It works.... but with difficulty...
Perhaps, but it does work.
and the high temps end up resulting in shortened component lifetimes.
Alledgedly. There hasn't been (to my knowledge) a scientific study that would prove that passively cooled computers have higher failure rates.
It also depends. A passively cooled Atom (or now Zacato or whatever) system won't run that hot. There are also many inefficiently cooled systems, that have a number of fans, yet single components still run entirely too hot.
If you have a Clarkdale/Sandy Bridge CPU, rely on the integrated graphics, put a Thermalright HR-02 on it and use a Seasonic X-400 (for example), I don't see why this system shouldn't be cooled adequately and how this would be any less upgrade-friendly then a fanned setup.
There have been quite a few users over the years right here on SPCR who operated passive setups with great success. Apple once sold the G4 Cube, which operated passively. I'm not aware that it had significantly higher failure rates than the rest of their line-up.
So absolute statements like "passive won't work" or "you need at least one fan" are definitely wrong.