Interesting read, looking at the 40d,50d,550d models it seems that if your shutter survives longer than the estimated rating (100k), it is likely to survive even much, much longer. The big question would be why - either these people got "lucky" and their hardware is better, or they shoot at lower shutter speeds or whatnot and there is a cause and effect.
That's pretty much standard for reliability. A item that has already exceeded the average life is likely to keep on working much longer. It usually means that the fit and alignment of the many small parts and bushings is ideal and can go on working.
What causes failure due to wear out is the tiny misalignment of parts so that motor bearings wear out, or other moving parts wear out. If alignment is perfect, the life can be many times the average.
R&M engineers predict life of a assembly using a manual that lists reliability of each component based on known information, or calculated information. Very seldom do they actually spend $500K to test a part. For high reliability parts, there is accelerated life testing involved, but it comes down to extrapolation. Its impossible to test rubber or plastic for a lifetime, or a motor bearing, they should last for 10 years - 1000 years. That's why they are tested with high and low temperatures, under vibration and shock, etc. In the end, life is a guess. In the case of camera shutters, the design from one camera to another is either similar or identical, so the probable reliability numbers are well known. They can construct a bathtub curve for a typical item.
The Japanese and many US industries have adopted a 6 sigma reliability process, I had to impose this on subcontractors for new components, and they resisted at first, but later found it paid back in fewer rejects and failures of expensive components. Doing it for a part that is so dimensionally accurate that it can't be directly measured is difficult.