IBM's mini computerslack thereof 832

Yup. It's the only reason I haven't given up yet. But we do need more mature OS types spending more time thinking. Before anybody blasts me, I am talking about organized, long-term, collective thinking where the concentration is on innovation rather than office politics. AFAICT, this doesn't exist.

That well-documented is all that is necessary. If you spec something and make it freely available, people will pick it up because it is work already done.

GIGO is usually cheap up front, but extraordinarily expensive over the life-time of the crap. If I were a god, I'd figure out a way to objectively measure user thruput from the POV of the user, not the damned engineers who wrote the specs on another system that gave them adequate user thruput.

Oh, cool. Is it possible that I-O is catching up to CPU speeds?

Yea, but you cannot design the protocol to be as complicated as a DECnet layer. It also has to be extensible, backwards compatible (which is definitely not the flavor of the month) and fixable.

I guess in my fairy tale world, I'm thinking of a (heh) multi-CPU IAS system. Just enough of a user interface for somebody to debug it, tweak it, and watch it. Stripping down TOPS-10 would be extremely difficult because servicing users was the goal.

Sigh! Note that I am not dismissing Unixy stuff because I think it's bad. I cannot consider it because I don't know it well enough.

BAH

Subtract a hundred and four for e-mail.