My coaching and consulting work is primarily with large companies and, as such, my approach is informed by that experience.  Rather than taking a purist, monolithic, and one-size-fits-all stance on adopting agile practices, I tend toward an evolutionary, organic growth approach which recognizes the give-and-take, assimilation-accomodation, nature of any kind of organizational change. 

Following this  very general heuristic, 'agility' constitutes a set of guideposts and beacons to help organizations and teams grow their own agile methodology, so to speak, more than it it does a fixed 'end state' to arrive at.   This seems to be all the more true in large, complex companies.

There are many different perspectives on the question of agile 'purity' and the issue is a very subtle one.  I will attempt to address the many facets of this topic in my blog over the coming months.

In my work with organizations, I tend to approach agile adoption as a learning process that has four aspects.

1. It is iterative and incremental.
2. It is native and organic.
3. It is holistic.
4. It is multidimensional.

I'll only briefly describe these here--fuller descriptions will emerge in writings on the the blog.

1. It is Iterative and Incremental

Pretty much all of the successful adoption strategies I've been involved with or read about tend to have an iterative and incremental character.  Architect Christopher Alexander uses the term 'piecemeal growth' to describe an approach to "growing" architectural structures, as opposed to designing entire architectural edifaces up-front.  Another term that describes this aspect is the term 'emergence.'  However we might term it, I believe iterative/incremental and emergent approaches work for organizational change:

Start with something small and learn what you need to know to get started.  Do a small part of a small project, learning as much as you can as you do so: what's working? what's not working? What impediments are we running into? What do those impediments tell us about ourselves as individuals and as an organization?  And so on.  These questions become the basis on which you form hypotheses.  These hypotheses, in turn, form the basis of alterations you make in your next iteration.  Following this iterative approach, each iteration will bring (a) greater learning and (b) increased competence and agile capability.

This iterative and incremental approach is very similar to the organization development practice of action research.

NOTE: Iterative/incremental growth as a stand-alone principle of organizational change probably won't work.  In the approach I tend to take--and which many other change agents take--iterative/incremental growth is to be combined synthetically with Holism, Organicicism, and Multi-dimensionality, as described below.

2. It is Native and Organic

Changing to a new way of working takes time.  Successful companies have evolved powerfully invested cultures and social structures.  These constitute the very heart of any given company's success and so people will fight hard to preserve them.  An effective agile adoption strategy will therefore respectfully observe the strengths already present and endeavor to build upon and amplify those very strengths.

Psychologist Jean Piaget described human development as a process of assimilation and accomodation.  At times, one assimilates a new thing or experience, which means that one changes something in that thing or experience in order to reduce the stress of dealing with it.   At other times, one accomodates oneself to a new experience or situation, which means they alter something in themselves.  It is this give and take which characterizes human growth and development: over-accomodation results in a breakdown of who one is whereas over-assimilation results in no new information entering the system, and hence little or no learning.

I believe we can make similar assessments regarding the Agile adoption strategy.  Over-accomodation results in the over-stressing of the current organizational culture, with likely concommitent breakdown in the functioning capacity of the organization.  On the other hand, over-assimilation can render little new learning and hence little organizational growth.  This latter situation (over-assimilation) seems to be a far more common experience: it is very common indeed for companies to out-and-out reject certain agile practices as 'undoable' or 'impossible' merely because managers and stakeholders are as yet unable to see a new way to cross over that hoop.  For them, something has become 'impossible.'  

It is critically important to point out, however, that when things get hard, or seem 'impossible,' we are given a uniquee opportunity to see something about our current culture and process that was previously invisible.  As long as we remain unjudgemental and noncondeming of what we see, such a moment can become a catalyst for choice.  It is a moment when honest, reflective inquiry can help people work together to see a new way of thinking and organizing.   Only then, I believe, does authentic choice become possible.  This is one reason why  'pushing harder' toward some process end-state by zealous 'change agents' can only make things worse; people dig in their heels, and we lose the benefit of openness and transparency.

3. It is Holistic

Effective change seems to involve the whole system.  Some companies approach agile adoption as a kind of grass roots affair, in which individual teams and their members spearhead, lead, and negotiate the adoption effort.  As they accrue early successes, they often find themselves, bewilderingly, facing at best a leadership that is ill-equipped to help, and at worse, unwilling to permit further growth of this new way of working. However, at the other extreme, we often see senior executives announcing a major agile 'initiative,' calling on their troops to realize 40% agility in the next year.  In such cases, people are usually not overjoyed to do what they see as the lion's share of the hard work of change.  Insiduous failures are introduced which eventually derail or, at best, significantly water down the entire endeavor.

Effective agile adoption seems to bring together various levels and various functional groups in an organization. We have to be especially careful about how we describe this aspect; there are a lot of caveats.  For instance, in VERY large companies, entire company-wide adoptions from  the very beginning may be ill-advised; better to stick with one department or unit at first (this articulates the iterative and incremental aspect).  Nevertheless, whether within a unit, department, or entire organization (if that latter is not too huge), you'll want to employ a cross-section of people's involvement.  This will bring in the variety of views and perspectives needed to continuously experiment and assess (as, again, the iterative and incremental aspect informs us), particularly during the early phases of an agile adoption strategy.  There are a number of tactics and practices for doing this (e.g. action research, appreciative inquiry, future search-type meetings, etc.).

4. It is Multidimensional

This one is a bit more subtle.  Multidimensionality points explicitly to the multiple perspectival views required to emergently design and evolve a transition to agile ways of working.  Agility, as performed within the context of a software development project will have a very different set of concerns, practices, and perspectives than agility as it might be practiced at the level of senior management.  This aspect directly addresses the different 'horizons' which different performers and stakeholders will have with regard to what it means to be 'agile'. Senior leaders will need to develop similar though different sets of personal capabilities than will developers or testers.  At the different management levels, it would almost be safe to say that what constitutes 'agility' will mean very different things depending on where you are functioning within the organizational hierarchy. 

This can sometimes be controversial topic in the agile world, in which some of us have come to regard agile as a way to liberate ourselves from command-and-control managment, and the hierarchical structures which seem to enshrine its existence.  Nevertheless, I feel that this is a critical aspect of the success of an agile transition, especially for large and complex organizations.