Viable computing

For a long time I have been generally frustrated by the stupidity of PC's (by PC I mean any computer). For most of my life I have been an "apologist", apologizing to non-techies for seeming random and unfathomable reasons why PC's have the unnerving ability to fail at the most critical moment, or in the most devastating ways, or dumbness of hierarchal file systems. Even the computer in our coffee machine at work now fails in its only duty, and on almost daily basis needs to be rebooted.

Ok so what? Well for my cents worth I think its time to fundamentally reevaluate what we do and how we do it. I think that in a simplistic way computing is (if you understand viable systems model) stuck in systems 1, 2 and 3 (i.e. its a ll about operation, management and resource contention) what its missing is system 4 and 5 ( an understanding of the outside world view and
what its identity is) Ok I know this all very AI'e but that's the issue.

So what would practicle befits would "I" derive from a system 4/5 based service? Well some basic thoughts are -
A computer service that "understands" and is far more personalized to fit "me" - from simplistic examples that it understands that when I go to "bed" it stops bleeping/disk sounds/fans etc to fact that I have a number task orientated modes (watching TV,writing software,playing a game.. doing a hobby etc) and that in these modes I need different support and operations and capabilities. Lastly that it "understands" the situation I'm in and reacts accordingly (e.g. so it does what I want/need it to do not what it wants to do) , e.g. when I'm desperately trying to print out an email it doesn't start a virus database update, windows update and a scan of the disk all at the same time (for a long time I have wanted a frustration input device for windows.. where by I could communicate that I'm very unhappy and to stop doing what its doing and instead ask me what I want to do, leaving everything else to later).

Ok all sounds very artistic and airy fairy AI'e and not the preserve of rational business focused computer scientists. But if we are to start "understanding" more about what the user wants and needs (in behavioral management terms we build in huge amounts or R+) and building system's that better meet these then I'm worried that we cannot simple back fit this on to an existing resource management system (i.e. its not a shell to MS Windows XP/Vista etc) the internal communications needed (the feed back loops) and resource allocation algorithms need to change drastically.

si