[carapace]

Essentially what you're arguing for is the desirability of untrustworthy software.

Typically people have insisted that it's too expensive to prove software correct, but as the machines and techniques improve proven-correct software becomes cheaper and cheaper.

The last argument standing is that it's unpopular.

[Pet_Ant]

On my current project the way totals and taxes are applied different on two different screens leading to slightly different totals. No one knows which one is actually correct or used by the customers in practice. Thus, you could say there is no correct answer.

Most software development is wrestling with malleable requirements.

As the old joke goes: writing software from requirements is like walking on water, both are easy when frozen

[carapace]

> No one knows which one is actually correct or used by the customers in practice.

Automating BS is bad.

> Thus, you could say there is no correct answer.

I object to such fatalism.

Kind of the whole point of computers is to find and eliminate such ambiguities.

I don't mind playing irrational games for entertainment, but they are "no basis for a system of government", eh?

> Most software development is wrestling with malleable requirements.

Sure but that's no excuse for automating irrational systems. There's no written requirement that the math to compute totals and taxes shall be mysterious?