[SystemSafety] What do we know about software reliability?

[Martyn Thomas]

On 16/09/2020 14:10, Lorenzo Strigini wrote:

> The statement "reliability is an attribute of an organisation not a software or systems product" does not seem to be about whether a product is reliable enough, or whether assessing product reliability makes sense. It is about one way that people may guess whether something will be reliable enough when in use: by assessing whether the organisation from which it comes is good. Nothing to object to that.  Yet some customers demand stronger evidence than "the vendor is a good vendor". One form of this extra evidence is statistical evidence about that specific product not failing when exercised in the same conditions (as far as feasible) in which it will be used. 
> (And if the vendor does try to gain that statistical evidence instead of just trusting "I am a good organisation with good processes, so all my products must be good enough", well, that adds to the trust one can have in that vendor).

One might ask "under what constraints will you warrant that your
software will not fail?". If a vendor won't bet their company on their
software, why should you bet yours?

I understand that Altran UK give warranties (and refuse to offer cheaper
prices without the warranty).  I don't know their contractual details.

Martyn