Why can't a role be a computer or a database?
Work is a social activity. Humans may create technologies that mechanize certain tasks, but machines do not make their own decisions about which activities they engage in. Only people make those decisions, determining what activities and transactions are important, and assigning the tasks either to real people, or technology enablers such as applications that can complete the tasks.
But don't applications sometimes take the place of humans to play roles?
Yes, they can and do. As technologies become more sophisticated a software program may well be capable of filling a role. An example might be using an online reservations program to book airline travel. In that case the technology supports fulfillment tasks by acting in the role of "travel agent" or "reservation service provider." So even though a software program is capable of filling that role it is fulfilling it based on the way a real human would typically behave in that role. The technology is merely a mechanism for a role to be executed. Focus on the role first; then consider what might be the most effective mechanisms to support the role activities.
I work in a small company and we all play multiple roles. Will this approach still work for us?
This is actually an ideal approach for a small company, because it helps people avoid the trap of organizing the company around key people rather than around the roles that are core for the business activities.
How do we represent roles the organization is planning to add?
You would first do the "as is" map of the current value network. Then you can create a "future state" map where the new roles are defined.
What if we have more than 8-10 roles?
You can of course create value networks with more than 8 roles. However, when you are using a visual map it is challenging for people to read more than around twelve roles and 50 or so transactions. With technology support, much larger networks can be combined. Even when visuals are too complex to read, other performance indicators provide insights into network performance.
What if we have loops going back into roles?
VNA is focused on what happens between roles, not within them. If you are creating circular loops then you probably have two roles packed into one and need to separate them. Or you may be drilling down into a level of detail that is not really necessary for the level of map you are working with.
What if we have different understandings from multiple people playing the same role?
You will need to reconcile the differences either by aggregating their responses or having the people who play the role come together to define it together.