Jeff McKenna's Agile Development Blog

Updating the blog

As you may have noticed, there has been little activity on this blog in the lasst few months. This has been my doing ... or rather my not doing.

Unfortunately Bill Ives is no longer able to add his expertize to our blogging efforts. His absence has been missed since he was a strong catalyst for me to write both through his interviews and regular pestering!

Recently an old friend of mine and past colleague David Socha and I have been collaborating on various topics related to software agility. David and worked together at Rational and also at QuickSilver during my time in Redmond, Washington. We have also collaborated in the past on the role of design in software. I am enthused about our working together again.

I have asked David to add some biographical information which I am sure he will do when he has a few moments. After all we all know how much time we all have!

Welcome, David!

Agile Webinar Q&A

Yesterday, John Scumniotales, Paul Dupuy and I participated in a Webcast for The Agile Journal. We could not answer all the various questions on line. I thought it might be interesting to see some of them and our answers.

You can see the Webcast yourself by visiting: http://w.on24.com/r.htm?e=155782&s=1&k=A98F0FB3D0AD7D354E60F0C65DB31BB7

Here are the questions and answers:

Question: What if the business process isn't defined how does Agile handle business reengineering?

Answer: When the business process is not defined, an agile approach provides an incremental model to explore and define it! Using Scrum for instance, you would work with your Product Owner (customer) to elaborate user stories that define some aspect of the business process. Don't boil the ocean. Pick a small "chunk" that can be easily described and understood. If the end goal is to have a system that is implemented using software, then you should strive to implement this process as well. To the extent possible, this should be accomplished in the same sprint. Avoid using the sprints to only document the process. Your goal should be to produce working software as early as often as possible.

Q: Thanks for this presentation. One issue we struggle with is using scrum teams that do all of the things you are describing but we have trouble bidding competitively on small professional services projects. There is some overhead with scrum when applied to smallish projects (lasting 2 - 6 weeks) that we struggle with. If you have any suggestions or examples for how to insulate the customer from our internal costs of using scrum on small projects I would love to hear it. We've been using scrum for about 4 years and it works great for our product team. Any thoughts you have would be appreciated.

A: When using Scrum on professional services engagements, we highly recommend giving your customers visibility (and getting their buy in!) to the process you use to deliver the requested features. If they've agreed to the approach, then this should minimize the impedance mismatch and associated overhead costs for spinning up the project. Also, try and keep your teams together. Remember, with Scrum, we focus on building and optimizing teams. If your building up and tearing down teams for each project, this will certainly introduce overhead and also decrease the predictability of teams velocity as it can take up to three sprints for a team's velocity to normalize.

Q: Which is more important between Individuals/Interaction and Processes/Tools? What I should do if my team members do not have cross-functional skills?

A: In agile we always favor direct communication. Whenever and where ever possible get people in a room together hashing through the problems and issues. The reality of "agile in the enterprise" is that teams are distributed by time and distance. This is where tools are essential in enabling team collaboration. When teams are first formed, its common that there is some specialization of skills. This moderates over time. Pairing, an extreme programming best practice, is an excellent tool to use to help cross train team members.

Q: How can you help a large, compliance-oriented, traditional development organization move to a more agile approach without jumping all the way into Scrum?

A: If you don't have expert help on staff, we recommend getting help with the organizational change management issues that you will deal with in adopting agile. That is why Serena partnered with Valtech, a leading provider of Agile transformational services, with our Serena On Demand product.
Also set it up for early success. Identify a few of your best teams that are starting up new projects. Plant an experienced agile coach on the team, get them norming and storming. You'll be able to use the experience from this team to seed (and motivate) others.
A tool like Serena Agile On Demand helps as well. It provides a prescriptive Scrum framework for the teams to follow.

Q: What tools to you offer to facilitate yearly budgeting and goal setting in a way that could feed into incremental deliveries while supporting financial accountability within key cost centers?

A: The "team" is the key unit for agile projects. We want to fund teams and keep them stable for as long as possible. This allows us to optimize predictability. So you want to fund as many teams as possible, keep them together for as long as possible, and bring the work (projects) to them. At the portfolio level, its also helpful to consider relative investment levels for key initiative requested from the business units. This will dictate the mix of projects that you direct to the teams.

Q: Can we implement these tools based on traditional methods and then seamlessly transition to agile?

A: No. Adopting agile has significant impacts on the people and processes within an organization.

Q: How does this approach handle complex systems integrations with third parties and systems involving end-to-end financial settlement processes such as GL posting.

