Someone asked a question on one of the live feeds. Would SpaceX learn more from a successful landing or from a failure. I can't answer that, but it seems they proved the concept.
Thankfully this failure was at the very end instead of, say, twenty seconds into flight or exploding on the launchpad. This way they got to test every subsystem which is phenomenal and about as good of a result as they could have expected. However, the landing system didn’t trigger and they’ll never know exactly why. The engines ate themselves alive and it would have been great to study how it performed in that scenario but now they’re in pieces. The header tank failed to provide enough pressure and it would have been great to test it further. Now they have to comb through the pieces and piece things together like a detective. They learned A TON but it’s definitely better to get it home in one piece.
Your greater point about it being better to land than not land stands, but these are experimental rockets. They're expected to fail catastrophically and you plan accordingly.
It must be possible to find most pieces, and certainly check the engines, even deformed, for overburn signs.