Just one thing …

Curly: Do you know what the secret of life is? [holds up one finger] This.

Mitch: Your finger? Curly: One thing. Just one thing. You stick to that and the rest don’t mean s***.

Mitch: But, what is the “one thing?”

Curly: That’s what you have to find out.

Jack Palance and Billy Crystal  from the movie “City Slickers”

I always found this scene amusing when I was a kid. The old ranch hand trying to teach a city guy the value of focus for clarity. Of course at the time I just thought it was funny. After all these years, I realize that Curly was teaching a true agile value. Focus.

Focus is that thing that often gives us clarity. It gives us the ability to think deep thoughts and really unravel an idea. It allows us to extrapolate from a big thing into discrete pieces upon which we can undertake. It applies to things in life as much as it does in software development. I have started to really appreciate this as I prepare for an upcoming house declutter and move.

Do I allow myself to always embrace this seemingly simple approach. Nope. I, like others, allow myself to get too “focused” (and I use this term loosely here as it really not focused at all) on too many things. Like those performers that are spinning plates on a skinny stick, I find myself focusing on keeping them all spinning sometimes. But, as I am want to do, which often results in a blog post, I find myself stopping, breathing, thinking, assessing and questioning my current chartered course. So here we are.

Focus is a core value in agility as it allows us to actually be more productive by allowing ourselves to work to accomplish “that one thing” as Curly put it.

Under scrum, we demonstrate this through the user story, the task decomposition in terms of work effort. We use focus in terms of daily risk mitigation through a focus meeting called the  stand-up. We focus on the features committed within the review and the team within the retrospective. If you look closely, focus is woven throughout

Kanban has WIP and the innate concept of “pulled work” which itself a way to focus and measures this through the idea of how effective our focus on delivery is through things such as cycle time.

But there is one dimension in which I often see less focus. In the process of delivery. What I mean by this is often I see scrum development teams use a “divide and conquer” approach in which teams members use their specialized skills to scatter and seek to deliver. They are in no way being less agile in this approach but I have always wondered what would happen if they allowed themselves to focus and iterate just like the larger product. There are examples of this. Just take a look at Woody Zuill’s mob programming or pair programming as outlined in Exteme Programming by Kent Beck.

My latest read, “Joy, Inc” talks about the use of pair programming as the standard way of how they hire and how they work. They see it as not only a way to create shared knowledge by this but also to ultimately be more productive.

So, I have posed this question many times to developers and often get the same response “well, development doesn’t quite work that way” to which I often reply “are you willing to try an experiment and either fail or be wrong”? *smiles*

My logic has always been that if a sprint backlog is supposedly composed of a top to bottom prioritized list of the highest business value, wouldn’t we want to work through the items (given we identify any dependencies and work to minimize those with the product owner) and focus to deliver each story in that manner during a sprint? Or even look at the absolute most complex item and focus on completion of that story and then work on the lower hanging less dense fruit?

I see 4 potentially significant benefits that could result from this:

  1. We get items in front of our team members specialized in quality quickly so they can assess features for compliance to acceptance, note any software issues or quality enhancements and they can become productive more quickly. They are reviewing and testing completed features, even if there are mocks behind some dependent needs of other stories. They also are able to learn a portion of the system under construction and build automation against it for regression more quickly if needed.
  2. If we for some unforeseen reason we are impacted to effectively deliver on all stories for our commitment, we can at the first responsible moment, assess remaining team bandwidth, current deliverable dependencies and make decisions so that we can have that conversation immediately with our product owner so we can remain transparent on the impediment and impact but emphasize our focus on the highest priorities or largest complexity first.
  3. We reinforce iterative development at the lowest level of development (by using refactoring and integration as part of the way we work) and can focus on ensuring that we meet our agreed definitions of done at each story level.
  4. We “should” end up with a set of features tested early which should allow us to as a team to swarm baking in quality for our product and preparing for our review at sprint’s end (given we did not over commit …)

So in my view this approach makes sense to me and I have worked on teams that worked this way and we remained productive and felt like we were accomplishing more towards our product goals. We effectively delivered features regularly for review for the stakeholders.

A lot of folks may be reading this and joining the chorus of previous conversations saying “yeah, sounds good but you don’t really understand” …

So I reply to you all … “are you willing to try an experiment and either fail or be wrong”?

 

Advertisements

One thought on “Just one thing …

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s