"two processors and memory on a backplate"
Are there any Transputer/Occam experts reading this? It sounds like there are jobs for you at Microsoft in Cambridge, you can get paid while re-inventing your wheels (so long as you let them think they only just thought of it).
Or maybe they could use massive-SMP or clustering or whatever expertise from the worlds of transaction processing and scientific computing.
In the real world, it's not just this year that "computers are beginning to look very different from two processors and memory on a backplate", but it seems that's the way MS see it.
Are there any popular widely used reliable (and ideally low cost, maybe even with source) OSes and/or application Platforms that are already in use in this market segment that might perhaps get in the way of MS's next attempt at world domination?
How is Windows HPC Server  doing, in the markets where engineers and scientists (rather than utterly clueless PHBs and MBAs) still make purchasing decisions ?
Shame research.microsoft.com is still unresponsive to me, just like it was after the last MS Research article a few days ago. It's not totally dead though, validator.w3.org finds plenty of warnings and errors so it's sort of alive(ish). netmonitor.org, where are you now?