>>7692052>The difficulty is the last few metres, right?
No, the difficulty is the whole descent, recovery, return to base, and preparation for next launch.
SpaceX has only attempted two landings on a solid surface, and only one of them failed in the last few meters (with the first, they ran out of hydraulic fluid so the "landing" burn became a wild divert maneuver to impact on-target, with the second, the throttle was sticky). They attempted several other recoveries before that.
The original plan was a parachute splashdown. The first difficulty with that is that the first stages were not aerodynamically stable or especially rugged. Without powerful propulsion, they would belly flop in the atmosphere and break up. So they basically needed to briefly relight the main engines to get down from hypersonic speed to something more manageable before resuming free fall.
Once they were planning on relighting the main engines, the prospect of a powered landing was far more attractive than a splashdown. A parachute splashdown is not a gentle landing, for any reasonable size of parachute it's more like a controlled crash, and even if there aren't waves to toss the floating stage around roughly, sea water is corrosive. So the value of a stage recovered by splashdown is in doubt. It will likely need refurbishing, if not to be treated as a salvage item where only some useful parts can be stripped.
A soft-landed stage, on the other hand, is likely be in conditition to simply inspect, refuel, restack, and use again. By its ability to land on one of its main engines, it's clearly in pretty good working order. The goal here is to have it work like an airliner: fly multiple flights before any signficant maintenance is needed, and last for thousands of flights without any major component replacements.
Falcon 9 isn't likely to achieve airliner-like reusability, but their next generation launch vehicle, based on real experience, might pull it off.