The Size of Your Software

By Carol Dekkers

January 10, 2005

(The Size of Your Software :  Page 1 of 1 )

Software project sizing is one of the most challenging and elusive aspects of predicting software project cost, schedule and duration. While traditional product-centric industries such as construction can use quantifiable measures, such as square feet, as a basis for project size, the software industry has always been challenged to find a way to objectively and independently size its projects. Software is abstract: How do you factor in differences such as programming language, complexity, and difficulty to build?

Traditionally, project size was extrapolated using a physical measure of the number of lines of software code, called SLOC or "source lines of (computer) code." The problem with SLOC, however, is that counting them is not standardized, and when more than one programming language is used, the SLOC count becomes inconsistent. Some organizations with a homogeneous, repetitive ("cookie-cutter") software development environment have succeeded using SLOC counts—but run into similar problems when the programming language or environmental factors change. For example, it may take 500 SLOC of COBOL code to program a particular function while in Java the same function might require only 25 SLOC.

In the late 1970s, facing challenges of rising software development costs and increased demands for fixed price software contracts, Allan Albrecht of IBM developed the first "Function Point" software sizing method. This original method remains the basis of the technique now widely known as the International Function Point Users Group (IFPUG) Function Point Analysis (FPA), and standardized as an ISO standard: ISO/IEC 20926:2003. Function Point Analysis was first presented publicly at an IBM Guide/Share conference in 1979, and within five years so many firms had embraced the method that Albrecht and IBM published the first solid rules for Function Point Sizing. By 1986, the International Function Point Users Group (IFPUG) was incorporated as a not-for-profit users group and close to 20 major companies in the US and Canada were willing to invest time and money to evolve and standardize Function Point Analysis.

 Page 1 of 1