Thursday, July 28, 2005

MSF for CMMI Receives Positive Feedback

The first industry review of MSF for CMMI Process Improvement has appeared on Devsource. Here are a few quotes,...

“What many people don’t realize is that if your organization is rigorously applying the principles of the Agile software development methodology, you’re already doing much of CMMI without knowing it,” says Bill Phifer, an EDS fellow and SEI-authorized lead CMMI appraiser.

“One of our main goals with the CMMI toolkit was to make the process of generating the artifacts [documentation] necessary for an assessment to be a natural outcome of using the product,” says Sam Guckenheimer, a group product manager on the Visual Studio team.

“VSTS uses several different approaches to help an organization get to a certain level with the target goal of Level 3. First of all, we have tried to instrument the process and treat evidence gathering as a by-product of the normal development cycle. Secondly, we use reporting as the artifact generation tool and not document writing,” says Guckenheimer.

Read the full story…

Caption Competition

What is agile ‘softie Ward Cunningham teasing colleague Randy Miller with at the Agile 2005 conference? Click the comments button and add your suggestion.

Wednesday, July 27, 2005

Stretching Agile to Fit CMMI Level 3

Here is the paper and the presentation slides that I presented today at the Agile 2005 conference.

Download Stretching Agile to Fit CMMI Level 3 - the story of creating MSF for CMMI Process Improvement at Microsoft [Paper PDF 342K] [Presentation PDF 829K]

I have to say that I was very pleased to see the room full - there is so much great stuff at the conference and to get 30 or 40 people for a CMMI experience report was a really good turnout. The audience also really seemed to like the presentation and greatly appreciated the work we’ve been doing to articulate the intent of the CMMI and make agile software development practices acceptable to the appraisal community.

[Update: Dec 14th 2005. I got some review comments from Jon Gross at the SEI. I’ve incorporated most of Jon’s comments and issued an updated version of the paper 1.6]

Sunday, July 24, 2005

Antibodies Swarm on Agile Business Unit

In my personal opinion, the death of Sprintpcs.com was a great loss to the shareholders at Sprint - even if many of the employees of the PCS unit rejoiced at its demise.

There have been reports of agile managers been axed in the literature and lore of agile development but I wonder whether .com is the first recorded death of a business unit?

.com was built to be agile - our VP used to talk about the 50 metre rule and the 5 metre rule. These were the maximum distances that customers could sit from developers and developers could sit from team members. Of course, they weren’t to be taken literally - no one was out with a measuring tape. But if you wanted to find an entire business unit with on-site customers and where customers met developers regularly then .com was it.

As I stated yesterday, .com was unusual within Sprint. It had every function it needed to operate. As such, it was seen as trampling on others territory. The argument that “layer 7 was different” didn’t cut ice with the territorial guys who would sit in a presentation and then at the end, stick up their hand, point at the architecture diagram and say, “so which box will I own?” Yep, if you couldn’t mount it in a 19” frame and monitor it with SNMP then it wasn’t anything a telco guy could understand. The idea that you would build an entire business model on layer 7 of the IP stack and all you needed from below was an IP connection and quality of service in layers 3 and 4 was bewildering to most people. .com’s marketing people were intruding (so others thought) on marketing elsewhere - this was easily fixed - everyone reported up the CMO. .com’s other development groups were intruding on IT territory with the exception of my mobile portal, wireless application manager group which was intruding on the network group known as TASD and on IT and to some extent on strategic planning and marketing. Luckily those latter two groups actually respected our efforts and encouraged us.

It’s easy to look back now and with the benefit of recent learning in tribalism in the workplace to see what went wrong. Sprint PCS was divided into tribes. The super-tribe of Sprint or PCS wasn’t strong enough to galvanize people together. Everyone was too comfortable. It was the telecom bubble. The stock price had soared to about $70 and then flopped a bit to $45. No one could see the $1.75 low coming. No one was threatened. They weren’t even afraid of competitors. The whole market was expanding. It was a land grab. Salaries were high. Bonuses were high. Fast track promotions were the norm. As an example, the CTO was in his early thirties and had joined the company 7 years earlier as a mid-ranking engineer. There was IS+ IV+ TS+ TV+. Time to back stab and in-fight. Who needs a common enemy when you can have one down the street in the next building? right?

Into all this came .com - a self-contained, vertically integrated business unit, in a functionally organized company. It was an alien and the antibodies swarmed on it. .com was its own tribe. I’ll try to update this post with a picture of me wearing my .com T-shirt. It had its own building away from the Sprint campus in Overland Park, Kansas. It had its own rituals, its own values, its own skillsets and tools - it wasn’t a telco, it was a pure play Internet company. It made software for layer 7. Period.

And it was agile. It was using Togethersoft’s Together product and Java and a collection of other IDE and testing tools that weren’t the Sprint corporate standard - Sprint was a Rational shop and still is. It was doing Feature Driven Development at an integrated program management level of competence. It was providing transparency with monthly operations reviews. And most of all, it was turning out a lot of product. So what, if there was criticism that a few high profile projects were late and/or buggy. When taken in context of the costs and productivity per employee, it was way above that of some other business units. There was envy of .com’s elitism. There was envy of its freedom to do its own thing. John Yuzdepski had galvanized a strong tribe. In the end it had to go. The end was fairly swift - a decapitation. The VP was removed. I can’t and won’t go into the details. His successor tasked with winding the business unit down and breaking it up - putting back all the integrated functions into their functional silos once again.

Of course, it is all water under the bridge now. Since then Sprint has undergone a major reorg with the disappearance of the PCS tracking stock and the realignment of the business into a consumer and corporate units. Now, it is going through its own super-tribe problems with a merger with Nextel. I imagine that those who successfully eliminate the .com antibody will find it somewhat harder as the Nextel and Sprint tribes fight for power and supremacy in the new larger entity.

If anything underscores the need for agile management at an enterprise level, it has to be the observation that a Fortune 100 (approximately) company can go through three major reorganizations in only 5 years during which time it killed or sold significant pieces of its business whilst initiating several entrepreneurial ideas (of which .com was just one). Change is the norm. There really is no such thing as TS+. TS+ is an illusion.

New Org Behind DoI Announced

The Agile Project Leadership Network is the new organization formed to promote the teaching and practices of agile and adaptive project management. It’s the organization created to reflect the values in the Declaration of Interdependence.

Membership is open. Join now and get a discount. Learn more about the APLN at the Agile conference in Denver this week.