No doubt they should do an investigation to see if the cause of failure can be determined and corrected. It's possible that the fault may no longer be an issue; this was flying the soon-to-be-retired Merlin C engines, and the design or manufacturing flaw may not have carried over to the Merlin D engines. But what they shouldn't do is spend billions trying to make the rocket perfect, which has been the traditional approach. A few million + redundancy is much better.
This has been done many times in history, in Saturn V and Space Shuttle for example.
But in this case, since the engines are so small, it seems the explosion tactic paid off very well.
Hopefully they can test Merlin-1D adequately. If you aim for reusability, your reliability must go up considerably. If you plan to fly something for 100 times, you probably can not tolerate 98% per-flight reliability. A lot depends on assumptions of course.
And you're right -- for reusability, they do need to get their per-engine reliability above 98%. That would produce a mission reliability of 97.1% (assuming that 2 engines out causes a loss of vehicle, which I understand is only the case during the earlier phases of flight). At 99% engine reliability, you get 99.3% per-flight reliability -- which strikes me as the point where reusability starts to become very sensible. I have considerable confidence that SpaceX will be able to do this.
Then again, it is unfair of me to make relatively uninformed declaratives about tonight's launch. The engine blowout could have been cause by an incredibly unlucky set of circumstances, and the engineers at SpaceX could have built a vehicle that survived a catastrophic failure that would have doomed any other craft to date.
Given what I've seen from SpaceX, I am positive that they will get to the bottom of the incident and have it corrected before their next launch.