Mathematicians have the concept of imaginary numbers. As odd as that may seem to non-mathematicians, it makes one wonder if achieving capability level 5 is similarly artificial and elusive concept for a number of organisations?

Mark Paulk, who led the team that developed "the CMM", recently deliver a keynote address at the ISSEC (co-located with PMOz) conference. Significantly, and unlike many management fads that have come-and-gone, the CMM has grown over time and been embraced internationally. Mark is now a Senior Systems Scientist at the Institute for Software Research, Carnegie Mellon University, and Alinement Magazine caught up with him to get his thinking on process improvement, measurement and, of course, the CMM brand.

What do you see in the future of high maturity?

I’m not sure about the evolution in CMMI/15504. There are a number of different agendas. I advocate organisations to think about the idea of choosing methodologies that are most effective and efficient for different environments (which has been part of CMMI for 20+ years). Agile is causing people to come to grips with what was put into CMM 20+ years ago (that is, that the practices are informative and expected, but the requirements are the goals of the process areas). SEI primarily deals with DoD contractors and large companies, so the advice is structured in that way. We meant for commercial shrink-wrap and small organizations to adopt the CMM work too. The process maturity profile shows that more non-DoD than DoD organizations have been picking up the CMM to use since about 1992, so the modelt is seen as useful to all kinds of software organisations. The guidance in the model focuses on high ceremony or formal processes because they were the biggest adopters of the original models. The CMMI user community includes organizations with 25 people or fewer also. CMMI for Development v1.3, published in November, added informative material on using agile practices.
Another orthogonal area is deploying innovative ideas. Building innovative organisations is outside of the CMM scope since the focus is on operational excellence, as is perhaps more important for custom software development than commercial shrink wrap. There is much literature on the kinds of factors that affect creativity and innovations in organisation – in the future that might be incorporated or addressed in separate models
Alignment with Six Sigma at corporate level is out of scope for CMMI but important for implementing the higher maturity levels. Multi-model integration frequently has to deal with ISO 9001, CMMI, and Six Sigma – and they do mesh fairly well while emphasizing different aspects of improvement.

What is the most interesting IT/process statistic you have come across?

There are many papers with data analysis. Of particular interest is the impact of the adoption of Personal Software Process (PSP) on process discipline and individual differences. At the beginning of the PSP process, the top and bottom performers have a ratio of 4/5:1 for defects and productivity. As you work through the PSP process, the bottom quartile of the class will eventually perform better than the top quartile although the difference continues to be about 4/5:1. In other words, the best people remain the best. The strength of the process is that variation goes down and performance improves for all individuals.

PSP is a “structured software development process that is intended to help software engineers understand and improve their performance, by using a "disciplined, data-driven procedure"” Mark added that:

PSP cuts across cultures and nations through the commonality of evidence-based thinking and skills building.

TSP is an “operational, “how-to” implementation of the principles and best practices that the SEI has advocated for over a decade.” Does PSP and TSP suit a specific culture? Mark characterised the ideal individual temperaments:

PSP and TSP might fit better with people who are inclined to use data to drive decision-making. People who are more intuitive may not find PSP/TSP very appealing.
National cultures are not so much defined in terms of the measurement-driven, evidence-based decision-making encouraged by PSP, as for TSP where there are differences in how people link together on a national and ethical level. In the Hofstedt model, five dimensions are used to characterise different national cultures. Power/status vs individualism/collectivism vs risk aversion vs confrontationalism vs short/long term perspectives. These cultural preferences indicate differences in how teams and organisations interact. The problems we face may be similar, but the solutions may differ depending on the individual and cultural preferences that folks have.
The Australian culture is reminiscent of the US culture, although much of what is done here is flavoured by Asia/Pacific cultures.

Talking at ISSEC, Alinement asked: Did you find the Australian audience had a different focus from your regular audiences?

No difference. The Australian audience was similar to others in interest in the kinds of questions that came up and issues of software and systems professionals were similar. Looking at the worldwide technical community, there is a lot of communication and alignment.
There are some cultural/ethnic differences in how issues are addressed, but this is also problem-driven by the complexity of the problem being solved.
With information availability and its free flow, solutions are reasonably available within the technical community.

As a result of offshoring, is there a national culture and how does this impact on eSourcing/service provider model?

The eSourcing Capability Models have more emphasis on relationship management compared to CMMI or ISO 9001.
Long term ongoing relationship needs to expend a noticeable amount of attention on cultural differences management. This is true of both national and organisational cultures.

Translated into process terms?

No specifics, but we need to think about the relationship issues. There is a difference between a US organisation working in France and a US organisation working in Japan. There is no simple recipe, only general guidelines on looking at national cultures.
There are courses on how people can work effectively in the context of different cultures.
Cultural differences need to be considered when making agreements and communicating with each other.

What is the one common mistake that all organisations make when doing process measurement or improvement? Mark had so much information to share!

A whole set – I’ve written a paper on common pitfalls and statistical thinking and other papers on successful measurement programs.
Common mistakes include:

• Biting off more than you can chew. Start small.
• Not good definitions. Be cognisant of the operational definitions of the measures you are putting in place.
• Use a goal-driven measurement approach (which is a variant of Vic Basili’s Goal Question Metric approach).
• Use goals to motivate collection of reliable data and analysis in making decisions and informing of the work of the organisation.
• Be wary of using measurement for motivational purposes (Rob Austin describes how it encourages dysfunctional behaviour and how to deal with measurement to avoid undesirable impacts)

How does the maturity of IT service organisations compare to IT software organisations?

There isn’t enough data to be statistically significant.
Many organisations picking up the CMMI-SVC or the eSCM for Service Providers already use CMMI for Development.
Organisations having existing capability in part of an organisation can adopt relatively easily another mode because they already have the basics in placel. Tier 1 organisations, e.g., IBM and Accenture find it easier.
Everyone else goes through the same learning curve of SW-CMM.

See Part 2 of this article for the conclusion to this enthralling interview where we explore agile and sourcing!