values under uncertainty, then his explanation makes some sense. But that's a property of measurement and uncertainty, not of numbers or arithmetic.