This article explores why computers can’t do floating point maths, which is what makes Excel and all those online calculators such lousy mathematicians – basically because computers are binary, they can’t calculate anything after a decimal point, so the workaround is to put the number including the decimal point in a register of a certain size (say 32 bits) and reserve a few parts of the register for the decimal. Should the number you need to calculate become too big for the register, you run into trouble with rounding errors, which can compound. It then shows how nasty compound errors can become by citing the example of why a Patriot missle battery missed a Scud attack, resulting in the deaths of 24 people.

Why computers suck at maths | News | TechRadar UK.