A: This is hard regardless of whether your following a traditional or agile approach! When we scale agile, we typically have many teams collaborating on a single appliation/product/service. The integration of the work developed by these teams of course needs to be closely managed. From a program management perspective, we can use the "Scrum of Scrums" to manage the inter-dependencies between teams and other cross team issues. We can also use the "Super Scrum" as longer range product planning tool to make sure that work is appropriately sequenced so that cross team dependencies will be satisfied.

Q: What if the business perception is that small, incremental releases do not generate adequate business value to warrant business change, training, and communication required? Can this process handle small internal increments that culminate into large releases?

A: Absolutely. Agile (and Scrum) are based on incremental approaches that produce potentially shippable code. Whether or not the increment is shipped is a business decision based on product, market, and customer readiness.

Q: (Loaded question) What if the user stories to be delivered in a sprint get delayed, and the spring demo cannot be completed on time?

A: We love loaded questions! The duration of a sprint never changes. If a story is not completed, its not accepted and not delivered as part of the sprint. The story will either be returned to the product backlog or moved into the next sprint.

Q: How is Serena differentiated from competition (such as Rally)?

A: Serena Agile On Demand was built from the ground up to deal with agile at the enterprise scale. Multiple teams, multiple releases, multiple products. We offer unique capabilities in our ability to deal with enterprise complexity. Serena Agile On Demand comes fully loaded for prescriptive Scrum development. While complete in its ability to deal with Scrum, its hyper-configurable to meet your specific needs as well. Serena Agile On Demand not only delivers the product features as a service, but it also delivers on demand agile coaching through Coach Live in cooperation with Valtech.

Q: How can the tool support allocating planned resource capacity to specific initiatives while allowing freedom from scrum masters to manage specific task assignments for team members. Need to measure that actual utilization is not exceeding targeted resource allocations.

A: A major difference with agile approaches from traditional approaches with respect to resource management is that the team is the unit of planning, not an individual resource. We try and put cross functional teams together and keep them together. The team then establishes a velocity - basically the amount of work they can complete in a sprint. Once the teams velocity is established, the team commits to deliver the amount of work that matches their velocity. Agile teams are always looking for ways to optimize their throughput (increase their velocity). This is contrasted with traditional approaches where management attempts to maximize utlization.

Q: Can the Agile On Demand task board be customized to reflect custom statuses ("swim lanes")?"

A: Yes. The swim lanes in the Card Wall are completely configurable.

Q: Who selects scrum team

A: When making a transition from traditional methodologies to SCRUM we think it is a best practice to "invite" members of the team to join the SCRUM. It's important in the early stages to get enthusiastic members on the team who will actively engage and support adoption. Hard to get people like that if they are 'forced' to be on the team.

Q: Since agile / scrum utilizes an Empirical model for managing costs, How can management make an assessment of the total cost of the program in order to fund development.

A: Plan the work as far as the release horizon using coarse grain items. Have the team size the items using rough estimates. Use historical information from the team's activity or make an educated guess about how much work the team can do in a sprint. Determine the cost of a team for a sprint. Calculate how many sprints you need to finish the work in the release. This gives you the estimated cost to complete the work. You'll need to use scope negotiation skills to prevent "scope creep" and decrease the priority of low priority items. You'll also need to recognize that as the software is developed, changes will occur (which is a good thing -- you'll find that what you wanted at the beginning isn't what you need) and trade-offs will have to be made.

Q: I manage a small unit responsible for several projects. Some were started over a year ago and were designed to follow waterfall. Subsequent projects are now developed via Scrum Agile method. With at least a year left on the waterfal project, how or should we switch mid-stream to agile?

A: It is fairly simple to switch to Scrum to manage your work. If your waterfall planning resulted in a pretty gantt chart with tasks, you can use them as the work items in your product backlog. Read up on Scrum and get a certified scrum master if you can.

Q: Have you ever heard of ACTUAL customers (in addition to product owners and PMs) participating in a sprint review? Any ideas how this could be implemented? Thanks!

A: Absolutely. This is done quite frequently in fact. If possible invite the customer to attend in person. Otherwise their are many virtualized meeting tools that can be used quite effectively.

Q: Could you expand some on non-product stories, how they get prioritized since the product owner may not care, how you track those vs. product user stories in metrics, etc?"

A: Pay me now or pay me later. Sometimes the team has to go slow for a few sprints to go faster in future sprints. There are many examples of non-product stories where technical debt is addressed that may not result in direct value to the customer. This is where the team and the Scrum Master must lobby and compel the product owner to include these stories into a sprint. Without this balance, an insurmountable amount of technical debt will build up.

