Is this claim* correct? “Function point metrics are the most accurate and effective metrics yet developed for performing software economic studies, quality studies, and value analysis.”
No. The claim cannot be correct because no-one can prove whether any software size metric is ‘the most accurate’ metric. All functional size metrics are based on assumptions about what might be a plausible size scale, so all scales are arbitrary. There is no physical object like the meter-long platinum bar against which the accuracy of various measurement methods may be compared.
As to FPA, this method relies on identifying components of the software to be measured and classifying them as simple, average or complex. Three possible sizes, that’s all. The most primitive counting system known to man is that of the Hottentots, a nomadic tribe in South-west Africa. Their counting system goes ‘one’, ‘two’ and ‘many’. If a Hottentot has three or more cows, he can only say he has ‘many’ cows. The FPA sizing metric is just as primitive. No matter how big a component, it can only be ‘complex’ (e.g. 6 FP for an External Input), no bigger. It would be more reasonable to claim that FPA is one of the least accurate sizing methods.
(As an aside, it could be argued that counts of Source Lines of Code (SLOC) are ‘the most accurate’ method of measuring a size of software since, provided you define your counting conventions, counting can be 100% accurate, and even automated. And a SLOC size represents all the requirements, both functional and non-functional. But we all know that there is no agreement on SLOC counting conventions and that sizes measured by counts of SLOC have all sorts of other problems, so are not very useful.)
Posted in: The COSMIC method