The way I change cars, 3 years sounds like a second compared to a year.
In any case, reliability is a very complex thing. As far as the scientific aspect (not the "statistical") is concerned, the last few days I have been digging into this (more precisely incorporating a reliability analysis in the design phase of a structure, as opposed to just follow a set of given design rules and call it a day) and I am stunned with the level of maths and statistical analyses involved in this. It's not a joke, or something to take lightly.
In any case, we should consider what exactly is what breaks down, or is considered a failure and at which mileage. For instance, most parts of a car have a finite life and will fail at some point. You can't expect a light bulb to last an eternity, but then again, an AC compressor shouldn't blow up after just two years. An alternator can't call it a day before 300.000km, and I'd be stunned to see one reaching the 400.000km mark. Two days ago the fuel pump of my Clio decided to leave this hopeless world and my mechanic didn't believe it was working for 350.000km faultlessly! Such incidents are normal. Also, this 25 years old piece of french automotive machinery has absolutely no signs of rust, yet it lived all its life right next to the sea.
On the other hand, the leak of the clutch bearing of my 150.000km old S60 is not normal.
At least in my field,
reliability is a very specific term and I try not to associate it with the levels of hassle a certain car causes me. Most, if not all, of my Clio's current issues either arrived right on time, or were caused by me.