Posts Tagged ‘QSMA’
Using 3σ Control Limits in Software Engineering
Source: Wikipedia; Control Chart
The July/August 2010 issue of IEEE Software features an article entitled “Monitoring Software Quality Evolution for Defects” by Hongyu Zhang and Sunghun Kim. The article is of interest to the software developer/tester/manager in quite a few ways. In particular, the authors report on their successful use of 3σ control limits in c-charts used to plot defects in software projects.
To put things in perspective, consider my recent assessment of the results accomplished by Quick Solutions (QSI) in two of their projects:
One to one-and-a-half standard deviation better than the mean might not seem like much to six-sigma black belts. However, in the context of typical results we see in the software industry the QSI results are outstanding. I have not done the exact math whether those results are superior to 95%, 97% or 98% of software projects in Michael Mah‘s QSMA database as the very exact figure almost does not matter when you achieve this level of excellence.
A complementary perspective is provided by Capers Jones in Estimating Software Costs: Bringing Realism to Estimating:
Another way of looking at six-sigma in a software context would be to achieve a defect-removal efficiency level of about 99.9999 percent. Since the average defect-removal efficiency level in the United States is only about 85 percent, and less than one project in 1000 has ever topped 98 percent, it can be seen that actual six-sigma results are beyond the current state of the art.
The setting of control limits is, of course, quite a different thing from the actual defect-removal efficiency numbers reported by Jones for the US and the very low number of defects reported by Mah for QSI. Having said that, driving a continuous improvement process through using 3σ control limits is the best recipe toward eventually reaching six-sigma results. For example, one could drive the development process by using Cyclomatic complexity per Java class as the quality characteristic in the figure at the top of this post. In this figure, a Cyclomatic complexity reading higher than 10.860 (the Upper Control Limit) will indicate a need to “stop the line” and attend to reducing complexity before resuming work on functions and features.
Coming on the heels of the impressive results reported by David Joyce on the use of statistical process control (SPC) techniques by the BBC, the article by Zhang and Kim is another encouraging report on the successful application of manufacturing techniques to software (and to knowledge work in general). I am not at liberty to quote from this just published IEEE article, but here is the abstract:
Quality control charts, especially c-charts, can help monitor software quality evolution for defects over time. c-charts of the Eclipse and Gnome systems showed that for systems experiencing active maintenance and updates, quality evolution is complicated and dynamic. The authors identify six quality evolution patterns and describe their implications. Quality assurance teams can use c-charts and patterns to monitor quality evolution and prioritize their efforts.
How to Combine Development Productivity Data with Software Quality Metrics
Consider the situation described in Should You Invest in This Software:
- One of your portfolio companies expects to ship 500K lines of code in 6 months.
- The company asks for additional $2M to complete development and bring the product to market.
- Using technical debt quantification techniques you find the technical debt amounts to $1M.
You are not at all comfortable “paying back” the technical debt in addition to funding the requested $2M. You wonder whether you should start afresh instead of trying to complete and fix the code.
Photo credit: @muntz (Flickr)
A good starting point for assessing the fresh start option is Michael Mah‘s studies of software productivity. Based on the QSMA SLIM metrics database of more than 8,000 projects, Michael will probably bracket the productivity per person in a team consisting of product management, development and test at 10-15K lines of code per year. If you use the 15K lines of code per year figure for the purposes of the analysis, 500K lines of code could theoretically be delivered with an investment of about 33.3 (500/15) man years. Assuming average loaded cost of $99,000 per man-year, the software represents a programming effort of $3.3M. Not much is left if you deduct $3M ($2M+1M) from $3.3M…
Five considerations are of paramount importance in evaluating the start afresh option:
- The comparison above ($3.3M versus $3.0M) is timeless. It is a snapshot at a certain point in time which does not take into account the value of time. To factor in the time dimension, the analysis needs to get into value (as distinct from cost) considerations. See the note on Intrinsic Quality v. Extrinsic Quality at the bottom of this post.
- Your “mileage” may vary. For example, best in class teams in large software projects have reported productivity of 20K lines of code per team member per year. As another example, productivity in business applications is very different from productivity in real-time software.
- If you decide to start with a brand new team, remember Napoleon’s quip: “Soldiers have to eat soup together for a long time before they are ready to fight.”
- If you decide to start afresh with the same team plus some enhancements to the headcount, be mindful of ‘Mythical Man-Month‘ effects. Michael Mah’s studies of the BMC BPM projects indicate that such effects might not hold for proficient Agile teams. Hence, you might opt to go Agile if you plan to enhanced the team in an aggressive manner.
- Starting afresh is not an antidote to accruing technical debt (yet again…) over time. But, it gives you the opportunity to aggressively curtail technical debt by applying the techniques described in Using Credit Limits to Constrain Development on Margin. For example, you might run source code analytics every two weeks and go over the results in the bi-weekly demo.
As long as you are mindful of these five aspects (timeless analysis, your mileage may vary, Napoleon’s quip, mythical man-month effects and credit limits on technical debt), combining technical debt figures with productivity data is an effective way to consider the pros and cons of “fix it” versus starting afresh. The combination of the two simplifies a complex investment decision by reducing all considerations to a single common denominator – $$.
Note: This is not a discussion from a value perspective. The software, warts and everything, might (or might not) be valuable to the target customers. The reader is referred to Jim Highsmith‘s analysis of Intrinsic quality versus Extrinsic Quality in Agile Project Management: Creating Innovative Products. See the Cutter Blog post entitled Beyond Scope, Schedule and Cost: Measuring Agile Performance for a short summary of the distinction between the two.
The Agile Executive 