The Mythical Man-Month

The Mythical Man-Month

Plan to Throw One Away is the eleventh essay of The Mythical Man-Month by Fred P. Brooks. In this essay Brooks expounds on the idea that system development and change are inextricably linked. Brooks begins the essay by making the argument that in many projects the initial design is replaced as more information is generated. Brooks uses the analogy of chemical engineers bringing a new process to production. ChemE’s recognize the need to prototype and pilot new processes in order to find the kinks. Software projects are no different.  Rarely is the first attempt useful to the end consumer, and the usefulness of that first attempt is less in the code than in the feedback it generates. Software development is no different. The initial conceptual design and anticipated technical architecture of a large project rarely stands up to the rigors of the discovery process, and those designs should be learned from and then thrown away.  Pilots and prototypes are often used to refine and validate requirements. Translating ideas from user stories, requirements or even use cases into a functional system is a journey of discovery and change.

Methodologies at that include prototypes, pilots or deliver functionality quickly generate change. Brooks drives the home the idea that change is inevitable with the phrase “Only constancy is change itself.” This section foreshowed the value in the Agile Manifesto that states, “Responding to change over following a plan.” On a cautionary note, while some change is inevitable, too much change can lead to a failure to deliver anything.  Time boxes are a tool to compartmentalize change so that change does not lead to a failure of delivery.

If change is both inevitable and good (within limits), then both systems and organizations (a type of system) need to be engineered to support and facilitate change.  Architecturally, techniques such as modularization, object-oriented design and other processes that foster simplification and incremental change create an environment in which change isn’t avoided, but rather encouraged.

Embracing change fosters and effects more than just the application code and technical environment; the human system needs to be designed to support change. Here again Brooks is a thought leader by espousing the idea that people with broad skill sets create flexibility. In 1991, 16 years later, Tim Brown of IDEO coined term ‘T-shaped people’ (or skills) to express the need to broaden skill sets in order to foster flexibility.  Just like embracing application architecture that supports and fosters change, the organization architecture needs to be designed to react to feedback so that the process of development can effectively react to change.

Brooks describes development as a dance that takes two steps forward and one step back. I recently participated in a discussion of application support with a manager of a support team. The manager observed that maintenance gets harder as applications age. Applications, when initially delivered, are at their lowest level of entropy.  Not zero because all applications are delivered with defects that must be repaired. As repairs are made the original structure of the application will be disrupted and the newly inserted defects will increase the entropy. Pilots, prototypes, flexible designs and T-shaped people are all steps to improve application delivery that maximizes value and minimizes entropy.

Previous installments of the Re-read of The Mythical Man-Month

Introductions and The Tar Pit

The Mythical Man-Month (The Essay)

The Surgical Team

Aristocracy, Democracy and System Design

The Second-System Effect

Passing the Word

Why did the Tower of Babel fall?

Calling the Shot

Ten Pounds in a Five–Pound Package

The Documentary Hypothesis

 

Putting the parts together!

Putting the parts together!

Hand Drawn Chart Saturday

Techniques for building an initial backlog can be classified by how the conversation between stakeholders and the project team is initiated. Some techniques are focused on asking, other techniques focus on observing, while the third category is all about showing something and getting reactions. Most practitioners blend the best from each of the categories. Here are some examples of the hybrid techniques:

Role Playing Prototype: This technique blends paper prototypes (show) with role-playing (observe) to get users and stakeholders to consider how they might act in an environment that has not been fully designed.

Straw man/JAD: This synthesis seeds a JAD session (ask) with an loose outline of a solution or set of potential solutions that are used to guide the discussion which are at the core of JAD. However, the seeding tactic can inhibit creativity. The technique is less constraining when a set of competing solutions is used as the conversation seed, however developing the range of solutions before the JAD session  increases the cost of the JAD.

Embedding: This techniques puts team member(s) into a department to actually perform the work (observe) alongside actual users and stakeholders. This generally requires the embedded team member to be trained and mentored to intimately see how the work is done. I have seen debrief sessions added to this technique to ensure that participants get a chance to discuss the nuances in the workflow.   As I have noted before, with any observation technique everyone needs to understand what is going on and why. This is not an episode of Undercover Boss.

