It's more important to have every flight well checked then to have a hard very process and then trust people afterwards.
I would argue the opposite. In the world of process control, prevention is preferable to detection. Applied to rocketry, that means make sure people building rockets know what they are doing rather than relying on someone else to check that they did it right.
I have all of the NAR and TRA L3 documentation in a folder on my computer and I'm slowly digesting it assuming the day will come that I want to go for it, and it is surprising how much simpler the TRA stuff is.
As a relative newcomer (built my first HPR kit in January '15), it seems like there should be separate systems in place for "licensing" people to use certain size motors, and "certifying" people to build rockets of a certain size, complexity, or design. I, for example, crashed my DD Loc Hyperloc 835 on my first L2 attempt because I had a brain fart and didn't put the wingnuts on my AV-bay bulk head. This, to me, is a good case in support of having pre-flight checklists for DD rockets. After I watched the lower air frame smash into the ground, I got up the next morning and flew a J330 in a single deploy, motor eject PML Explorer. It landed a mile away, but I got my cert. It felt like cheating.
Its probably been discussed and subsequently axed, but I think it would be both informative/beneficial as well as fun/rewarding, if there were certifications for HPR Kits, HPR Scratch Builts, HPR DD Kits, HPR DD Scratch Builts, Multi Stage (Kit / Scratch), Cluster (Kit /Scratch), etc, and then subdivide those into weight classes. All of the scratch building certs could require demonstrating at least a fundamental understanding of the engineering (basic aerodynamics, basic strengths of materials, etc.).
When a rocket fails, its usually the result of an assembly error / flight prep error, not the motor, and when it comes in ballistic, its the size of the rocket, not the motor, that dictates the damage it could cause. I get that by regulating motor size, it indirectly regulates rocket size, but I think its too broad.
BC