Q: Is Serena overkill for a team just looking for a requirements/project management?

A: No, not at all. Serena's Agile On Demand has the ability to manage many types of backlogs and is ideal for requirements/project management. Internally here, the Marketing department has several teams actively sprinting and using Agile On Demand to manage projects and requirements. Multiple product backlogs, about nine is our case, contain stories for different projects the team is working on, each managed by a separate 'product owner' (usually a Product Marketing Manager). Then there are separate TEAM backlogs that contain the sprints and cross functional teams of marketing types who all contribute to the completion of stories with the project backlogs. The benefits of have a team manage requirements this way include time management, capacity planning for the sprint and minimizing the randomizing effects of fire drills.

Q: How did you convince your managment to use XP?

A: XP is a collection of best practices that increase the likelihood a team will deliver running tested features that meet customer needs in a predictable and repeatable way. Surely management can see the value in that result? ;-) The book "Extreme Programming Applied" has a good discussion of this. Also see the discussion on the c2 wiki: http://c2.com/cgi/wiki?WhyItIsSoHardToSellExtremeProgramming

Q: So, Product Managers can equate to Business Analyst in a traditional organization and Scrum Master is played by Project Managers?

A: The Product Manager's roles and responsibilities align quite well with the Product Owner's. BUT, the approach they take to elaborating requirements and engaging with team varies significantly. A Project Manager can be a Scrum Master, but be cautious. Project Managers are typically in a top-down command and control roll where they plan and dictate to the team how to complete work. A Scrum Master is a passive leader that facilitates the team. The Scrum Master coordinates meetings, interfaces with external teams, and manages the impediments that the team can't resolve themselves.

Q: How do we handle Standup meetings for geographically distributed teams?"

A: Tools are essential. Use planning and collaborations tools like Serena Agile On Demand and Virtual Meeting tools.

Q: What estimation method do you use from sprint planning?

A: We use story points with "unit-less" value. Our teams use planning poker to estimate sizes. There is an interesting video of it's creator talking about it here: http://www.youtube.com/watch?v=1oIaz7aZsno&feature=channel_page and more info on this site: http://www.planningpoker.com/

Q: Would you recommend adding to the backlog a "regression testing" item?"

A: Using stories to manage work that is difficult to complete or track can work well. If the team is having trouble doing regression testing in a predictable way during the course of a sprint, using a story is recommended. If the team does a consistent amount of regression testing every sprint, you could instead consider it "overhead" and choose not to track it.

Q: is serena agile license per user based?

A: Yes. $35 per user per month.

Q: Where would a Business/System Analyst fit into the Scrum team/role

A: Often times we see Business/System Analysts as part of the Scrum team in primarily a Product Owner role.

Q: IS a user story different than a story board. Or is the story board where you post all the user stories for selection?

A: The card wall (story board) is a technique for showing stories and their current state in a single view. It is typically used to manage the state for stories in a sprint.

Q: Where is the current "high-bar" for portfolio management across an enterprise for full Scrum methodology used on across large teams of developers? Team size of 500 + developers? 800+ developers?

A: 1000s of resources.

Innovation Games® workshop

I just finished a two day workshop on Innovation Games® by Luke Hohmann of Enthiosys. In short it was great, stimulating, chaotic, and filled with lots of ideas and games to play. Most of the attendees were independent Scrum Trainers/Coaches with some additional perspectives given by two market researchers. A couple of us were embedded coaches/trainers like myself. It was a stellar group with lots of often expressed opinions! The games we played with are all described in Luke's book, Innovation Games which I certainly recommend.

All good training for agile methods include liberal amounts of experiential exercises. I was first seriously exposed to this approach when I studied with Jerry Weinberg. He called them simulations and from him I learned how precise in terms of teaching objectives they can be. Luke has taken this to even greater levels and he uses the more currently politically correct term, game, to refer to this type of exercise. These days when agile trainers gather together they often share games they are using and practice on each other.

In my view the power of games is that a well designed game will engage more of the our capabilities. Some games get the body involved - somatic involvement. We know that learning for most folks is enhanced when the body is more involved. Some games get both sides of the brain involved - right brain (random, intuitive, synthesis, subjective, wholes) as well as left brain (logical, rational analytic, objective, parts). We need both modes to solve complex problem in new ways. Some games encourage our emotions to play a more active role. Our passions can be expressed and utilized.

