I can understand the sense of "we just spent 8.7 billion dollars on this thing; we'd better not screw it up", but I wonder what the replacement cost would actually be if they had to launch another? Assuming they spent most of the money on R&D, it might actually be relatively cheap. Maybe we could even launch a couple smaller versions just to have backups and do the science tasks that don't require a full 21' mirror.

There's an interesting contrast between NASA's and Elon Musk's idea of what space exploration should be like; the former spends most of its efforts on one-off projects, whereas the latter is focused on making things cheap and repeatable and achieving reliability by iterating on a design rather than getting it perfect the first go-around.

Both approaches are needed, and certainly NASA paves the way for others to come along and do the same thing cheaper once it's been proven.