Doug Messier has a long but useful piece on the current state of affairs.
We’ve been trapped in a moribund, high-cost space-transportation industry for decades because the cost of each flight is so high that it doesn’t allow for anything resembling real flight test. That has traditionally meant that much verification of design is by analysis, rather than test, and design changes are careful and rare. Working on more of a software model with frequent upgrades, SpaceX has broken that mold, with probably every vehicle slightly different, continuously improving the system, but walking a fine line with risk of failure from something new and untested. But a software change rarely has unrecoverable catastrophic consequences, as a rocket launch can. I suspect that ultimately the finding will be that such a change had an unanticipated effect that caused the recent loss.
ULA can’t (or at least traditionally hasn’t been able to follow) the same philosophy because of the conservatism of its customer, which values reliability above all else for its expensive, critical payloads. This makes it difficult to quickly evolve its own designs (one of the reasons, no doubt, that Tory Bruno wants a new rocket, rather than a re-engined Atlas that he knows won’t be competitive with SpaceX). One of the challenges they will have is how to break out of that mode.