It's a feature of any application that uses IEEE double precision
floating point math (e.g., every commercial spreadsheet I know of).

It's the result of having finite precision representation of numbers.
Just as 3 * 1/3 = 3 * 0.3333... will be less than one for ANY finite
number of decimal digits, so will most numbers not be exactly
represented in binary.

Using parens changes the order of calculation, so it's not unusual that
there would be some rounding errors causing a very small remainder.

You can compensate that by using ROUND(), or for setting an error limit,
e.g.:

   =ROUND(A1-B1,10) = 0

or
   =(A1-B1)<10^-12

both of which will return TRUE for your conditions.

See

  http://cpearson.com/excel/rounding.htm

for more.

Please do not separately post the same question to different newsgroups.

As respondants in both threads have pointed out, this is a well known
property of finite precision arithmetic exacerbated by binary representation
of numbes.  It is not a bug and is not unique to Excel.

The blogger that you cited contributed to his own confusion by using fixed
decimal formatting, which obscured what was really going on.  His examples
can be simpified to
 =115-113.2-1.8
returns -2.88657986402541E-15 instead of zero, and
 =215-213.2-1.8
returns 1.13242748511766E-14 instead of zero

In decimal, these expresions would return zero, but most terminating decimal
fractions are non-terminating binary fractions that can only be approximated,
just as 1/3 can only be approximated as decimal fractions.  What is happening
in these examples is comparable to doing 1-1/3-1/3-1/3 as decimal fractions
and not getting zero
  1.0000
 -0.3333
 -0.3333
 -0.3333
 -------
  0.0001
i.e. when you do math with approximate inputs, then it should be no surprise
when the result is also an approximation.

Specifically, the IEEE binary approximations to 213.2, 112.2, and 1.8 have
decimal values of
 213.19999999999998863131622783839702606201171875
 112.2000000000000028421709430404007434844970703125
 1.8000000000000000444089209850062616169452667236328125
If you do the math, you will find that Excel's answers to the example
calculations are correct, given these initial approximations.

For a simpler rule of thumb, use Excel's specifications documented in Help,
where it states that Excel only supports 15 digit math.  Thus you can think
of the calculations as
  115.000000000000???
 -113.200000000000???
   -1.80000000000000?
 --------------------
    0.000000000000???
and
  215.000000000000???
 -213.200000000000???
   -1.80000000000000?
 --------------------
    0.000000000000???
which are consistent with the results that Excel gave.

Jerry