It turned out that parachutes are great for soft uncontrolled landings, but if you want a soft controlled landing you need some form of propulsion. So if you want propulsion why not use the main engines, and since you are going to use those formthe soft controlled landing why not use them for the descent too.
At this point you have traded several tons of parachutes and many more points of failure which could cause loss of mission (parachutes are complex mechanisms) for a smaller weight of fuel and exactly the same points of failure as a rocket always has. Then SpaceX added grid fins, since purely propulsive landing was not going to be anywhere near accurate enough.
Classic example of working to your strengths, where the greatest strength of a rocket is acceleration along its primary axis :D
The ULA are now planning a reusable rocket to counter SpaceX, and they plan heat-shield then deploy parachute and finally capture by helicopter. This seems much simpler than SpaceX's landing approach as helicopters and large aircraft have been snagging things parachuting back from space since the beginning.
>It turned out that parachutes are great for soft uncontrolled landings, but if you want a soft controlled landing you need some form of propulsion.
They could do something like the Russians do with Soyuz - small solid rocket boosters that fire right at landing. But yeah, that's a lot more things that have to go right.
Also solid rocket motors are a lot less weight efficient than liquid fuelled rockets and are not throttleable and are awkward to gimble. Also where would you put them without increasing air resistance on the way up?
Also even with the rocket motors Soyuz hits the ground pretty hard.
As I understand it, the F9 first stage can't actually hover because a single engine at the lowest setting gives the overall stage a TWR > 1. Thus the "hoverslam" - they calculate the landing burn such that the stage reaches 0 velocity and 0 altitude (from the pad) at the same instant. But this means the only margin for error is in the landing legs, and they've had some hard landings.
I was originally thinking about solid rockets in context of a parachute system, but the more I think about the more I believe truly gentle landings aren't really a feature of the current design.
I'm not sure how much weight it would add, but it seems like having the ability to fine-tune the thrust at the very end would allow them to land much more gently and feel better about not introducing metal fatigue or other damage. Of course the addition of a smaller LF engine probably makes the most sense.
The kinds of parachutes you'd need for a first stage are surprisingly heavy. I doubt they would save any weight, since you have to use the rocket anyway to slow down to the parachute's operational range.
In addition you are more mission-flexible with the weight expended for fuel. It can be used to either soft land the stage, or give extra boost to a GEO mission or a heavier payload.
Early on they tried parachutes but ditched them, iirc.