The shuttle is still an interesting science experiment in what it did teach us about reusability in space craft, failures included. At the time reliable booster recovery and reuse seemed unlikely and command module reuse seemed the best way to drive costs down. The shuttle was a good command module design for reuse based on available data at the time (even with all the design by committee compromises), but proved to be difficult and expensive to maintain/refit between missions.

Today we've seen incredible advances in booster recovery and reuse in part spurred by learnings from shuttle reuse including the learning about how expensive heat shield maintenance/refitting turned out to be on the shuttle.

(A bit of an aside, but ironically and sadly the Challenger disaster didn't "fail hard enough" to drive costs down or do much to reimagine the shuttle program: that problem was a dumb booster problem that had nothing to do with the shuttle directly and could have happened to any payload. (It was still terrible that it happened to that shuttle and its crew, specifically, of course.) It's curious to wonder if something more like the Columbia disaster, which did more reflect shuttle maintenance issues, may have had an interesting scientific impact if it had "switched places" in a different timeline.)