27.2. Range of Project Sizes

< Free Open Study >

27.2. Range of Project Sizes

Is the size of the project you're working on typical? The wide range of project sizes means that you can't consider any single size to be typical. One way of thinking about project size is to think about the size of a project team. Here's a crude estimate of the percentages of all projects that are done by teams of various sizes:

Team Size	Approximate Percentage of Projects
1-3	25%
4-10	30%
11-25	20%
26-50	15%
50+	10%
Source: Adapted from "A Survey of Software Engineering Practice: Tools, Methods, and Results" (Beck and Perkins 1983), Agile Software Development Ecosystems (Highsmith 2002), and Balancing Agility and Discipline (Boehm and Turner 2003).

One aspect of project size data that might not be immediately apparent is the difference between the percentages of projects of various sizes and the number of programmers who work on projects of each size. Because larger projects use more programmers on each project than do small ones, they employ a large percentage of all programmers. Here's a rough estimate of the percentage of all programmers who work on projects of various sizes:

Team Size	Approximate Percentage of Programmers
1-3	5%
4-10	10%
11-25	15%
26-50	20%
50+	50%
Source: Derived from data in "A Survey of Software Engineering Practice: Tools, Methods, and Results" (Beck and Perkins 1983), Agile Software Development Ecosystems (Highsmith 2002), and Balancing Agility and Discipline (Boehm and Turner 2003).

< Free Open Study >

27.3. Effect of Project Size on Errors

Both quantity and type of errors are affected by project size. You might not think that error type would be affected, but as project size increases , a larger percentage of errors can usually be attributed to mistakes in requirements and design, as shown in Figure 27-2.

Figure 27-2. As project size increases, errors usually come more from requirements and design. Sometimes they still come primarily from construction (Boehm 1981, Grady 1987, Jones 1998)

Cross-Reference

For more details on errors, see Section 22.4, "Typical Errors."

On small projects, construction errors make up about 75 percent of all the errors found. Methodology has less influence on code quality, and the biggest influence on program quality is often the skill of the individual writing the program (Jones 1998).

On larger projects, construction errors can taper off to about 50 percent of the total errors; requirements and architecture errors make up the difference. Presumably this is related to the fact that more requirements development and architectural design are required on large projects, so the opportunity for errors arising out of those activities is proportionally larger. In some very large projects, however, the proportion of construction errors remains high; sometimes even with 500,000 lines of code, up to 75 percent of the errors can be attributed to construction (Grady 1987).

As the kinds of defects change with size, so do the numbers of defects. You would naturally expect a project that's twice as large as another to have twice as many errors. But the density of defects—the number of defects per 1000 lines of code—increases. The product that's twice as large is likely to have more than twice as many errors. Table 27-1 shows the range of defect densities you can expect on projects of various sizes.

Table 27-1. Project Size and Typical Error Density

Project Size (in Lines of Code)	Typical Error Density
Smaller than 2K	0-25 errors per thousand lines of code (KLOC)
2K-16K	0-40 errors per KLOC
16K-64K	0.5-50 errors per KLOC
64K-512K	2-70 errors per KLOC
512K or more	4-100 errors per KLOC
Sources: "Program Quality and Programmer Productivity" (Jones 1977), Estimating Software Costs (Jones 1998).

{% if main.adsdop %}{% include 'adsenceinline.tpl' %}{% endif %}

The data in this table was derived from specific projects, and the numbers might bear little resemblance to those for the projects you've worked on. As a snapshot of the industry, however, the data is illuminating. It indicates that the number of errors increases dramatically as project size increases, with very large projects having up to four times as many errors per thousand lines of code as small projects. A large project will need to work harder than a small project to achieve the same error rate.

Cross-Reference

The data in this table represents average performance. A handful of organizations have reported better error rates than the minimums shown here. For examples, see "How Many Errors Should You Expect to Find?" in Section 22.4.

< Free Open Study >