You might be underestimating the timescales involved.
For example, grass - super simple organism, right? Should have appeared early on? No, dinosaurs appeared before grass did. But when the first grasses did evolve, wow, they were successful on a scale that is hard to overstate.
The beauty of sexual reproduction, from an evolutionary point of view, is that by its very nature, it allows many experiments to take place at once. The success criteria of each experiment is how many babies can the subject make.
Little wings evolve on seed pods for the same reason they evolved on anything else. For whatever reason, each step along the way made them a tiny bit more successful at having babies than those without.
Maybe a little spike makes them slightly less likely to be eaten, a bigger spike less likely still. A flatter spike helped them catch the wind and scatter further afield, and broader ones further still.
There’s no feedback needed for individuals in this system - it’s literally a numbers game based on who/what can make the most babies. They’re the ones who, millions of years down the line, end up winning.