In article <email@example.com>, Bill Rowe <firstname.lastname@example.org> wrote:
> On 2/3/13 at 8:22 PM, email@example.com (Matthias Bode) wrote: > > >The fact that WRI does not even "recommend the use of the Product" > >in instances where it could "threaten" ... "injury, or significant > >loss" does indeed constitute a most serious limitation to "the > >Product's" usefulness. > > Why do you reach the conclusion of "serious limitation"? All > that is really happening here is Wolfram is essentially > transferring legal responsibility for problems to the user. Not > any different than is typical of software developers. > > I don't think you can find any software with comparable > complexity/power to Mathematica that is bug free despite best > effort/intention of the software developer/programmer. Given > that, why would any software developer want to be held legally > responsible for damage etc caused by a bug he failed to find. > > Expecting Wolfram to willingly accept legal responsibility for > damages due to bugs in Mathematica is simply unrealistic. And it > is equally unrealistic to expect a developer of any similar > software to take willingly legal responsibility for damage > caused by bugs.
Exactly. For a view of what it takes to make software almost bug-free, look at the DO178 certification process, which is used for software that controls airplanes - it the software fails, you lose the plane.
As one would expect, DO178 is *very* heavy, causing at least a factor of ten cost increase. Or one hundred. If anything as complex as Mathematica could be certified at all - the scaling is far from linear.