The biggest problem with this approach is that all the process is, at its heart, intended to do is allow average quality staff to reliably deliver “acceptable” systems. And most of the time it delivers. Now, if all you want is an acceptable (read mediocre) system, then by all means go for it. For big process to work properly the project team should come onto the project already immersed in the company’s methodology based on actual prior experience with it. A very real problem with this approach in the government contracting world is that the development contractors do not keep stables of ready and willing people who have been deeply immersed in the company’s methodology on prior projects. When they win a contract they go out and hire mostly new people for the project, who then have to go through a long period of training and adjustment to the company’s methodology.
Big process tends to force big project size - and big project cost. Typically there exist separate sub-organizations for requirements management, systems engineering, software engineering, development, testing, configuration management, quality control, risk management, etc. And all of that process increases the amount of work (much of it administrative in nature), and also increases the communication load on the project. On a project with a smaller team, much can be done via informal channels where formal process would be needed in larger projects. Proper choice of support tools can improve productivity, and those tools come with built-in process. So using them gives you desirable best practice methodology without the need for separate formal written processes to deal with.
The primary uses for project documentation are (a) to allow the system to be developed and fielded, and (b) to support post-implementation system maintenance. I recommend limiting formal documentation to that which is truly needed. I believe a project needs a high quality set of formal system requirements, and a matching system test plan. Some level of design documentation is needed, with the higher level design being more important than the more detailed design. Depending on system complexity as well as security needs, other documents will be necessary. I believe it becomes a waste of money to continue to maintain anything but the highest level documents for purposes of system maintenance. Below a certain level of documentation needed for orientation purposes, maintenance people assume (largely correctly) that the documentation is out of date, and in any event they often prefer to just directly go to the code.
Given that other conditions are extant, namely there is a small project; it is staffed with a solid leadership team; the project team is in a partnership mode with its customer; and the project is not overburdened with formal process, it becomes possible to take the final step to achieving the perfect project. The two most successful projects I have personally been involved in met these criteria, but also exhibited characteristics that now are associated with what is known as the Agile Methodology. The essence of “Agileness” is the ability to let things flow freely, and to operate intuitively versus slavishly following a rigid set of rules. It requires a high tolerance for ambiguity, and a willingness to quickly change direction, perhaps even radically. The team has to be fast on its feet.
Agility can be exhibited in any manner of ways. From the perspective of the project manager, that person may express agility in being able to rapidly shift assignments, based on knowledge of their subordinate team member’s individual experience and capabilities. This might be in response to a technical roadblock, a customer shift in priorities, or loss of a staff member. An individual analyst may need to multiplex multiple assignments, and be willing to instantly shift what they are doing based on project needs of the moment.
Agility also means thinking out of the box about how things are done. Consider system requirements. While I am a big believer in starting out with a solid set of formal requirements, one has to understand that those requirements are just an initial guess at what the system’s users truly need. It is right and proper to put together an early version of the system based on those requirements. But early pre-release user exposure will almost certainly identify various needs for adjustment. There may even be a need for significant rework. The agile team will, quickly and gracefully, make accommodation for them, even if major rework is needed. The users will deeply appreciate it. But at some point, typically after the first system release (but perhaps even earlier), it will make sense to just abandon those original requirements, and any attempt at traceability to them. The agile team then goes into change management mode and allows the system to morph into whatever it wants to become, as user complaints and suggestions for improvement pour in.
Not everyone is capable of what I used to call “go with the flow development”, and not every customer will allow it, but a project team that is able to operate intuitively can produce absolutely spectacular results.
John L. Dean is a seasoned IT professional with 40 years of varied experience, including systems engineering, systems analysis, requirements analysis, systems architecture and design, programming, quality assurance, testing, IV&V, and system acquisition, equally split between technical and management. He has worked both the contractor and government sides of the fence, has been an employee of large (IBM, CSC, Booz-Allen, MITRE, ACS) as well as very small companies, and is currently an independent consultant. He has a Masters degree in Operations Research from Clemson University. John is the “Project Whisperer". His depth and breadth of experience gives him an unusual perspective on what does and does not work. If your project is running you, he can use his experience and intuition to diagnose why and apply calm, assertive energy to start corrective action, so you can and regain control! He can be contacted at email@example.com.
© John L. Dean, 2008