A few weeks ago I was lucky enough to spend a week in Toronto at the Agile 2008 conference. Here’s my summary of proceedings…

Agile 2008 Summary

• Something like 62% of the attendee’s were new to Agile.
• The race is on to monetise Agile! Oh dear.
• The Agile community is growing both in terms of expertise and adoption.
• There were a huge number of presentations on Lean and related topics - it’s a growth area.

Two areas of focus seem to be emerging beyond applying tools and techniques at a team level. The first is the development of models to introduce Agile processes to the enterprise in a structured way. The second is scaling Agile processes to an enterprise level - there seems to be a growing recognition that only introducing change at a team level is not going to cut it in the long term.

One gem of information came courtesy of Sue McKinney, a VP in the product division of IBM. Sue and her team started to introduce Lean concepts in 2005 across a number of teams (including Lotus Notes & CICS) with some help from the Poppendiecks. Sue stated that to date their efforts have saved them $90.6M (USD)!

The Wisdom of Crowds - James Surowiecki (Keynote)

James delivered a really interesting keynote that seemed to be a one hour summary of his book The Wisdom of Crowds. James explained that under the right circumstances the collective opinion of a crowd can be far superior to any individual expert within it. The story he told began with a chap called Sir Francis Galton who developed a social philosophy called Eugenics. Apparently Galton was a big believer in the ‘individual expert’ until an experience changed his mind. One day he was at a village fete and observed a competition where folk were guessing the weight of a cow - the closest guess got to take the cow home. Galton somehow got his hands on all of the entries following the competition and found that the average of the entries was better than any individual guess. He repeated the experiment and continued to get the same results.

James went on to explain that this is a consistent phenomenon, however some criteria are required to make it work. For example there needs to be an effective way of aggregating the collective judgement and there needs to be a certain amount of thought diversity within the group. Groupthink and dominant characters can lead to poor results.

James tied all this back to IT by talking about estimating and planning - these techniques have been used in other areas to improve forecasting. It’s really a very detailed and effective explanation of why Broadband Delphi estimating works and the conditions required to make it as effective as possible. I’ll definitely get my hands on the book.

Expanding Agile Horizons - Mary Poppendieck

This was an interesting presentation from Mary. The presentation was effectively a journey through the history of software engineering starting at the 1968 NATO software engineering conference up to our current position. Mary used an interesting metaphor of plank roads to structure the presentation, asking if Agile is a plank road. Apparently plank roads were very popular in the early US and were very effective at serving a short-term purpose. However, they didn’t last and were quickly replaced by more durable replacements. Mary seems very focussed on whether Agile is sustainable and where things will go next for the community.

Ten Terrific Transition Tips - Joshua Kerievsky

This was a great presentation. One of the best of the event for me. I found a lot of the materials at the conference targeted at an introductory to intermediate level or worse, a thinly veiled sales pitch (more on that later). This presentation was clearly from someone who is dealing with the same challenges that I am at an organisational level and coming up with innovative solutions to them. Joshua spoke about applying transition readiness assessments, providing managerial and technical coaching, defining team values, managing change and a whole host of other areas related to transitioning to Agile. What will I take away? Well, I’ve got notes to take another look at Virgina Satir’s change model (and the Dreyfus change model from a conversation with Dan North) and I’ll probably buy a camcorder!

Money for Nothing - Jeff Sutherland

The abstract on this presentation promised details of contract structures for Agile projects. I was expecting innovative ways to develop risk-reward pricing and encourage the right behaviours in customers and service providers. 45 minutes in and Jeff had talked through the Nokia online SCRUM assessment and showed no promise of ever getting on to contracts - I left and caught the last half of Mike Cohn’s presentation on Agile Estimating and Planning. Dire Straights indeed.

Experience Reports - Channel Advisor, GAP, Babycentre

I went along to three 30 minute experience reports where representatives from Channel Advisor, GAP and Babycentre each discussed their experiences in introducing Agile. It was very interesting to hear the real-world experiences of three very different businesses, particularly around the similarities and differences in the challenges. Here are a few lessons that I took from the presentations…

• Take a pragmatic approach to applying practices based on value.
• Until you know how to collaborate don’t use supporting tools.
• Create generalist by rotating within and through teams.
• Experiment to find the optimal team size.
• Invest heavily in continuous integration.
• Start with an ambitious pilot.

Don’t Sell Buzzwords - Richard Sheridan

Another really interesting presentation this one. This time because Richard explained his social experiment in creating his own company with a very unique culture. Richard is the boss at Menlo Innovations; a smallish and growing web application shop. Richard spoke about the culture that he’d attempted to create, how he’d gone about building it and how he maintained it. It gets particularly interesting around areas such as the recruiting practices (assessment centres are run in large groups of 40/45 where candidates must pair and take part in other Agile activities - resumes aren’t even reviewed until the 3rd stage), permanent to contract staff ratios (it’s around 2/3 in favour of contract staff), client relationships (clients have to take a one day course to work with Menlo) and incentivisation (developers get paid per story!). What was clear was that the Menlo folk that Richard had brought along were very happy and enjoyed working at Menlo. Richard has plans to keep his business growing and I wish him all the success in the world; he’s a true innovator.

