Mint, maybe it's mucky terminology but I think maybe you're being a bit harsh. Certainly, at least in the US (and in my mind to any sane designer;-)) function should be the primary driver of tolerances, and this will vary by application.
However, you also have to ensure that your tolerances can be achieved at acceptable cost. This is where there are some rules of thumb and in some cases there are indeed tolerance standars such as (the in my opinion flawed) ISO 2768. While I have to say that I think this particular standard may tend to swing too far toward suiting manufacturing at the cost of function, it does exist.
There's a big difference between designing to stay within "standard tolerances" that can be achieved at reasonable cost by almost any competant supplier (obviously looser isn't usually a proplem for manufacturing, I'm talking about tighter) while ensuring function and not considering tolerance at all and just relying on some industry standard, block tolerance or other rule of thumb.
The former should be encouraged, the latter should be outlawed;-).
KENAT,
Have you reminded yourself of faq731-376 recently, or taken a look at posting policies:
What is Engineering anyway: faq1088-1484