If the STOVL variant is spun off into a seperate design the air force and naval variants could be more optimized. It is hard enough to make a stand alone STOVL aircraft, trying to give it commonality with standard aircraft is always going to be an issue. Many experts have said that separating the programs from the start would have been the best choice, and I think I read something about how the differences are so great they are effectively two different aircraft families at this point anyways.
I also would have switched the design priority and focused on making the naval aircraft first. There have been many naval aircraft that have successfully served with air forces, but when air force aircraft try to make the switchover to navies they tend have compromised maintenance and performance.
This is it basically. Cancel the Air Force version as anything beyond a cheapened version of a naval fighter, and drop the idea that the STOVL aircraft is related in much more than being developed simultaneously by connected teams and you should be able to get a decent successor to the Harrier and a stealthy F-16 replacement that can do carrier ops. You should even be able to get a decent amount of commonality, it's largely just a matter of dropping the idea of it actually being a single type.
That said, I also have to agree that what we have isn't a bad aircraft. It's expensive and late, but everything military is now. The price will come down with production numbers, and the capabilities are pretty much what was asked for. The concerns I've seen are largely about it's capability as a front line air superiority aircraft, and it never was supposed to be that; as with the F-16 the JSF was supposed to be a lighter, cheaper, multirole platform to sell overseas and to COMPLEMENT the air superiority fighters, and I don't see anything that contradicts that. Realistically the failing of the program isn't the aircraft, but that unreasonable expectations were put of what became a massively complex project.
Of course, the other option (and this is one I rather like) would be to make the project only semi stealthy. Whatever you do about the three in one nature of the aircraft (and I think in this scenario a semi stealth F-16, F-16N and fully independent V/STOL program are the way to go) this will help bring cost and complexity down a lot. The problem here, of course, is that when the program was launched we were very much thinking in terms of Desert Storm, and the late 90s, before it became clear that stealthiness is so expensive by its nature (rather than as a result of it's newness), and LONG before question started being asked openly about it's long term technical viability. In short, at the time, to propose a next gen platform that wasn't stealthy would have been a non starter.