Most of us don’t want to be famous, even if it brings great wealth. We want to be admired. Being admired is not the same as having strangers hate you just because you were on TV, or because you wrote a piece of software that made some money.

Real fame is having people write computer viruses specifically targeting your company. Against this sort of customised attack, it’s hard to see what anybody can do. Anti-virus and anti-spyware software is generic, to counter mass-market attacks.

But suppose you work for a merchant banking group and one day you find all your systems crashing. The only clue is an anonymous email saying ‘Ha ha. That will teach you to fire me!’, and you realise a disgruntled ex-employee has written a virus specifically designed to use inside knowledge of your network to bring it down.

That’s fame. It isn’t good. But it is what you need to be a successful giant business; which means you’re making yourself into a target as surely as if you painted concentric circles on your face.

It is also a direction in which the anti-virus community fears we’re headed. What the PC community can do about this isn’t obvious to me. I’ve endured many presentations from Microsoft about how it is making the PC more secure, and I suspect we aren’t talking the same language here.

From the perspective of security experts, there are choices. First, you must have an inherently safe environment. When you have that, you must support it by writing inherently safe code. Finally, you have to use the computer in a safe manner.

I had a chat with the guys from Fortify Software, who don’t believe an impenetrable firewall is the answer.

Rather, they believe software-development tools have to produce code that is pre-verified against common errors such as buffer over-runs as part of the testing process. I expect to hear a lot more from them over the next year or two.

But their point was a good one, generally. Put simply, the more secure your environment, the more careless you are likely to be.

If there are no cars, why look before crossing the road? If the network is controlled by foolproof intrusion-prevention technology, are you likely to trust an executable piece of code you find on your hard disk? Or are you more likely to be cautious if it’s a machine used on the Internet by a 10-year-old?

I think the problems require a computing environment where it isn’t possible to bring down the network by running trojan code. That means, to my mind, that the environment has to allow user stupidity, but isolate the user’s own private environment from the main system.

The question that then arises is the old, sad, Microsoft-bashing one. Can this honestly be done with a Windows system focused entirely on digital rights management and backwards compatibility with the IBM PC Bios from 1981?

I’m working on a story relating to what I think is a scandal in local government computing. In essence, it involves business practices by Microsoft agents (I can’t yet tell if Microsoft is even aware of the practice) that amount to ‘buying the business’ in order to create a publicity firewall.

This is a system where people write ‘security analysis’ white papers about their proposed new system, and base those white papers on unsupported assertions from Microsoft. They publish the white papers as ‘our research’ and then Microsoft quotes this ‘independent assessment’ as proof of its suitability for a high-security environment.

The way to achieve security isn’t by publishing corporate flannel as independent research. It is (obviously) the view of my Linux friends that a proper operating environment is the only way forward. It is the view of my Windows friends that there’s no way of replacing the universe of Microsoft-based software, or of making the universe of Windows users into people who understand Unix.

Perhaps the solution is to create a virtual environment, with Unix as the host, in which protected Windows environments can be set up and run with limited privileges.

It must be possible, with virtualisation technology, to run something like that on any new dual-core PC, with a little extra system memory. And if it is, maybe we can move forward into a world where computers actually work without causing more problems than they solve.