Forgot your password?
typodupeerror
This discussion was created for logged-in users only, but now has been archived. No new comments can be posted.

Excel Math Bugs Continue Unnoticed

Comments Filter:
  • So, I guess this is now a possible snag point from which people can dispute things on spreadsheets..which they already do but this time they might have some truth to their arguments. And, of course, they want to stall you so they don't have to pay.

    "This table here shows the invoices that were billed to your company, and here's the total at the bottom."

    "But Excel is full of math bugs. How do *I* know these numbers are right? I'm not paying."

  • That's shameful. Has anyone tried to reproduce it on older versions of Excel?
    • Nothing "shameful" about it. Just the way computers store numbers. You'll need to read the entire page here: http://www.dfanning.com/math_tips/sky_is_falling.html [dfanning.com] and follow the links. And then (and that's the hard part) rethink. IEEE 754 yields about 14 or 15 digits on a 64bit double float, which means 9.999999999999999 and 10.0000000000000 are exactly the same number. The bits in your computer representing the one will be the same zeros and ones in the same order as the ones that represent the other.
  • In order to completely avoid results like this, software would have to use an infinite number of bits to store floating point numbers. That is not possible, and therefore we have to live with the occasional 10 showing up as 9.99999999999999. If 9.99999999999999 is not close enough to 10 for your purposes, use a software package that uses more bits than Excel does to store floating point values.

Computers will not be perfected until they can compute how much more than the estimate the job will cost.

Working...