Theoretical Count of Function Points for Non-Measurable Items

Introduction

The function point analysis (FPA) is a standardized technique for measuring software development, aiming to establish a gauge of the software size based on the functionalities to be implemented, considering the viewpoint of the user (IFPUG, 2000). Some usage thoughts have been made to the extent that software systems are becoming more complex (Micro Focus, 2008), such as:

• 1.

  “Function points are not a very good measure when sizing maintenance efforts (fixing problems) or when trying to understand the performance issues. Much of the effort associated with fixing problems (production fixes) is due to trying to resolve and understand the problem (detective work)”;

• 2.

  “FP analysis (FPA) is not useful to size Web Design. FPA is useful to size web development, but not web design. [...] FPA is not useful in estimating the time necessary to create graphics, images, page layouts, so on and so forth”.

During the eighties and until the beginning of the 21st century, a number of authors have discussed the metric procedures in vogue (Albrecht, 1983; Dreger, 1989), analyzing its applicability to object-oriented software (Whitmire, 1993), its advantages and disadvantages (Jones, 1994). Also, Kemerer (1993) studied the reliability of the FPA technique. Important contributions to understanding the complexity inherent to software engineering were brought by Indian school (Jalote, 1998; Ram et al, 2000). More enthusiastic works about FPA appeared since 2000 (Garmus & Herron, 2001).

The function points measurement technique has generated much controversy since its dissemination as an ISO recognized tool to size information systems, both as regards its general purposes (does it measure productivity, size, complexity or functionality?) and in relation to mathematical rigor under the concept of metric. In particular, with respect to the latter, being the number of function points a dimensionless quantity, some authors claim that there was no way to analyze and seek information from numbers not associated with a reference system (Abran & Robillard, 1994). This is not entirely true. In science there are many dimensionless numbers widely applied in several fields as hydrodynamics, geophysics, optics and others. A dimensionless metric, being independent of the reality beneath evaluation, is useful to compare two or more objects abstracting a lot of details of these objects, placing them in the same plane of observation and providing a perspective that would be inconceivable without a standardized approach. Perhaps the difficulty is to precise the mathematical structure and the semantics of the metric, that is, what the metric formally measures.

One of the major contractual problems faced both in the governmental sphere and in the context of private enterprise is the remuneration of the activities not measurable by the technique of function point analysis. Some devices have been adopted, but with high degree of arbitrariness, making the calculation uncertain and often unfair, and vulnerable to critical assessments of the organs of control. In addition, an arbitrary control of the estimates may be evidence of unprofessional management. Also, something more is missed out as observed by Lokan (2008):

The present model, although it stemmed from the need to measure the development effort of sites and portals, aims to incorporate not just the typical non-measurable items but all function-strong software processes, including operational system migrations.