As we discussed in the workshop, a current constraint on successful agile projects is the difficulty of writing those stories that are the work to be done. Scrum books assume that they exist. They don't.

Surprise, surprise. For over 40 years from the engineering side I have heard 'If we could just get good requirements we could '. Typically blanks are 'Take less time', 'Have fewer bugs', 'Deliver what is needed', 'Cost less' or 'Get the design right'. This problem has not gone away. And agile has never promised that it could fix this problem. Well executed agile only promises that engineering can reliably deliver what is asked for at a known quality level at a known time.

The short time box aspects of agile allow more in-course correction to be made based on what I think of as reality based feedback. This behavior gives the impression that 'requirements' are better. That is not true. Agile projects can produce bad products faster and more reliably than more traditional techniques.

Innovation Games® as we learned them are a set of tools that can seriously help with the difficult problems of where the product is going. We experienced the ease and FUN that can be experienced as difficult problems are addressed relatively quickly and innovative potential solutions are discovered. High level plans such as roadmaps can be addressed with these techniques as well as prioritization of high level backlogs. Filtering the infinite world of the possible down to the practical world of the doable.

Some of the games are now available online for distributed teams and we got a chance to play one of them. An impressive beginning to a difficult problem.

I was impressed.

Junkies and Zombies: Patterns of Project Behavior

Yesterday I finished reading Adrenaline Junkies and Template Zombies - Understanding Patterns of Project Behavior by Tom DeMarco, Peter Hruschka, Tim Lister, Steve McMenamin, James Robertson and Suzanne Robertson. Whew... with a list of authors like that it must be good or at least long.

It is good and it is not long. It is a very readable compendium of 86 patterns with a couple of ‘interludes’ to break things up. The patterns are in a short, fairly informal format of some two to six or seven pages. Each describes a way we operate or operate in organizations. There is some gentle discussion of details. And an even more gentle treatment of those patterns that are judged less than ideal.

The book supports achieving a goal of mine for all teams and individuals on those teams: Becoming more conscious of the work that we do and more conscious of the relationships that we are in as we do that work. The patterns are presented as a way to think and reflect on how we work together. Some are more technical, some are more relational, and some are individual.

The names give many of them away: Face Time – about talking together; It’s Always the Goddamned Interfaces – exploring the congruence of teams and components that they build; Sanctity of the Half-Baked Idea – notices that if only ‘good’ ideas are allowed the great ideas are more difficult to find.

I recommend this book. Enjoy it one small bite at a time.

P.S. As I was reading the book I found myself wanting to score myself and score my current job situation. How was I doing? How were we doing? What was our Zombie Score? Perhaps the authors can provide a simple spreadsheet and do some analysis. If the Zombie score of an organization is good, are they more successful?

My Eight Current Agile Blogs

When I put together this blog I wanted to make reference to other bloggers. I noticed that there are a lot of Agile blogs. What I chose to do is to pick people I knew or blogs that I found particularly interesting. There are many, many others. Have fun browsing!

Tobias Meyer writes the blog, Agile Thoughts. He is one of the most interesting trainers that I know. He is very opinionated. I know him personally and really enjoy being around him. I have learned a lot from him. Tobias is always very thoughtful and I really respect his opinion. He is particularly interested in the games we play to help people learn.

Mike Cohn writes the blog, Succeeding with Agile. Mike is a classic and was on the Board of Directors of the Scrum Alliance for a long time. He has written some of the best books on Agile. He is very strong in what he believes Scrum should be. I have trained with him. I enjoy being with Mike a lot and he is very knowledgeable. He has thought out a lot of things and is not afraid to make controversial statements.

Ken Schwaber writes the blog, Edge of Chaos: Edge of Chaos, Agile Development Blog. He is the man who put Scrum on the map. He went to conferences starting in the mid-90s and talked about Scrum and learned more about it. Ken co-wrote the first book defining Scrum and is the “father” of Scrum in a sense of getting the word out. He put together some of the first training programs. Ken is still active. I see him about twice a year at the Scrum gatherings.

John Scumniotales writes the blog, Agile Product Development. He was on the very first Scrum project, which I also had the pleasure of being on. John was, in essence, the very first Scrum Master. He is now a VP of Engineering at Serena Software. He is taking Agile up through the enterprise. He has been very concerned with how to bring Agile practices as you move up the management chain. We have disagreed at times but he has thoughtful articles about his experience, more on the management or executive side of Scrum. I read his blog fairly often.

