Exploring Five Pair Programming Techniques

A collaboration by Antonio González Sanchis and Mario Moreira

Extreme Programming (XP) introduced a programming technique that’s been famous since its early days: pair programming. It sounds simple, two people working together with the same computer. However, what people don’t know is that there are many ways in which you can approach this technique. Let’s explore a few:

Driver-navigator: This is the best known style of pairing. It consists of one person ‘driving’, taking the keyboard and coding or doing the work, while the other one is ‘navigating’. The Navigator’s job is to pay attention to the work being done by the driver while keeping the big picture in mind.  They should guide the driver in the right direction. It is very important that driver explains every decision they make, otherwise the navigator might lose interest and may stop paying attention. It’s healthy to switch roles every now and then.

Ping-Pong: For this approach you might need two keyboards connected to the same computer. In contrast to the driver-navigator mode, in ping-pong both people can be driving at any point in time. A good strategy for this approach is to have one person writing the tests while the other one tries to pass them. As in the previous approach, you should be switching roles often.

Backseat navigator: A typical approach when there’s a new team member in the group or a junior colleague. The navigator (usually the more senior team member) tells the driver, who has the keyboard, what to do to solve a problem with every little detail. The navigator should provide insights on how to solve the problem but also on why that’s the best solution in mind. It’s a good way to ramp up new members in your team as they build relationships with other people in the team while learning about the working style.

Tour: If there is an even number of team members or someone who you used to pair with went on holidays, you are not going to stop working. However, when those people return or you start pairing with someone, you don’t want them to start right away. Therefore, a good way to start would be to do a ‘tour’ through the code or work you did in their absence. The person who doesn’t know the most about the work takes the keyboard and the mouse and starts driving through the code while the ‘expert’ acts as a navigator and explains every detail.

Distributed pairing: Often people tend to think pairing is for collocated teams but that’s not entirely true. While pairing face-to-face is definitely much more effective than remotely, there are many tools and ways colleagues can pair. A good suggestion would be to use a videoconference messaging tool with a screen sharing app that allows the navigator to see in real time what the driver is doing.

These are not the only pairing styles that you might encounter but they are the best known.  They are also the easier techniques to try when you want to start pair programming. Also, examples seem to focus on software development but pairing is applicable to almost every field.  For example if you are in marketing and are launching an email campaign, you can pair with someone to navigate through the email while you write it, etc.

There might be anti-patterns when pairing. For example, you might find cases when the navigator is ‘sleeping’. This means the navigator can be checking emails, texting or even coding in parallel without paying much attention to the driver. Try to identify the root causes of this behaviour before you jump in into action. Usually it might be because the driver is not explaining properly the reasons why he is doing a specific task in a particular way.

Lastly, keep in mind that while pair programming may work in every role or department, it might not be suitable for every work. There will be times when you just want to discuss the approach with the team and go do it by yourself. That’s also a way of getting feedback, which in the end is the main purpose of pair programming: getting feedback early.  

If you are interested in trying pair programming, the first thing to do is review the list of possible techniques (above) and select one that you will experiment with.  Then identify your pair.  Brainstorm on how you might approach pair programming, specify how long you will try your experiment, and begin.  

------------

Learn more about Antonio González Sanchis at https://www.linkedin.com/in/antoniogonzalezsanchis/

Read more of Mario's article at: http://cmforagile.blogspot.com 

Reducing Teamicide with Lightning Bolt shaped Teams

Teamicide is the act of purposefully disbanding a team after they are done with a task or project.  While this may not sound particularly negative at first glance, an organization loses the benefit of achieving team productivity and cohesion each time they disband a team.  When teams form, they take time to gel as a team. This is an organizational investment that often isn't realized. 

To gain some perspective, let’s take a moment to review Tuckman's model that discusses the gelling process.  Established by Bruce Tuckman in 1965, this model has four sequential phases (e.g., Forming, Storming, Norming, and Performing) that teams go through to effectively function as a unit, know each other's strengths, self-organize around the work, with optimal flow, and reduced impediments.  Regarding teamicide, if a team hasn't achieved the performing state, they will have invested in the time and team building effort without actually gaining the benefits of a performing team.   The irony is that while companies focus a lot on return on investment (ROI) regarding the product, they inadvertently achieve no ROI since they disband teams and not allow them to achieve performing.   

The next question is, why does management disband teams?  Do they not understand the harm they are doing to their organization when they disband teams?  Do they not respect the benefits of a performing team?  Or maybe they apply a move the team to the work approach when they really should be applying a move the work to the team approach.  Exploring the “move the team to the work” approach, this may occur because either there is a “form a team around a project” mindset or there is a belief that teams don’t have all of the skills or disciplines needed to handle the new types of work.   

How do we solve this problem and gain the most from performing teams?  The first change is to move to (or experiment with) applying the “move work to the team” method.   This assumes that we have teams that have the skills and discipline to handle a variety of work.  Therefore, the second change is to invest in building Lightning Bolt-shaped teams. These are teams where each team member has a primary skill, a secondary skill, and even a tertiary skill. 

The shape of a lightning bolt has one spike going deep (primary skill) and at least 2 additional spikes of lesser depth (secondary and tertiary).   The purpose of having various depths of skills is for the team to be able to handle a broad range of work and for team members to be able to step up and fill gaps that other team members may not have or need help with.  Note: some have used the term “T-shaped” teams, but I find that the lightning bolt shape is more appropriate to the several spikes of skills and the various depths that are needed.  

Creating a lightning bolt-shaped team takes an investment in education.  This takes a commitment to educate each team member in both secondary and tertiary skills.  As an example, let’s say that a developer has a primary skill of programming code.  As a secondary skill, they can also learn how to build database schemas and as a tertiary skill, they can write unit tests and run test cases.  The long-term benefit is that if the team members can develop additional skills, there is a greater likelihood that a team can work on a much wider range of work and be kept together allowing the organization to gain the benefits of a high-performing team.   This can reduce teamicide and increase the organization’s ability to produce more high-quality product.

Have you seen teamicide occurring in your organization?  Have you seen the benefit of allowing a team to remain together long enough to become a high-performing team?  If so, what level of skills were or are prevalent on the team?