Is it really "Serious" though? I mean, this isn't exactly devastating. Office 2007 has been out for almost a year and this being the first time we've heard about it seems to imply that not many users have run into this issue. It's poor that this got by their internal testing, but I'm not sure it's Serious.
Well, how many have actually gone undetected? There are quite a few papers out there that show that Excel-induced accounting/financial errors are a lot more common than previously thought. Some financial services have gone as so far to prohibit the use of Excel in their financial planning and forecasting.
"Microsoft in recent weeks approached Facebook with proposals to invest in the startup that could value the fast-growing site at $10 billion or higher, said people familiar with the matter."
Who puts the valuation for Facebook? was it MS or WSJ assumption?
According to later posts on this, many calculations which involve storing 65535 or 100000 really store the other one:
Dana DeLouis writes: """
And if you multiply be say, 2 you get something else:
=A1 MUL 2
returns 131070, as if A1 had 65535. (which it should have been)
=A1 MUL 1
Keeps it at 100000.
=A1-1 returns 65534
=A1/1 is still 100000
=A1/2 retuns 32767.5
"""
Given that it would more often be the case that 65535 or 100000 comes up as an intermediate number than the final result, it could be that there are many calculations that have been done in error silently for those using Excel 2007.
EDIT: Replaced asterisks with MUL because I don't know how to escape them.
Probably not. All I know: using one star on each sides of something for italics, two for bold, and if you put a space at the beginning of a line, you get fixed-width font (for code).
Under infinite precision the probability of selecting this particular expression to evaluate is zero. Under finite precision it is small.
However, it raises the question of what other errors Excel has. Excel has had other serious mathematical errors in the past - like the standard deviation function. In Excel 97, if you wrote down a sequence of large numbers with a standard deviation much less than the mean (1e10, 1e10 + 1, ..., 1e10 + 10, say) it would say the standard deviation is equal to zero!
Furthermore it doesn't help that there's always a total lack of transparency in how Microsoft handles bugs. We have no way of knowing whether Microsoft knows about this, if or when they intend to fix it, and what other problems this could possibly hint it. Megalomaniacal megacorporations are so banal.
Any problem like this could be potentially very serious, when so many people around the world still use Excel for making important business decisions. Not being able to trust your numbers 100% is a big issue to some..
I'd say it's fairly serious, since something so basic doesn't inspire a lot of confidence in Excel. It only takes one bad PR moment like this to tarnish a product's reputation forever.
I wonder if some disgruntled ex-Microsoft employee just took the crown for best Sabotage-By-Easter-Egg ever.