Jeff Sutherland writes the blog, Scrum Log. He was the intellectual founder of Scrum. He did all the research with the Japanese. He was the CTO of the first Scrum project that John Scumniotales and I were on. Jeff is a very interesting fellow with two MDs. He was a fighter pilot in the Korean War. John is full of energy and goes around the world training people in Scrum. He is constantly pushing the boundaries of Scrum. You should read his articles about red and blue Scrum. He has said that if the project team does not have a velocity in place after two or three meetings you should disband the team and start over again. I think of what he does as tough love or “tough Scrum.” Jeff is a great guy who always listens carefully to what you have to say.

David Anderson writes the Agile Management Blog. David is from the Lean thinking approach and I find him very interesting to read. He has a lot of good ideas. He is very much into measurement and its effects on projects. Lately he has been chatting more about Enterprise Agile and more focus on Lean.

Corey Ladas writes the blog, Lean Software Engineering. I do not know him but I find his work in Lean very interesting as we bring the Lean practices more directly in front of the Agile process. Clearly Agile and Lean are closely tied together. Corey makes this very explicit.

The Agile Advice blog is subtitled Working with Agile Methods (Scrum, XP, Lean). I do not know this writer but it is a blog I saw that I liked. I thought it was interesting and wanted to include it.

The Importance of Quality Engineering Practices in Agile Development

As I have working with a number of Agile development efforts I have noticed the importance of quality engineering practices in Agile adoptions.

Agile work management is about orienting people to make the right work at the right time. I sometimes think of it as the TIVOing of software development. We are timeshifting work. You wait and do things at the right time. Ward Cunningham, the inventor of the wiki, talks about only doing what you know. Don’t do what you don’t know. This is a critical quality in how we experience development in Agile.

A clear prerequisite to being able to do this is quality engineering practices. We have discovered that many organizations do not have a clear understanding of this. Now quality is a fairly broad term. It can be interpreted in many ways. People say, well, of course you need quality engineering. And I have a specific take on what I mean: a quality of correctness. It is correct.

It is little things like the code works and it continues to work. It does not have bugs in it. The code actually fulfills the purposes for which it is intended.

It is finished. Sometimes the code never seems to get done. There are always some bugs associated with it or the documentation did not quite get done.

And we have experience with this. It can be achieved. The project I mentioned in the post, Achieving Immediate and Continuous Bug Fixes in Agile Development, is an example. Because we wrote with sufficient unit tests, because we had good automated integration tests, because of a 100 regression test way of thinking, and because of continuous integration that ran these tests, when we finished a piece of code it was done and there were very few bugs, never more than 5 at one time for the entire project.

Now this does not mean that we had all the features you might ever think about. In my mind that is the trap we fall into. We say: ‘Oh, that feature should do X and Y should be done in a different way.’ That is not what I mean by done or quality. Quality means that the system does what you want it to do at that moment in time. We take small steps and make sure that each step is done in a quality manner.

Later, you might add some different features or change features to respond to market pressures, better ideas, customer input or other conditions. However, when the new stuff is put in, these additions also have the quality of correctness.

This requires good engineering practices. It is interesting to consider this because what happens is that the teams that achieve this way of thinking/acting with their code are very productive. They feel better. It is fun to be around them. They are always thinking about new things because they do not have to think about the old things any more.

The social side is that they are more willing to try new things. It means that there is more innovation in the products. They know that if something is done it will be quality or they will just take it out.

I see Agile being like the train running down a well laid set of tracks, constantly moving forward on a solid foundation. When the foundation is kept strong (the attitude and practice of quality) the train can easily move. As you go forward everything is working and you are just advancing. More alignment is naturally achieved.

In fact we use the term release train to refer to a particular way of deploying software on a regular basis. It is this feeling. You are on an engine that is running and you drop new requirements in as you go forward and working functionality comes out the other end.

This causes some other issues. At times, developers never feel they are done. So we need to address this human need for accomplishment and build in pauses, vision setting, and reflection to address this human need. You need to have the team take time to stop and look at their accomplishments.

This means that way down deep inside the team has good tools and a good sense of what testing is. The developers know what good design is and how to recognize it. They know how to change the design if the requirements change. Extreme programming really pointed the way to achieve this quality. If you do not do this then Agile becomes just another process that sort of works. The old ways come back: schedules slip, bugs proliferate, morale declines.

I wanted to discuss quality engineering because it is important for people to understand that Agile is not just about making short iterations. An attitude of quality is important. It is an old software adage to get the best people you can. Here we are suggesting that the best people as those who know quality engineering.

