Caught me in a more than usually argumentative mood:

Ah Ha! a topic for next week

I think putting all the wood behind one arrow makes strategic sense for a company, but not for a community.

So now you've inspired me to argue (next week some time?) that fragmentation is actually good for innovation - and if you'd like to argue the contrary.. my blog is your blog etc etc. ?

There are two main claims here: that users need to pick winners, and that market consolidation is a good thing.

On the global economic level the history on this is that governments have never successfully picked or created economic winners - and that monopoly control leads to travesties like Canadian Medicare under which those who can afford it pay our exorbitant taxes - and then go to the U.S. when they need actual medical help.

In the IT context we have the classic exemplum horribilis of market consolidation: the Microsoft desktop - a product featuring lots of paid media (and uninformed enthusiast) support praising its power and innovation but an actual history of predatory practices, almost total stagnation, ever increasing costs, poor scaling, and low reliability.

And on the Unix side we have the opposite: Solaris setting new records for both scalability and reliability while adapting the "out of the box" Unix ideas from Plan9; the BSDs driving single box OS research; Linux making enormous strides in commercial x86; and Apple's market leading desktop technologies.

Value perceptions about this vary, of course, with your point of view: if you're an IBM or Microsoft shareholder, predatory pricing looks like a good thing - but if you're being asked to pay IBM $23,000 for an activation code on a CPU core you've already paid them $11,500 for, the fact that you can get a faster system from Sun for less makes that kind of thing rather less attractive.

What's going on is simple: Unix diversity and change demonstrates that competition drives product improvement while Wintel stagnation drives home the point by demonstrating the opposite - and if you don't think SunOS improved HP-UX you didn't use Domain/OS; if you disagree that Linux replicates a lot of stuff from System VR3, you've been drinking the groklaw Koolaid too long; and if you'd like to argue for Wintel innovation I'd suggest you compare all the wonderful things you can do on tomorrow's i7 combo to what Mac users could do in the mid to late nineties without invoking things (like most of the internet and lots of software) that didn't exist then and got started on Unix.

The bottom line on competition is just dead simple: when everybody tries to beat everybody else, the survivors and new entrants get better - and that has an important corollary: in a diverse, competitive market buyers can expect that the next generation will be an improvement on the current one.

Once you assume that anything you buy now will be improved on tomorrow, it's obvious that picking winners is silly: what you need to pick is something that works now, has acceptable costs, and doesn't impose significant downstream barriers to change.

Pick a "winner" -like Microsoft- and you need to ask yourself what they won: because the answer is your money and your freedom to adopt alternative technologies. Pick, instead, a competitive market product - something built to open standards - and your future cost of change becomes negligible as a determinant for present decisions.

When that happens you can make today's decision on cost and suitability to purpose in the full knowledge that when something better, cheaper, or otherwise generally more wonderful, comes along, you'll be able to use it.

Focus does, of course, make perfect sense for vendors: when Sun put all of its wood behind the SPARC arrow, Sun prospered; and, conversely, HP's current confusion about whether they want to be IBM, Dell, or NCR is contributing to their decline as focused elements from what's left of Compaq eat them out from inside. Basically corporate focus sharpens corporate competitiveness - and while competition can be painful for individuals, it's good for both the companies and their customers.

Stagnation - isn't. Thus it makes no sense for the Unix customer community to want homogeneity because that's the route to stagnation - to the Windows and mainframe worlds where there are no serious competitive checks on the seller's power and a touch of some other company's lipstick on the product front passes for fundamental change.

And that, fundamentally, is what's wrong about Roger's position: he claims that the Unix wars kept Unix from obtaining a market monopoly or near monopoly like Windows - and my response is to thank God it did because otherwise we'd all still be using Xenix.

Think of it this way: whenever government or other predators pick and enforce winners, stagnation sets in and product values trend down - but whenever open competition encourages diversity and change, product value averages trend up.