New Shepard of Blue Origin separates the crew capsule from the booster. The crew capsule is in free fall for a while, then the parachutes open and finally the landing rockets give a brief impulse to land it softly. Meanwhile the booster free falls without parachutes and fires to almost hover to also touch down softly. Actually, it looks to me that the crew capsule dingles more than the booster does (during the last test linked to below).
Wouldn't it be simpler and safer to skip the separation and the parachutes and the double touch down impulses, to instead keep the crew capsule on top of the booster to land both at once still integrated? The crew capsule could rely on its launch abort system to function also as a landing abort system. To separate only when it increases safety to do so, instead of every time.
In this video of the test launch 19 June 2016, the booster landing looks softer than the crewed capsule landing. The time in free fall for the booster seems to last 5 minutes (from about 2:00 to 7:00 in the video). For the crew capsule, free fall lasts only 40 seconds longer (until the drogue parachutes unfold) or maybe 100 seconds longer until the main parachutes are unfolded. Is that difference of free fall time, which I suppose is what their customers are in demand of, worth the extra complexity and risk? Couldn't it be made up with a simpler integrated design? Is it maybe designed to give experiences more relevant to planned orbital launches, and thus by intent in itself not optimal for suborbital crew launches?