National University of Singapore School of Computing Software Project Management (CS5212) Critique Paper: Web Development: Estimating Quick-to-market software Donald J. Reifer ieee software, Year 2000 paper

In this article, the author described the difficulties in estimating web development projects in terms of size, cost, effort and duration. He purported that the objective of traditional software projects are driven by cost whereas the principle of web-based projects is time-to-market. Web projects have shorter SDLC as their foundation is built on rapid application development and continuous prototyping. The proliferation of visual programming web development tools quickens the pace of web component development, reusability and assembly of building blocks. Due to the nature of the web development projects, the inherent traditional software projects estimation process, model and matric has become inapplicable and unsuitable.

The author studied 46 web projects and proposed 2 new metrics:

1. 
   Web Objects to estimate size,
   
2. 
   Web Mo model to forecast effort and duration
   

In Table 2 - web-based estimating challenges, we do not agree with the author that the quality estimation for web-based projects is hard to measure as opposed to traditional software projects. The author does not elaborate his definition of quality clearly such as conformance to requirements, user acceptance criteria and so on. There is also certain degree of overlapping with conventional software projects.

In web object predictors, certain operands and operators are disproportionate. For example, interface to databases, extraction, transformation and loading requires significantly more effort than the other. The Web Objects does not consider the weightings of the operators and operands to estimate sizing, effort and duration.

The author acknowledged that the component reusability is important in web objects but failed to incorporate it in his matrics. It is also unclear what constitute reuse and the degree of reuse. The author has identified reusability in Table 9 but has flagged it as “Not Used”. No explanation was provided.

The author claimed that he has achieved accuracy of estimation within 30% range with 60% success rate but does not further elaborate or substantiate his findings/

The author does not consider the effort and complexity of different programming languages. He assumes that they are identical.

The author should consider comparing WebMo model against COCOMO model to prove his case of accuracy.

The author should also consider the extent of reusability in web development projects and how it contributes to difficulty in estimation.

There is an error in Table 6, the acronym to represent cost driver - Product reliability and complexity is used inconsistently. “CPLX” should be read as “RCPX” instead.

This paper deserves a rating of 6.5 out of 10.

This article is a short article with 8 pages. The introduction and the comparison of traditional software projects versus web development projects are easy to read and understand. However, in explaining his metrics, the author does not provide examples to illustrate his web objects counting and WebMo model. The article is not clear in the explanation and its usage. In the author’s subsequent publication “Estimating Web Development Costs” (June 2002), a better illustration of his WebMo model was demonstrated.

We have reviewed the following papers published by the same author: