One of the most common agile tools in the toolbox is the good old iteration. Call them what you want, sprints, cycles, releases, but iterations while a potentially valuable and useful exercise for teams new to agile, can easily become dangerous to the quality of the software and the morale of the team if not used in moderation.
Iterations are useful for a number of reasons, they force teams to break down the product development, they encourage small batches and they help to keep work manageable and time boxed. Honestly, what’s not to like about iterations?
So what’s wrong?
Unfortunately the reality of iterations is often miles from the idyllic picture Agile paints. Teams and even entire businesses tend to get so caught up in the process of iterating that they lose sight of the purpose of it. SCRUM is certainly the worst offender with it’s PASS/FAIL style and it’s use of the term “sprint”.
The problem occurs when teams time box a bunch of work that is often hastily planned and loosely estimated and mold it into what feels like a good iteration. If all the planned work gets done and the results are good the team gets a pass, if they are not they get a fail. Usually the product owner or some other team authority figure decides on PASS/FAIL for the entire team (I’m aware this is a bit of a straw man and not the textbook principle of the SCRUM sprint but this is typical of how it actually ends up working).
The results of following and repeating this process are predictable. Teams are constantly pressured to improve or at least maintain their velocity. They cram more than they can handle into the iteration, they rush to get everything through, and they cover up or ignore the rough edges. When things don’t go as planned they “fail” and team moral is left battered and bruised.
The long term result is decreasing quality from pushing through half baked features, increasing technical debt from all the rushing, stress and low morale and the situation piles on itself as iteration after iteration are performed.
The road to recovery
The path back out is to return to the reason things like iterations and velocity were conceived in the first place. The core idea here is pacing which is why I completely loathe the term “sprint” because it portrays the exact opposite principle we want at play here. I don’t think anyone needs to be convinced of why rushing to get this type of work done is bad.
The importance of pace is so underrated in software. We often feel like “crunch time” is still a valid concept. Some SCRUM team literally have rules like “no one goes home if the work isn’t finish for this iteration”. Agile iterations are not an instant fix and the reality is that continuing to do the wrong thing faster and time boxed into iterations isn’t going to improve anything.
Are features being hastily planned and there isn’t enough clarity? Slow down and make sure the team gets clarity. No time to learn TDD? Slow down and learn to do TDD. The code is becoming total crap and needs to be fixed? Don’t put duct tape on it, slow the @#$% down, and learn how to rewrite it properly. Need time to understanding new tools, design techniques, architectures to do something right? Well, you get the idea.
I personally feel that Kanban provides a deeper level of guidance and exposes the root of the fundamental problem here. Kanban focuses on maintaining a continuous process rather than one that stops and starts in fits. It requires that teams are honest about what is a realistic and sustainable pace for building quality software. Finally, Kanban focuses on completing small batches by driving the delivery of individual releasable features and pushing them through a queue with a minimum of waste and waiting while ensuring each batch is still deliverable and of value to the business and the customer.
Let’s face it, no one wants to buy your iterations so why tie releasing and features to this completely superficial concept?
This is the first of what I very optimistially hope will be 30 [#NaNoWriMo] blog posts. I’ll probably miss a few, but it’s the thought that counts. I probably won’t be releasing them daily though, I’ll pace them out a couple per week.
Reference: No One Wants to Buy Your Iterations from our NCG partner Chris Nicola at the lucisferre blog.