Even if we knew what an acceptable rate of failure was, there's no way to determine what the current rate of failure is. The sample size on this forum is small and people tend to only post online when they have an issue. Trying to extrapolate failure rate based on a forum thread where the subject line is "rubish product design" is not a good way to go about it.
Actually, this is incorrect; there is one fundamental incorrect driving assumption, that there is an acceptable failure rate in this case, backed by two irrelevant premises: A, that there is no way to determine the current rate of failure and B, that it may be a poor methodology to deduce it from the forum. Here is the short rebuttal:
A- Easy to determine, from the # of Shimano issued replacements. That Shimano does not publish such a number does not mean their warranty department, or Global HQ, cannot obtain data- whether 1, 3, 10 or 0.15%.
B- Extrapolating failure rate from forum is indeed impossible, but this is also backing an incorrect assumption and preset conclusion. Forum extrapolation is not necessary; the forum has already demonstrated a real serious issue.
Unlike industrial tolerances, there is no acceptable, preventable critical failure rate in the automotive, aeronautic or road/highway safety sector
. The accepted failure rate for a road, maritime or aerospace vehicle is zero if it endangers human safety, if it is known, if it is a design issue, and IF it can be corrected. GM had 13 fatalities (out of millions of vehicles- or 0.0013% for one million) caused by faulty steering or switch, and once it was proven the failure was known, rest is history, US Govt, consumers, litigation. Same for the aeronautical industry. Tolerance for a known failure that can result in an aerospace or road crash is ZERO- certainly in North America. No part known to potentially give away, fatigue, snap or break can be used in any commercial or recreational vehicle if it may lead to road, maritime, aerospace safety hazards.
My point is that any cassette failure, resulting in a fall or a traffic accident, rider injury or worse, may have (‘has’ is reserved for actual regulators) failed the acceptable regulatory threshold.
Why I think this forum is invaluable, and actual numbers do not matter. We are not discussing hard drive or smartphone failure, nor are we discussing the freak hazard bike frame crack, but critical component and road safety at speeds of 30-70 km per hour. The chain drops I experienced, and I avoided stressing the carbon to dodge a cassette crack, made me temporarily loose control of my direction(1-2 m at speed) luckily, no car was next to me.
The good news is that Shimano adapts its designs rapidly, and am very confident these issues will be resolved. 1.5-2 summers of DA feedback (how many people do 8000 kms per summer) is perhaps reasonable.