It is daft.
It is a nice idea, but software is just too complicated to make it 100% reliable and bug free. If the software had to be 100% reliable, the cost of extra development and testing time would mean we would be looking at several years to market and the costs would go back up to 1980s standards, with operating systems and applications costing 10s of thousands of pounds.
The current project I am working on is developed on WAMP and runs on LAMP, with a different version of PHP. Some bugs appear only on the live server, because they have been fixed in later versions, but our provider won't install a newer version of PHP...
Anyway, the application is around 50,000 lines of code and it still has several bugs, which I work hard to fix, but getting even a simple application like this 100% error free is nearly impossible.
And who do you blame? If an application doesn't work, is that because the application has a bug, or the underlying OS has a bug, or there is an incompatibility between the hardware and either the OS or the application.
Making sure that the developers are responsible for fixing the problems is one thing - look at the patching programmes that most major vendors and many of the larger open source projects put a lot of effort into tracking problems and fixing them. I don't think there is much more they can do, without significantly raising their prices to cover extra development costs and increased legal fees...
