I think the source of confusion is events aren't really exactly like IF statements. The event "Sprite X < 500" does not work like "IF Sprite X < 500 THEN run actions", because that only tests a single Sprite instance. Conditions filter down instances. So it's more accurate to describe it as working like "FOR EACH Sprite IF Sprite X < 500 THEN run action (on that sprite)". The "for each" iterative aspect of conditions is essential to the way events work, but means it does not act like a normal single IF statement, so it's not always correct to compare it to that.
In the example I gave, if you want an action to run on Sprite in that event, then you can only choose a Sprite that did not meet the condition. Therefore it makes it possible to run actions on instances that did *not* meet the conditions. I think this is totally counterintuitive and nonsensical, and although the current system may be sometimes confusing, the alternative is even worse.