Does Agile Development Require UML?

I am often asked about UML or Unified Modeling Language. A lot of young developers don't seem to be using UML or any kind of modeling language when doing development.
I think this is a big mistake.

When we go back to the roots of Agile, which I consider to be extreme programming, we see the issue of quality elevated to the highest level of importance. Quality is very important and low quality can affect new feature development, team morale, and drown an organization in bug triage, reporting, fixing, regressing, documenting and forward fixing.

We know that unit testing and even better, test driven development, can drive the quality higher. But what does that have to do with UML?

The best way to have much better software is to have better design. Having a better design is the product of communication with your other team members about design choices. A great language for that communication is UML.

Text or just word-based communication about design is slow and subject to interpretation errors. Most folks find communication about patterns, which are really just examples of good designs for common problems, much easier with UML. (This is why we use UML in my Emergent Design Class.)

When I talk about UML, I do not mean UML models that are so specified that code can be
generated directly from them. I mean 'blackboard UML'. Usually this is the class diagram, the sequence diagram and the interaction diagram - just three of the possible diagrams in UML. Other diagrams can be used in specific situations. State and component diagrams are seen as well. I would say that if you only use class diagrams, you have achieved a significant percentage of the value. For a simple introduction to the class diagram go here. The book on UML I like best is UML Distilled by Martin Fowler.

High performing Agile teams tend to have visual things around their work areas or on their wikis. In particular I see UML on white boards or on wiki pages to show high-level designs of areas of their work. People have good discussions around these. And make changes as the project unfolds. Product Owners use UML to capture Domain Models.

I believe that every person (or at least every developer) on an agile team should be able to use simple (blackboard) UML to talk about their work. And everyone should be able to diagram the high level abstractions in the team’s code - the architecture as it were. What do you think?

Comments on Amr Elssamadisy's new book Agile Adoption Patterns

I just took a detailed look at Amr Elssamadisy's new book Agile Adoption Patterns - A Roadmap to Organizational Success.

I like it. It brings together a wide range of agile practices and gives good analysis of each. As with good pattern books, it shows why, how, why not, and such for each pattern. Since there are 38 (!) practices, finding the one that matters for you is important. They have been clustered a bit and are still hard to find.

I personally find patterns a useful way to capture and present what is needed and at the same time difficult to understand. Since I already understand these patterns I found them useful and occasionally found a new way to think about a particular one. If I did not already know the pattern, I am not so sure how useful the book would be. I will defer on this to others in that position.

And one comment early in the patterns did catch my attention and has, I hope, influenced my thinking: Amr comments on the importance of learning. Much of the agile feedback mechanisms can be viewed as opportunities for the team to learn.

The importance of the learning cannot be dismissed. Getting better in every aspect of creation - planning, design, measuring, coding, testing, communication, delivery - is critical to create a high performance team. Such teams look at every change, comment, bug, and hick-up as an opportunity to improve.

I use the repetition of activities as an easy way to identify an area for improvement. If the team is spending time on activity over and over again then an opportunity exists for improvement. In particular these opportunities often are automated to eliminate the human time/errors that are present.
So, Amr, thanks for the stimulation.

Conducting Release Retrospectives in Agile Development

In this post Bill and I discuss the concept of release retrospectives that I briefly touched on in the blog overview.

Bill: You mentioned how you turned a software development retrospective on its head and discussed what went right rather than what went wrong. This is a nice twist on lessons learned and should encourage participation and engagement. After all what you want to repeat is what went right. So let’s have the details.

Jeff: This project had been going for six months. I talked with management about having a release retrospective. Most people are familiar with having a retrospective after each iteration or spread. This usually occurs about every four weeks for an hour or two with each team doing this independently. But we also recommend that after longer periods, say after a major milestone is reached, two practices. First, the teams take a week off and stop the sprinting and breathe for a bit. And second, the teams have a more relaxed release retrospective.

In the book Agile Retrospective, Esther Derby and Diana Larsen talk about the need for the shorter sprint retrospectives while you are still in the fray. These typically last one hour or so and involve one team. In contrast, the release retrospective involves the whole group of teams and takes a longer look.

Bill: What was the problem that set up this release retrospective? Why did you do it?

Jeff: The problem is a generic one. People need time if they are going to learn. One of the Agile tenets is that teams need to be able to learn and reflect. The release retrospective is looking at it from a longer time perspective and from the perspective of all the teams on the project. Typically problems are solved at the individual team level. So as we move to the multiple team level I am a little less concerned with specific problems and more concerned with attitude and making sure that people feel they are on a larger project where teams are working together.

