In article ,
Derek Lyons wrote:
software tools are in error (which is not infrequent, it's all proprietary
code that you are not allowed to see, examine, and verify).

I'm not certain that Open Source code would change the situation any.
I don't think an engineer is going to open the code to verify...

You might be surprised. Not every engineer using it, certainly, and not a
complete audit, but it's not unlikely that occasionally somebody would
spend a while digging into the code on some specific issue.

so who does the verification?

Don't overlook the possibility of having audits done by researchers or
industry consortia.

Certainly some efforts at verification would seem in order. Les Hatton's
experiments about a decade ago, comparing the outputs of "production"
software packages used for seismic data processing, revealed barely one
digit of agreement between packages (supposedly implementing the same
well-known algorithms) whose outputs were routinely trusted to three or
four digits in choosing oil-drilling sites.
--
MOST launched 30 June; science observations running | Henry Spencer
since Oct; first surprises seen; papers pending. |