In D&D 4E as with many RPGs, characters have a fixed armour value or AC. When attacking, the attacker must roll a die to overcome the target's AC. In 4E offensive magic works the same way: the target has a fixed Fortitude/Reflex/Will defence determined by their stats, and in order to hit, the attacker rolls to overcome whichever defence is relevant.
However it was not always like this. Some spells in 3E and now again in 5E work the opposite way. In those editions the target is the one who rolls against a fixed value determined by the caster's stats, in order to avoid spell-damage.
Both methods compare the attacker and target's stats, and have an element of randomness. I personally like the 4E system as it simplifies combat without leaving anything out. Why was the decision made, when designing the early D&D editions (and still when iterating upwards into (A)D&D 1e–3e), to use saving throws to control spells?
Answers should consist of designer quotes and not personal opinions.