>>Do I think it is possible to produce zero-bug software? Yes, I do,
>
>I disagree, and so does the rest of the professional software testing industry. You can hit a piece of code 10K times, and it may be the 10K+1th time that it fails. You can do code coverage and think that a piece of code is unreachable, but your customer will find a permutation that reachs that code. In a complex system, there are enough possible permutations that you could spend a lifetime testing it.

I told JohnK that I'd happily accept a 99.9999% objective.
But I do feel that, sooner than we think, 100% will be achieved. Sure, 100% is presently felt to be unattainable but that will change. It will be a matter of need (driven by the marketplace) and bucks, but we will get there.
Right now the need is much smaller, and those with the need spend extra dollars to compensate for this deficiency. But you notice that people do not have 2 fridges just in case the other breaks. Or 2 of much else just as backup.

People simply will have too much of their life dependent on properly functioning computer software. They will ask that one tough question - how come, when every single copy is identical, that you can't make sure that it works properly first? They won't care about billions of lines of code.
At least physical products have the excuse that, though patterned identically, each is itself a separate and distinct entity. Imagine if GM said 'yea, we sometimes deliver a car without it's steering wheel (or working brakes or engines or transmissions or whatever) because it was just too complicated to get it there on time that time'.