Logical Levels and Statistical Games: A Powerful Strategy for Agile Adoption

Level: Expert

The use of metaphorical games as a strategy for adopting an agile culture has shown to be weak because most of trainers don’t know the principles of changing beliefs and values of a human mind. The Neuro-Linguistic Programming (NLP) Logical Levels of Learning and Change (LLLC) is a powerful framework to be considered when we need to challenge skeptical and analytical minds in traditional software development environments. This tutorial presents the structure of the LLLC and a set of three statistical games that can be used as a seed for an agile training or a movement into the agile culture.


Neuro-Linguistic Programming (NLP) is a theory of language, communication and thought that provides certain principles and techniques to improve the way a human being can interact with the world and produce better results. Since I’ve started practicing NLP and improved the way I used to gather and analyze business and software requirements, I decided to extend the benefits to other people teaching requirements engineering classes based on NLP approach. The results were impressive and I decided to apply the same concepts to move my new traditional team into the agile culture. Again, a structure of change and some mathematical facts challenging the analytical minds were powerful enough to bring everyone to a new paradigm.

The NLP Logical Levels of Learning and Change (LLLC) is a model introduced by Robert Dilts in the mid 1980s and it refers to a hierarchy of levels of processes within an individual or group. The function of each level is to synthesize, organize and direct the interactions on the level below it. Changing something on an upper level would necessarily “radiate” downward, precipitating change on the lower level. Changing something on a lower level could, but would not necessarily, affect the upper levels. These levels include (from the highest to the lowest): 1) Identity, 2) Beliefs and Values, 3) Capabilities, 4) Behavior, and (5) Environment. [Dilts, R. & DeLozier, J. Encyclopedia of Systemic NLP and NLP New Coding, NLP University Press, 2000]

Based on this model, we can say that in whatever organizational environment, bad or good results can be attributed to bad or good behaviors, respectively. Bad behaviors can be attributed to human incapabilities (skill limitations) and good behaviors can be attributed to human capabilities. Most of human capabilities and limitations can be attributed to empowerment beliefs or weakening beliefs, respectively, where all of them are supported by personal values. Finally, personal values are the essence of identity, a self concept that sustains a purpose in life. So, the most effective way to promote a change in an organizational behavior is to purpose a new set of beliefs based on real benefits while keeping the main identity and personal values. But to do that, the instructor must be mindful about the language used in agile training.

What we intend to discuss in this tutorial is that most of people participating in a metaphorical game, based only on practices focused on behaviors, report this kind of event as an enjoying dynamic but not necessarily an effective way to change their minds. What we would like to show is that behavior can be easily changed if communication is effectively done in the levels of beliefs and identity.


Mathematics is a language understood and respected by every computer professional. Mathematical facts are strong evidences that can be used to challenge unsustainable beliefs. The strategy of games suggested in this tutorial is to run experiments, analyze statistical data and associate bad results to bad identities, typically related to traditional software development. By doing this, it is possible to trigger bad feelings when participants repeat bad behaviors (traditional development).


To be presented in 90 minutes, this tutorial has to be limited to 30 attendees divided in 6 groups of 5. Six tables will be necessary to accommodate all participants and allow them to write records and play dice during the experiments. All material will be provided by the instructor. Only a projector must be provided by the organizing committee.


1) Introduction [Duration: 20 minutes]

The audience will be questioned about feelings they’ve got from typical games played during agile trainings. Participants will be asked to analyze his or her instructor’s communication strategy. The LLLC will be presented as a better communicating structure to be used in promoting changes during training.

2) Game “Large Lots x Small Lots” [Duration: 20 minutes]

In this first experiment, participants will play a simple game that simulates two types of production logistics: large lots (waterfall) and small lots (agile). The five group members will act as a Customer, an Analyst, a Designer, a Programmer and a Tester. The main goal of this game is to deliver to the customer, as fast as possible, 10 features solved and signed in a single paper containing a particular task. Based on the concept of a JIT production, we show how it is possible to almost triplicate productivity working with small lots. A “waster” identity will be explored in this game with people who continue to develop in a waterfall manner. Data will be collected from the experiment and statistical facts will be discussed with teams.

3) Game “Dependent Events x Productivity” [Duration: 20 minutes]

In this second experiment the same groups are asked to deliver 40 features in 10 iterations simulating the same production logistics as the previous game (analysis, design, coding and test). At this time, each step will have a capacity determined by a die. Participants will testify the appearance of inventory piles (waste) and team velocity will become smaller than individual velocities. This game is based on Eliyahu Goldratt´s Theory of Constraints and shows how statistical fluctuations from dependent events lead to the performance of the system becoming worse than the average capacity of the constraint. In this case, we associate an “individualist” identity to the specialized way of work because a team made by generalists (agile developers) can deliver the whole project running at the lowest team member velocity. Data will be collected from the experiment and statistical facts will be discussed with teams.

4) Game “Product Backlog and Velocity Forecasting” [Duration: 20 minutes]

In this third experiment the same groups are asked to deliver 50 User Stories completed in 10 iterations time boxed by 2 minutes each one. At this time, each group will work as a team and productivity will be measured as the amount of User Stories delivered per iteration, where a User Story is a fact from each participant life. Participants will testify similarities between statistical data from each group behavior, allowing forecasting how much time is necessary to complete the whole project. This game is based on Queueing Theory (QT) and shows how productivity can be predictable when we see the whole process. In this case, we associate a “team” identity to a desirable behavior. Data will be collected from the experiment and statistical facts will be discussed with teams.

5) Final Discussions [Duration: 10 minutes]

By the end, participants will be asked to discuss strategies, games and their feelings about the dynamic.

This session was presented last year in the First Latin-American Agile Conference (http://www.agiles2008.org/en) and it was well received by all attendees who awaked to the power of changes based on the LLLC model.

Learning outcomes
  • Understanding of how the NLP Logical Levels can help us in cultural changes
  • Understanding of how mathematical games are more appropriate for changing beliefs from analytical minds
  • Learning of three mathematical games based on JIT, TOC and QT for a faster agile adoption
  • Changes in beliefs and values for a better concept understanding
Featured participants
Primary target persona
Other target personas