While this method is quite slick, it won't work as expected with numbers that are of the form 0.5, as it will always round up to the next number instead of rounding to the nearest even number (a rule which tries to make sure that most rounding errors cancel each other).

Update: nuance has updated the above post, so this post has become meaningless ;).

Here is an example for a case when always rounding up and using the results in further calculation. Adam calls this "premature rounding", but I think there are valid cases to do this, for example when doing accounting or printing invoices. In this example, using the rounded values in further calculations increases the error introduced by rounding :
Consider (for simplicity) 2.5 * 3.5 :

Method	Result	Error
exact	8.75	0
round up	12.00	3.25
round to nearest even number	8.00	0.75

In this (extreme) example, the rounding errors don't cancel each other out when always rounding up. There are also cases where rounding up and rounding to the nearest even number give the same rounding error (consider 3.5 * 1.5).

And by the way, I'm not pedantic, I'm just trying to be accurate :)