<... in my book, redundancy is job one ...>

The engine has demonstrated its ability to get you to the next airport without coolant or oil.  Ignition and fuel are another matter.  Dual EWPs might be necessary for TO and climb and redundant in cruise.  Redundant oil pumps seem to me a whole lot more trouble than they're worth. As for CAS and the like, NAPA, dealerships etc. could provide the least UNreliable available as     to what needs to be backed up and what doesn't.  Might even discover that you need to back up the MAP sensor or something of the like before the CAS on account of they fail all the time.  My point is:  *who knows??*

In my view, Al's discipline and methodology work beautifully in a factory where the same machines are doing the same thing hundreds of thousands of times a day.  It works less well for say, cars because it's so much more difficult to gather accurate data.  How you gonna determine how many of some component failed after how much use?  Dealerships are a help - they track what failed and when - but the independent shops do not and I'm pretty sure they do the great bulk of the repair business.  On top of not knowing with any degree of reliability what failed and when, we are totally clueless as to fleet mileage.  Therefore MTBF is impossible to track unless you assume (remember that acronym) that the fraction of data gleaned from dealerships reflects the entire fleet accurately.  Bit of a stretch IMO.

So here we are examining airplanes.  Homebuilts no less.  How in hell on earth are you gonna' get data around failures?  Al got his little slice off the Canard or Cozy list(s), but it was a *tiny* slice of reality submitted by volunteers.  Just for openers, I'd have to question the *motives* of the guys who participated in his little study - and the guys who *didn't* as well.  Beyond participants' agenda, the small size of the sample is troubling.  If one could identify all of the suppliers of parts and components, one might estimate MTBF of this component relative to that component, but life expectancy in actual hours is still an impossible dream.  Add to that the fact that Lycs on planes that fly all the time (like instruction, air freight, air taxi and the like) almost always make it to TBO while the same engines on privately owned, rarely flown planes almost never do.  Quantify *that* Tarzan!  Does that phenomenon apply to auto conversions too?  Does it matter?  Damned if I know.

I may be mistaken here, but it seems to me that the very soul of Al's methodology is accurate data around the incidence of failures.  He can speak to that better than I can, and I sure hope he does.  Additionally, only he can speak to the degree to which data that is dirty and/or unreliable skews the results.  Basically, it seems to me that Al is trying to quantify something that's qualitatively intuitive to everyone on the list.  We can guess which components need attention and set rough priorities as to which ones to address first.  The main issue is quantifying these occurrences failure.  At this juncture Al simply does not have the data to do this with a lot better reliability than we could do with educated guesses.  He needs a *lot* of *good* data IMO.

I'm wondering if it's possible that he will ever get it ... Jim S.

-------------------------------------------------

Was said ...
I see guys doubling up on water pumps (2 EWP in some cases) yet because there hasn't been a CAS failure (that we know of) they are not worried . If you want true redundancy, you must double up on the systems that have a possibility of killing you engine (and you,no matter how remote) The CAS failure could be between the ECU & the CAS due to our application. Not to be too critical, but I've seen some pretty scary (sloppy) installations. If you follow the certificated aircraft procedures as far as securing, protecting & routing wires, hoses, pipes & components you can be on your way to a safe installation.
Georges B. (in my book, redundancy is job one)

and ...

I fail to see how installing another CAS will 'dramatically' reduce
risk of all ECM causes.

 We have already said we have no history of failure of the CAS, how can
installing another CAS (with no history of failure), 'dramatically'
reduce the risk of failure?

And how can installing another CAS have any influence on "the risk of
all ECM causes" ?

I also have great respect for redundant systems, but I cannot see your
logic in this one. It is the 'dramatic reduction' that troubles me.