herbaltea001-winter (4/22/2013)
The report seems to show that none of the popular spreadsheets are up to the job. I doubt if they were ever intended to be used for such critical calculations. This is a little reminiscent of the warning that comes with Java in that it should not be used for aviation or nuclear facilities.
It's a few years out of date so the conclusions might not hold for later excel versions, but it's a question of whether a spreadsheet *should* go to that level of accuracy in such hugely niche cases, particularly when there are programs specifically designed to handle these situations with ease?
Hopefully, what the devs think is something along the lines of:
A spreadsheet is meant to be good at a very broad range of tasks and time should be spent delivering great functionality to >0.1% of users instead of spending time making it correct for the 0.1% who may or may not wake up and realise that they needs a stats package and can get them for free.
I never knew that about java, but it makes sense that a still evolving language can introduce risks into an existing codebase and you certainly wouldn't want it to affect super important things. Of course this is partly why FORTRAN and COBOL are still used in banks :crazy: