Sizing software by counting ‘one, two, many’ – does it matter? (Yes)

Symons, C.R.

Studies have shown that the sizes of software that has relatively simple elementary processes
measured by both the IFPUG and the COSMIC-FFP methods correlate quite well. But as the
average size of processes increases, the fact that the IFPUG measurement scale suddenly stops,
causes serious distortions. We conclude that IFPUG sizes will be seriously misleading when
used for performance measurement or estimating for software with larger, complex transactions.