We know that there are many species of software - whenever a branch is made in the development, we get (eventually) specification (or is that a specious argument) - but the real question is: is this by design, or is it just random? Does the Xen Master play Dice? Is the Flying Spaghetti Monster Code better, or worse, than IBM's 10 lines a day of commands meant?
meanwhile, also, what is better or worse, intelligence (see the short story by Bruce Sterling about the short term survival value of intelligence) or stupidity?
virtualisation is a way to avoid making decisions (the ultimate random branch, if you like) since you just continue to run all old and new versions....however adapted they are (indeed, if enough of you do this, you effectively run all the eco-systems in history - i.e. world 0.1 alpha thru world 2.0 beta, in parallel universes (if oprah is anything to go by, at least as far as the Onion claimed today...)
so its clear that the virtualization thing is virtually stupid - it allows multiple isolated types of stupidity to co-exist, without learning from one another. This is a good thing (spreading stupidity is a Bad Idea) and a bad thing (clueful people would find easier to recognize all the stupidity if it was all isolated in one place rather than N places). Also, with remus, stupidity is now hilghly mobile and hard to eliminate with Denial of Stupidity attacks. Indeed, SANs (Stupidity Area Networks) with virtualisation mean that stupidity will come to dominate the cybersphere, if it hasn't done already. This, unfortunately is in line with many grand challenges such as Sustainable Stupidity, and Curatable Stupidity.
I am thinking of starting a project on CAS (no not compare-and-swap - Clutching At Straws) - campaign against stupidity
but I suspect I would just get stuck in dom0 an ignored by all the Guest Operating Stupids