Well, the thing is that you don't know if you're "over engineering" up front. That's the whole challenge.
And rewriting down the road because of the wrong gamble can be very costly. Make the wrong bet enough times and you'll see that it's not over-engineering but just engineering for changing requirements.
A classic example is thinking it will be simpler to avoid React and use a few .show()/.hide() toggles. Simple at first. Then as you take on complexity, you start to feel like the lumberjack who thought he could save time by skipping the axe sharpening.
