Bigger but this [1] test by Armadillo was more impressive in my opinion for the additional complexity involved in cutting the engine deploying a chute then cutting the chute and successfully landing from a turbulent freefall.
Its an intriguing question isn't it? Here we have two companies which together have all the moving parts necessary for a lunar landing and return mission. It would be fascinating if Bigelow Aerospace could put a habitat at the Lunar L1 point. Then SpaceX provides a boost to L1 capability, and Armadillo provides a transport service from L1 to/from the surface of the Moon.
Yeah, they seem to have stalled in recent years. Either Carmack has lost interest or they've reached the point of complexity where a few volunteers aren't going to cut it.
Armadillo has recently flown the Stig and Stiga rocket quite high with succesful recovery.
Since they have the landing and low speed part pretty much covered, they're now looking at the high speed and altitude problematics.
The Stig lands with a steerable chute.
High speed stability, drag and turning for powered landing are a nasty problem field where testing is hard and expensive. Interplay of aerodynamics in a wide speed range, center of gravity shifts and everything else.
Burt Rutan's SpaceshipOne solved it elegantly with feathering: the descent was safe and easy. The mechanism could be troublesome though.
And for a first stage, returning is not nearly as hard.
At first, the reusable first stage might provide only a modest portion of the delta vee and an expendable second stage would do most of the work. That would mean a low payload.
I don't think the SpaceshipOne solution would work "as is", but something like it might work. It only went vertically to 100 km.
When the Space Shuttle was being designed, Max Faget toyed around with different ideas on how to arrange re-entry, with essentially something similar as SpaceshipOne in a sense: a high drag configuration at high speed and then flying at low speed. http://www.astronautix.com/lvs/shuenara.htm
Sure, the first stage is easy from an energy perspective, so SpaceshipOne is a good model for that. From what I understand they intend to try to use the same idea for reentry from orbit, though, and I think they'll end up with something expensive and dangerous as a result.
The space shuttle was a disaster from a cost perspective. I don't know how many lessons we can draw from the program beyond "don't do it this way".
Though Virgin Galactic is very far from orbit yet. They've been doing glide flights with SpaceshipTwo for years since their big hybrid rocket motor has not been progressing very fast and even had a ground accident. Stuff like that can happen if an airplane body company builds a rocket ship.
We can draw a lot of lessons from the space shuttle - it was very successful in lifting re-entries. It takes a huge amount of expensive wind tunnel and test time to develop something like this.
Why STS was so expensive must be analyzed more deeply - we definitely can not afford to waste all the lessons learned in that program. I think, in details, it contained many technologies that were not robust and required lots of checking and hand work (people with salaries). This also made it less reliable. Strategically, it was a huge one off program, meaning there was no possibility of iteration and trial and error: feeding information from operations back to improving design. (Cue Gall's law.)
Since development is so expensive (afaik a lot of the testing infrastructure has been driven down), this is also why most space capsules use known-to-work shapes: Vostok, Mercury, Gemini, Apollo or Soyuz. Some small capsule designs have even used Corona spysat film return canister shape.
[1]https://www.youtube.com/watch?v=9u0qlIoSSkQ