Combining tactics from different categories of techniques that teams use to develop an initial backlog can fundamentally change the dynamics of the requirements definition session. A group of stakeholders will generally have a diverse range of learning and interaction styles that they favor. Combining backlog building techniques gives the project team a better chance at making a connection. Combining techniques should not be done randomly or an ad-hoc basis. Selecting which techniques to combine has four prerequisites:

  1. Someone with experience and training (perhaps a business analyst).
  2. A knowledge of the user community (knowledge the product owner can provide).
  3. Planning (time and effort).
  4. Involved users and stakeholders (call on the product owner and project sponsor to help with this prerequisite).

Developing an initial backlog is a step to get projects going and moving in the right direction. It is, however, only a first step. Backlogs will evolve. Teams, product owners, users and other stakeholders will gain knowledge and experience as the project move forward that will continually shape and reshape the backlog.

The evacuation instructions could be a form of paper prototype.

The evacuation instructions could be a form of paper prototype.

The most powerful single technique for generating requirements is showing users and stakeholders something and collecting reactions. There are numerous techniques for developing something to generate feedback, ranging from functional code that could implemented in production at one extreme, to functional prototypes in the middle, to paper prototypes at the end of the scale. Functional code is used to generate feedback to evolve the backlog in Agile projects, however showing techniques are not used as often as they should be to generate the initial requirements backlog. The reason these techniques are not used is the perceived level of effort needed to generate the prototypes or an impetus to begin building the solution immediately.

The power of the showing techniques is based on the theory that many people only know what they want when they see it. The process of generating feedback and requirements is also risk management. A prototype reduces the risk that the project will either build the wrong thing or build the right thing wrong. Functional prototypes also, to an extent, are useful to prove that a solution is at least conceptually feasible (prototypes are generally too small and not fully functional therefore do not truly serve to prove technical feasibility). Paper prototypes address generating requirements and reducing risk but are faster and cheaper to generate because there is no code or code related infrastructure. A very simple example of a paper prototype for a customer relationship management system (CRM) might be set of drawings of screens to show the rough flow of work.  Users to use the paper screens to imagine the process and discuss the flow. In comparison a functional prototype would have mock screen on a computer show system flow but typically without any background functionality.

The first issue with prototypes is that developing them requires time and effort. Project teams are often presented with a goal, a budget and a deadline. In most corporate IT organizations, performance against the project budget is considered a critical measure of progress (and in the longer term, project success). Therefore teams and project/program managers mercilessly manage the budget to improve the possibility of project success. Unless prototyping is built into the budget and the planned approach for the project, there will be pressure to use the less costly, albeit less effective, asking techniques to generate requirements. Teams, sponsors and project managers make a rational choice to manage the budget risk against the possibility of not generating a good enough backlog to get started. However, generating requirements using prototypes is a process that can used to balance requirements and budget risk. The higher the risk of not having a more through early backlog the more important techniques like prototyping will become to mitigate that risk.

The second issue that the use of prototypes face is what I call “starting fever.” In many methods, including Agile, the whole cross-functional project team is assembled to start the project. There are many individuals that don’t believe that gathering requirement is important, therefore unless there is something for them to build or test, they will find other things to do. There are numerous ways to deal with this type of structural slack; here are two extreme cases will illustrate the range. The first solution is to have a subset of the team (like the Three Amigos) generate the initial backlog before kicking the project off. The second solution is have the whole team spend a day as the project kicks off to generate a quick initial backlog and then use the demonstrations at the end of each sprint to continue to flesh out the backlog. I lean towards scenario one or variants in which the other portions of the team work on framing the technical and physical infrastructure.

Another way “starting fever” can be triggered is because of the panic a fixed budget, fixed scope and fixed date project that someone actually said yes to before requirements are known can cause. Before you protest, regardless of how much every developer, tester, BA, project manager or CIO believes in that this an irrational situation they happen ( I see this as the norm in some organizations) and they casue teams to abandon good  practices like prototyping. In the long run project teams have to pick up the pieces when these types of projects had problems. When teams are put under immediate schedule, budget and scope pressure, leaping into action and then creating and firming up a backlog later can look like a great solution. Action is still equated to progress.

Prototyping is an effective tool for driving out requirements that can’t be expressed until they are seen. The backlog building techniques that show the users and stakeholders something to react to also serves to mitigate the risk of building the wrong thing or the right thing wrong. The power of these techniques is offset by the cost and time required to generate prototypes and a fear that unless you are building something the project has started and the due date is at risk.