Scaling Software Agility - Dean Leffingwell

Dean was one of the few presenters at the conference that spoke about managing change at an executive and organisational level - much of the focus was around introducing change at a team level. I took a lot from this presentation. Dean has a long and successful history in the IT industry and has recently published a book on Scaling Software Agility: Best Practices for Large Enterprises. Dean seems to apply a combination of the technical practices of XP and management practices of SCRUM to good effect. In the presentation Dean presented a model built on 7 practices for scaling Agility to the enterprise level. My notes are a bit scruffy when it comes to working out exactly what they are but here’s what I think I wrote…

1. Define / Build / Test component teams (cross-functional).
2. Master the iteration.
3. Apply two-level planning and tracking.
4. Smaller more frequent releases (fix the dates).
5. Intentional architecture.
6. Lean requirements at scale.
7. Component Agile is not system Agile.

These are probably wrong (but still interesting triggers for thought). I’ll be getting hold of the book so I should be able to correct myself soon - assuming someone else doesn’t correct me sooner!

Future Directions for Agile - Dave Anderson

This is a hard one to write about. I find that I am either in passionate agreement or violent disagreement with Dave’s views; there’s no middle ground! So, I thought I’d give a quick synopsis and then start to tackle things. The presentation, as the name suggests, asked the question of where we go next in the Agile community and where our opportunities lie. Simple enough? Yes and no. I agree with Dave’s views that scaling Agility to the enterprise level is our next big challenge. I also agree that the Lean world has lots to offer in helping solve this problem. I think where we have contrary views is in the ‘how’. Dave challenged the Agile manifesto on the basis that it’s superstituous and not ’scientific’ - I think this is missing the point on what a philosophy is. Dave continued to talk about how he thinks tools like Kanban, Real Options Theory and CMMi are the solutions - I agree that they’re part of the picture but I’m pretty sure they’re no silver bullet. What Dave is doing well is applying some innovative tools to a high level of maturity and success.

Dave argued that we should embrace division of labour rather than challenging it and use scheduling practices to make activities work as efficiently across the divided labour. He is also actively involved with the SEI in writing the positioning of Agile practices for CMMi - this will be interesting to follow up and seems quite a controversial step within the Agile community. In fact at the start of the presentation Dave said he’s been trying to speak at the conference for a while but this is the first time that he’s been approved.

I think Dave is a great character to have around the Agile community. He’s controversial and he challenges everything which is great and although I don’t agree with all of Dave’s sentiments I think it’s good to have someone pushing us and keeping us honest.

5 Success Factors for Enterprise Adoption - Michael Mah

Michael is an analyst for QSM Associates and Director of Benchmarking at Cutter. Michael presented some case studies from BMC (Distributed SCRUM) and Follet (Collocated XP) against industry averages from Cutter archives of 7,300+ projects over 20 years. As you’d expect it told a good story compared to the industry averages, with Follet saving $6M over 6 release compared to an industry average project of it’s type and size. At the start of the presentation Michael stated that CxO’s really have three options when it comes to reducing IT project costs…

1. Send it to India.
2. Go Agile.
3. Go Agile and send it to India.

There are a number of papers and articles that Michael has authored on the subject that can be found on the QSM Associates site.

Economics of Agile Development (IBM) - Sue McKinney & Ted Rivera

I wouldn’t usually attend an IBM presentation at an Agile conference, or any conference to be honest, but this time I did and I’m glad. As I said at the start of this post, Sue McKinney told how IBM started to introduce Lean to their products division in 2005 and estimate that they have saved $90.6M to date. IBM are working with large, distributed teams often using dated technology (CICS) and seem to be getting really good results. IBM have developed an internal workshop process that they are rolling their staff through. If a large, distributed and often bureaucratic organisation like IBM can apply these ideas and save $90.6M then there is a real hope for everyone else.

Kanban Systems - Corey Ladas

I like Corey. His presentation was interesting and his blog is good too. It’s taken me a while to get my head around what Corey and Dave Anderson have been calling Kanban. I learned Kanban in the manufacturing plants so I’ve got a pretty strong view of how it works and I’ve always felt that using story walls on Agile teams gives us as much of a Kanban process as I’d ever want. Having seen Corey’s presentation it’s more clear to me what they’re doing. Firstly, I have to say that it’s not that different to what a lot of high performing XP teams are already applying. Corey talked through a few examples on teams of differing maturity at different stages of adoption. Effectively what I saw was a scheduling solution. My view is that we have 2 options at a team level. We can drive towards poly-skilled people on cross-functional teams which will minimise our requirement to perform person scheduling as we have few knowledge and skills silos (my preference for lots of reasons) or we can maintain our seperation of duties in which case we have scheduling issues for individuals time which Corey is using Kanban techniques to manage. Corey and Dave argue that often skills separations are necessary for legal and regulatory requirements or organisational reasons; I think that if you have this challenge then the scheduling tools applied here will help. So far I’ve always been able to drive towards poly-skilled teams, even in highly regulated banks. If you haven’t read Corey’s Lean Software Engineering blog then I recommend you go and check it out for an alternative perspective on Lean in IT.