A lawsuit filed against Microsoft in Los Angeles this week is attempting to hold the company responsible for the damage wrought by the systemic failures of security in its sofware, and for its conspicuous failure to fix them adequately. The suit follows hard on the heels of the publication of a paper on Microsoft security which, among other things, suggested that the company should be held legally responsible for such damages.
The Los Angeles case is being brought by one woman who says she was the victim of identity theft, but it is designed to form the basis of a class action. It argues that the vast majority of successful attacks occur because of major vulnerabilities in Microsoft's software and - crucially - claims that the disclaimers in Microsoft's licence agreements constitute an unfair business practice under Californian law, because consumers have little choice but to use Microsoft software.
Licence agreements have traditionally operated as a catch-all 'get out of jail' for all software companies, so the argument of this suit can be seen as quite narrow, success depending on establishing the existence of a Microsoft monopoly, rather than on exposing licence agreements as the outrage (you can tell we're disappointed, can't you?) they are. Microsoft's recent practice of stopping you fixing the software you've got if you refuse the new, even more horrid agreement could well provide some support for this suit's argument.
Aside from being able to argue that the licence agreement means it doesn't have to do anything about its software being broken (which from a marketing perspective is probably not an ideal first line of defence), Microsoft is currently attempting to shift the blame ground away to the "criminals" out there. Pitching the new-look line on security recently, Steve Ballmer denounced all hackers as criminals, and signalled a move away from the 'patch and patch again' approach to 'securing the perimeter.'
The patch regime certainly doesn't work, as it's just plain ridiculous to expect all consumers to keep their patches up to date, even if they have the bandwidth (for the record, a Register test of a bringing a virgin Win2k installation up to date as far as last Tuesday, over dial-up, took in excess of three hours, without us bothering about WMP and DirectX updates). Large numbers of unpatched machines out there generate 'cascade failures' which bring a cost (in terms, for example, of bandwidth and crippled email systems) even to patched systems, so you still get hit no matter how good you are.
Securing the perimeter, though, is under the circumstances a dubious alternative for Microsoft to be proposing. If you currently think you have a pretty secure perimeter, we expect you still feverishly patch your Microsoft clients, and there's good reason why you do this. What Microsoft is really saying, we think, is that security should be addressed further out in the network. This is certainly sensible, but doesn't have an immediate effect on where we are today, and how we got there. It does hold out the prospect of the blame being shifted from the vendor of the client software to the vendors of the firewalls, to the ISPs and so on, but we're sure this can't be why Microsoft now thinks it's the way to go.
Given that Microsoft's security pitch is shifting, and it's at least arguable that the company concedes that what it's been doing doesn't work, defending it in court could be particulalry tricky, if the company can't stop it getting as far as that argument.
But this is still a half a loaf kind of a lawsuit. It is consumerist, certainly, but doesn't look likely to test the broader issues of product liability and 'fit for purpose' on the products of the software industry. In order to do this it would be necessary to overthrow licence agreements on more general grounds than are being applied here, but we reckon there's a case, at least under European consumer legislation. Software companies in general have managed to avoid test cases over licence agreements - we think there's a good reason for that. ®