The old argument about which is cheapest, Linux or Windows, and which is best for the enterprise, refuses to go away. In one camp live those who are comfortable with a commercial licence which has its own advantages and disadvantages, while in the other sit open source proponents, who argue that the best products arrive via a collaborative route.

Both however are starting to agree that the parameters of the debate moved on from straightforward cost of acquisition to the total cost of ownership sometime ago, as reflected by the popularity of Techworld's own white papers on this issue.

What's happening now is that the debate has moved on one more notch, and is starting to show signs of maturity.

Microsoft fan recants?
One exemplar is a recent column by Yankee Group analyst Laura DiDio, usually seen as belonging to the pro-Microsoft camp -- she has even been described as a Microsoft spokesperson. In the column, DiDio argues that the Windows v Linux debate is "not an either-or, all-or-nothing proposition" -- an uncontroversial position to be sure.

She has however in the past allegedly made disparaging comments about the viability of the open source business model and how it's run by hippies. Yet what has changed in recent months is that even Microsoft accepts that the open source model has commercial legs. It has responded to the in-roads being made by open source products in a highly aggressive fashion, as DiDio points out. It's even publicly toyed with the open source idea.

Meanwhile, most major open source players have acquired the trappings of the standard IT company, including a CEO, marketing managers, PR companies and the like. I'm just not talking about the companies that came from the grubby world of moneymaking either -- those such as Novell and IBM. The description also fits Linus Torwalds' Open Software Development Labs (OSDL) and Red Hat, which arrived at a similar position from the bottom up, as it were.

They also, of course sell things, though the purer ones sell only services rather than products. In effect, they have moved in the direction of Microsoft's business model. However, the all-important (to developers) means of production differ from those practised at Redmond. Instead, as IBM's vice president of technical strategy and innovation Irving Wladawsky-Berger points out in his blog, "open source is all about collaborative innovation, that is, working with smart people all over the world as a community to solve important problems."

It works too. IBM, now a fervent advocate of open source, is not known for following blind alleys to the exclusion of commercial viability. Recent examples where it has turned tail and run in face of huge potential losses include its PC hardware business, and OS/2. Though prepared to experiment, for IBM open source remains centre-stage. It's a big vote of confidence.

So the key difference between open source and commercial products is not that the software is free -- very little is really free, as Windows' advocates have gone to considerable lengths to remind us -- but that the method of production and therefore product quality are better.

TCO -- a blind alley?
But the means of production are not top of mind from an IT manager's point of view. They're interested in practical matters: getting more for less, technical support, and especially lower TCO, for instance.

DiDio's column makes this point -- kind of. She argues, as many others have, that the TCO argument is not cut and dried, but depends on your situation and the needs of your business. TCO is also very hard if not impossible to calculate, since the exact state of every single asset and the cost of getting it to its current state, and the exact cost of maintaining that state is unlikely ever to be known with much precision.

This is a position that I, along with other writers on a large but now-defunct IT publication, adopted when Gartner first publicised its invention of the TCO phenomenon over ten years ago. Little has altered since then to make me change my mind. And the recent Yankee Group report which sparked DiDio's column also supports this position.

The report found that, even after a thorough TCO analysis, 75 per cent of enterprises "could not answer explicit questions about their own environments". Respondents knew neither how much their IT kit cost nor how much they were spending to keep it working, in other words, let alone everything they owned.

So TCO is a bit of a chimera. Instead, in the real world, technology choices usually boil down to what works with the existing setup -- interoperability and integration, in other words. Linux, with its multiple distributions and functionally overlapping applications, each with its own dependencies, still has some way to go to ameliorate that problem. Despite that, as DiDio points out, Linux servers "are legitimate contenders in the corporate enterprise". The debate has grown up.

We've come a long way since the invention of Linux; it's taken longer than many observers imagined it might to gain this kind of acceptance. But despite the scepticism, the disparagers would, in my view, be doing themselves a service by declaring this degree of acceptance a minor milestone.

And the winners here are IT managers, who are likely to find their needs addressed in a more focused manner than before.