Originally Posted by JimVonBaden
Not really. ANY issue is an issue. Reliability, by the definition I am using, is any issue, whether it strands you or not. Any other definition is a matter of degrees, and therefore subjective and would make this conversation too muddy to pick one over the other.
Your definition is not how Reliability Engineering sees it. Issues vary in their effects and there is a method for accounting for that.
For instance - engine failure in a single engine airplane vs engine failure in a 4 engine airplane. The 4 engine is 4X as likely to have a failure but the consequences of the failure are likely to be much less.
There is a Reliability Engineering analysis activity called Failure Modes and Effects Criticality Analysis (FMECA) used to work out the issues and their consequences. Since it's time consuming it's costly. But it can help change design decisions if the reliability or maintainability goals aren't met. And of course it depends on some assumptions that may be proven wrong in actual service when theory encounters manufacturing defects and production tolerances (" BMW final drives are bulletproof" :-)
But to give a thumbrule of Reliability - The more stuff in the system, the less reliable the system. (You're gonna be doing more maintenance to keep it running 100%.)
Another - the more lines of code in the software, the higher the probability of a defect lurking in the software.
Another - jack with it long enough - you're gonna break it. :-)