The heads down sprint-to-sprint activity in a team can be very focused and sometimes isolating. It can be harder to come up and breathe. So in the release retrospective there are discussions at a higher level about such things as the vision for the next six months or big changes that are being considered. People can reestablish the larger context that they are working in.

There is a type of inquiry called appreciative inquiry that is a technique for exploring issues that allows people to focus on solutions and positive aspects of their experiences. It is referred to as the discipline of positive change. There is some data in research that indicates that larger groups of people who focus on things that work and spend less time on problems such as we typically do in our industry, people feel better, are more productive, and their customers are happier. I was exposed to this research and I thought it might be interesting to bring this approach into the retrospective.

Here is what we did.

First we had everyone introduce themselves and provide a fact about themselves that they felt no one else knew. This activity helped establish a relaxed atmosphere.

We we used a time line. We had put a 20 foot roll of brown paper on the wall and listed key events during the six months (e.g., the big snow storm, the hiring of key people). Then in the retrospective individuals put on the key events that mattered to them. We ended up with this long time line with all these sticky notes that documented the things that mattered to individuals.

As each person put up a note they told the group about the event. This generated lots of laughing and feeling that they were together on this thing. It was fun, interesting and informative.

Next we added a temperature reading. People went along and added below the time line their memory of their general feeling at that time.. This is a technique where people say things like I was feeling really good about things here and not so good here and all that. All thirty five people did this. In this particular time line case was only one situation were there was an alignment of feelings. Everyone felt bad when we slipped a date by three weeks. Other than that some people were up and others were down at any particular time. People began to realize to that people were different. People enjoyed this.

Next, we had people tell stories about when they worked on teams that went really well and they loved the experience. This is appreciative inquiry technique.

Bill: Do you mean they told stories about events not just on this project but in general?

Jeff: Yes, in general. Some people called out things on the project. Some people called out things from the other parts of their life. This was fascinating as well. We told stories in small groups of five. Each group picked the best story to present to the larger group. As the stories were told we abstracted out the commonalties. We recorded what were the things that mattered.

A this point in time we were at five hours since we had had breaks and lunch. People were feeling jazzed. They felt they had heard what was important from each other and had expressed what was important to them. So we quit early which always makes people feel good. Hopefully, I will have a chance to go back to the same group after another major cycle and continue this.

People came up to me afterwards and said they had been very worried that we were going to spend six hours talking about what went wrong in the prior six months. There was even one guy who did not come into work that day and started the session on the phone. He got so interested that he drove in to participate live and be more involved. The most senior person there was excited as he valued building a cohesive team. Everyone was pleased with the reverse in expectations.

Report on the Scrum Gathering

I am in Orlando at the biannual Scrum Gathering put on by the Scrum Alliance. The gathering alternates between the US and outside the US and is attended by Scrum Masters, Practitioners, Trainers and Coaches. There are some 260+ folks attending this one.

Jim Cundiff has been running the Scrum Alliance for a bit over a year now and I am glad to report that I am seeing considerable improvement in the professionalism, focus and activity level of the Alliance. The program for this meeting has brought in a number of folks from outside the immediate Scrum community. Ron Jeffries and Chet Hendrickson from XP, Dr. Alistair Cockburn of Crystal, Dr. Mark Paulk of CMM, and Gregory Balestrero from PMI and Jim Coplien of Patterns are all here and adding to a more inclusive feel of the gathering. In particular Mark is beginning some academic research on Scrum and the PMI is 'embracing' Scrum with a large number of PMP folks attending.

The Alliance is showing off the redesigned Scrum Alliance website. I suggest you take a look. It appears to be much more comprehensive and there is a lot of content now easily accessible.

Last night I met with 8 other Certified Scrum Coaches as we considering our next moves. I was involved with the initial definition, construction and roll-out of the Scrum Coach certification program. It is a rigorous certification with a non-trivial application and thoughtful evaluation. One really has to be a Scrum Coach to get certified. Initially there were six of us certified. Now the number is up to 12 and increasing each month. The coaches I met with last night were from Australia, Germany, Denmark, South Africa and the US. I am proud of the work we have done on this program and we are hoping that it grows and becomes recognized as a guide level certification.

With Alliance support we are seeing more and more Scrum User Groups getting started. Now over 50 world wide. And also growing. Some folks have wondered if at over 50,000 certified Scrum Masters if the 'market' was becoming saturated. It appears not to be so. Altho training rates have slowed somewhat, the demand is still reasonably strong. The requests for coaching are increasing as well.

