Earlier I stated that FAR Part 23 guarantees a certain level of safety.  In retrospect, this may have been somewhat overstated.  But there were some responses that have their genesis in ignorance of the certification process.

There is a quantitative process followed by manufacturers whereby they provide an agreed-upon level of safety based on failure probabilities.  The consequences of failures are determined, and depending on how bad it is (minor, major, hazardous, catastrophic) it is allowed to happen with a xx% probability.  The worse the effect of failure, the less likely it is supposed to be.

The companies predict the failure probabilities.  These estimates are only projections.  They are not actually guarantees (and that's where my overstatement occurred).  Does anyone actually believe these projections?  The safety guys do, but I don't.  What I *do* believe, though, is that this methodology has proven to provide a level of safety that stakeholders are comfortable with. 

This quantitative process is totally absent in the experimental world.  If it were followed, you'd probably find that most airplanes are just as safe as the certifed ones in many areas, safer in rare instances, and worse in other areas.  But blanket statements like "all homebuilts are safer than certified airplanes" and "the certification process adds no safety -- it is purely bureaucratic BS" are based in ignorance.

And no, I don't suggest that the FAA mandate that this process be followed by amateurs (as in "amateur-built" -- people like you and me). 

- Rob Wolf