IMO it all boils down to efficiency...

If a formula can be condensed into one efficient formula then it makes sense to do so.

However, it is often the case that an "elegant" approach to a complex problem will be inefficient and "helper" / intermediate calculation approaches are the better approach long term for the sake of usability.

The above is perhaps most prevalent with multi conditional calculations where people tend to be over reliant on Arrays / SUMPRODUCT ... when such formulas are either a) over used or b) used sparingly but say in a Volatile manner they will quickly impact overall performance and ultimately the viability of the model in which they reside.

As a thread subject - it's pretty broad ... and it all boils down to context, the particulars in each scenario - what holds true in one model might not in another.

If you're interested in optimisation I'd suggest having a read through of Charles Williams' site - he's considered by many to be one of the real experts on this subject.

http://www.decisionmodels.com/calcsecrets.htm

He's also written a white paper for MS re: XL2007

http://msdn.microsoft.com/en-us/library/aa730921.aspx


All that being said there's little value to be had in an elegant model if those using & maintaining it do not understand it... things going wrong is a fact of life and it's important that there are people around able to fix it...