In short, Scrum is alive and well and in fact, thriving.

Achieving Immediate and Continuous Bug Fixes in Agile Development

As I mentioned in our blog overview, most development shops spend noticeable time prioritizing and dealing with bugs. However, that effort only manages the symptoms. It can be more useful in the long run to look at the operational side of development. An interesting issue is the time that elapses from when a bug enters the system and when it is detected. I have been in situations were this time was measured in minutes. Then developers could quickly learn from their mistakes, make immediate corrections, and move on. They learned and over time produced fewer and fewer bugs. Then we did not spend time prioritizing and dealing with long lists of bugs.

Bill and I had a conversation going into more depth on how this might work.

Bill: How do you get the bug reporting time down to minutes? I am used to seeing many bugs getting all the way out to users as an accepted byproduct of software development. Some vendors are famous for this.

Jeff: The first thing you need to do is to establish the practice of continuous build, or better continuous integration. Most successful Agile development
shops have this practice in place. This means that every time a developer turns in their code, the application is rebuilt and tested with the new code. So you are able to quickly see if the new code brings in any unwanted behavior, aka bugs. This integration often takes a few minutes to build a typical change. The book Continuous Integration by Paul Duvall, Steve Matyas, and Andrew Glover is a good resource on this topic.

So the first part is having the ability to quickly turnover the integration of new and old code. The second step is to have good unit tests. These tests should be part of the automatic build. If a unit test fails, the ‘build’ fails. If you have unit tests in the area being changed, the effects of the bad code will show up right away.

Bill: I have often heard of unit tests but usually just assume I know what they are. I think I need to make sure I know what you mean here.

Jeff: Unit test are the lowest level of tests. They test classes and methods and small assemblies. They look at code function to see if it does what the developer wants it to. Functional tests on the other hand look at code to see if it performs what the customer or end-user wants. Unit test are designed to run very fast and look at all the code on a specific section of the application. Test coverage measurements should be made on unit tests.

If you use a technique call “test first,” then no line of code is written except in response to a failing test. That means that there is not a lot of code in the system that was produced without a test informing the code’s development. This gets interesting because what this means that every line of code has a test associated with it. So if something changes the code later the unit test can see if anything breaks.

Because the unit tests are so fine grained you will know right away what the problem is without having to debug. Debugging is figuring out what is wrong and with unit tests you know what is wrong. So if you make a code change and it breaks right away you know right away the cause. You still need to understand it but you do not need to spend time finding it.

So the combination of continuous build and unit test allow you to get to the condition of bugs being found right away. In the particular case I was thinking of we did not need a bug tracking system. In this effort we had twelve developers over eighteen months and only had about five bugs at any one time. There was no need to prioritize bugs. We said all bugs are important and need to be fixed right away.

Bill: So this makes total sense to me. Why doesn’t everyone do it?

Jeff: There are a couple of reasons. One reason is that historically we have considered testing not part of development. We have considered testing as something done by QA, often by a different department. Code got thrown over the wall for someone else to deal with and quite a bit of development was done before testing occurred. Responsibility was passed on and not shared.

The other reason is that we do not tend to measure debugging effort so the cost is not apparent. It is almost always much more than people think. When we do not measure fixing bugs there is not any drive to improve the process. Also, folks just assume there will be lots of bugs. ‘That is just how software is.’ Well, it does not have to be that way and I think it is time to change our attitude.

Bill: So is the concept of immediate fix one of the cores of Agile development?

Jeff: Let’s say that most successful Agile teams just do it. One of the practices in extreme programming (XP) is the concept of collective code ownership. In this case an individual does not have ownership of code. So no one can say, “well I cannot fix this code because it is someone else’s code.” So when in Agile we all own the code so if I see a bug, I just fix it.

Lightning Interview by Ward Cunningham

I thought it might be fun to show you an interview that Ward Cunningham did with me late last year. For those of you who do not know of him, Ward is a brilliant developer. He has many claims to fame including the development of CRC (Class Responsibility Collaboration) cards as a design tool and invention of the wiki.

I had the pleasure of working with Ward in the late 1980's at Tektronix in Oregon. We were working on Smalltalk and working on Patterns and the beginnings of the development style now known as Agile. Other interesting folks were at Tektronix at that time including Kent Beck, Rebecca Wirfs-Brock and Norm Kerth. It was a heady time and place with lots of passion and lots of great software.

Ward and I have kept in touch and this interview was an early one in his lightning series.