XLP brochure

Introducing XLP

Ben Koo Department of Industrial Engineering, Tsinghua University

Jan Van Maele Faculty of Engineering Technology (Campus Group T), KU Leuven (University of Leuven)

Wei-Tek Tsai

Ron Rusay

Gautam Dasgupta

Context: Why it is needed

We live in a digital era that calls for a paradigm shift in the way we learn . Throughout everyday experience, people with access to the internet have the natural rights to integrate a set of open source e-tools into a unifying workflow. Learning is data-driven: all social transactions are digitally captured, providing ample data for analysis and feedback. In this way, people of all ages need a repertoire of competences that are generally applicable to solve the new breed of problems in this world saturated with so many computing and communication devices. A highly digitized and networked society is filled by diverse diversities. A good learning environment should be populated by students, teachers, and professionals from various places, disciplines and age groups to enhance learning at the group level. Indeed, the complex challenges we are facing as a global community can only be solved by working together with people who differ in the perspectives they adopt and the competences they have acquired. We are also living in a world direly in need of talent. The learning experience of designing and taking challenges provides a training ground for grooming talent. What is more, by consistently measuring both individual and group performance, We need a way, say an Extreme Learning Process (XLP) to serves as a method for identifying much sought after talent for tomorrow’s world on a broad base.

What is XLP

Extreme Learning Process (XLP) is a transdisciplinary learning activity design methodology that organizes learners from different disciplines to design and execute learning activities for other learners. It empowers learners by giving them access to open source technologies to keep track of participants’ contributions, conflicts, and learning outcomes throughout learning activities. The aim of XLP is to become a crowd learning operating system that enables the creation of rich learning ecosystems.

XLP is being developed by Professor Ben Koo at Tsinghua University, Department of Industrial Engineering. So far it has been tested in undergraduate as well as graduate education, at universities, vocational schools and high schools, in different formats such as the 4-day (80 hour) Orientation Program, the 8-week Lab Exploration Program, and the 16-Week Global Manufacturing Strategy Program. Participants have included students, scholars, and ‘hackers’ from Asia, Europe, and North-America.

WHAT IT IS

As a design methodology for learning activities, XLP can refer to educational practice in a number of interrelated ways. It is at once a training ground for developing a range of professional skills, a crowd-learning operating system, and a large-group intervention. These usages of the term can be briefly described as follows. training ground

First, XLP serves as an interdisciplinary and intercultural training ground for all participants, ‘challenge designers’ as well as ‘challenge takers’, who execute the mission they have been given. That is to say, XLP embodies the ambition of developing the right mix of competences that is expected of global professionals and leaders, including team building, communication skills, intercultural competencies, project management skills, leadership skills, information literacy, lateral thinking, cognitive foundations, creativity, and innovation capacity.

One of the reasons why XLP makes for such a powerful testing bed is that it brings together participants from different cultural backgrounds who represent a variety of professional and academic disciplines. Indeed, there is a strong drive in XLP to cross boundaries by involving a diversity of students, teachers, and ‘hackers’ from various places, disciplines, and age groups. XLP also welcomes and integrates practices that originate from a variety of academic fields, including information system design, law, and communication. It is through this interdisciplinary interaction that participants can discover the unique skillset that they bring to the group. As such, the diversity of the group will lead to greater self-awareness of one’s talents. operating system

Secondly, XLP refers to an operating system for cognitive learning at the group level. That is to say, XLP provides an environment within which interconnected sets of protocols are mobilized for data-driven learning. This environment comprises a range of software programs and e-tools that differ in granularity and can be related to the different functions fulfilled by operating systems (Figure 1). First of all, the XLP universe is called into being by granting a digital identity to each agent. In the Internet of Things this universe can be populated by individuals, groups and organizations, as well as by devices and objects. Each of the agents is provided with a unique code from a virtually bottomless supply, resulting in a disambiguated and scalable universe. Secondly, conduct within the operating system is regulated in XLP by virtue of three principal e-platforms. Evernote collects all kinds of data snippets tagged for agent, time, and location as input throughout the XLP event. Git is a distributed file system for recording, storing, and sharing all data through pull (i.e. input) and push (i.e. output) actions. Wiki can be situated at the output end as it provides a platform for digital open-source publishing. Thirdly, sequencing is regulated through Teambition, a web-based project management and collaboration tool.

Figure 1.

This system can obviously only be effective to the extent that all meaningful content is systematically recorded in digitized form and that emerging patterns are identified, analyzed and interpreted. Modeling the various types of social transaction in XLP can facilitate this effort. Figure 2, for example, provides a model of the market trading performance of a given team, building on mathematical category theory. It shows that each transaction results in an updated ‘capital’ (monetary as well as all other assets) that is owned by the team, and that capital is itself a function of the ‘resources’ proper (commodities, products, services, finances, knowledge, skills, ...) and of the respective ‘value’ that is attributed to these resources. Such accumulated data will reveal patterns of changes over time, which can be analyzed for evidence of learning. They also allow continuous feedback to all participants, stimulating self-directed group behavior.

Figure 2. Conceptual model of trading in XLP for a given team

Viewing XLP as an operating system for cognitive group learning is related to the basic tenets of the ‘Probably Approximately Correct’ [PAC] model of learning (Valiant, 2013). The first tenet holds that the coping mechanisms with which life abounds are all the result of learning from the environment. As long as the environment is rich enough, learners will pursue ‘learnable’ targets even in the absence of any teacher. The second tenet states that learning can be done by concrete mechanisms that can be understood by computation. These mechanisms are formally noted down as ‘ecorithms’, i.e. step-by-step procedures for processing information that operate in highly complex environments. Ecorithms make use of generic learning techniques to acquire knowledge from their environment, aiming at effective performance in new situations. In addition, the attention within XLP for crowd learning, time pressure, and realistic context also all find a theoretical ground in PAC.

Both XLP and PAC address learning with regard to ‘theoryless’ aspects of knowledge, where definite answers are not available, for instance the capability of effective trading during an XLP event. TRIZ, which was developed as a method for inventiveness in technical creative work, can offer a useful procedure for learning in such situations. As a starting point, it takes the identification of contradictions between incompatible requirements that prevent the achievement of primary functions, e.g. increasing capital. The cognitive foundation for resolving contradictions in TRIZ resemble the principles of category theory, namely moving from the concrete to the abstract, followed by an operation at the abstract level before moving back to the concrete (as illustrated in Figure 3). This method can also be applied as a generic learning method for effecting change in XLP.

Figure 3. Effecting change through moving between concrete and abstract

intervention Thirdly, an XLP stands for an experience-driven large-group intervention in which participants need to engage with others as they tackle a set of open-ended tasks. The mission, not the contents, is driving, and product development is an integral part of the experience. XLP can therefore be regarded as an exponent of hacker education, which finds its roots in the wider maker movement in China and elsewhere, sharing its thoroughly collaborative spirit. Although the number of teams may range from four to forty (and more), typically ten self-organizing teams of eight need to muster all the skills and tap all the talents of the group in order to rise to meet the demanding challenges they have been set. These tasks are defined by the ‘challenge designers’ and are tailored for each intervention, which can span anything from an intensive four day immersion event to a continuous four year process (and why not an entire lifetime).

II. WHAT IT INVOLVES roles and tasks There are two main roles for participating in XLP: the role of challenge designer and that of challenge taker (also referred to as ‘missionary’). The work for the challenge takers coincides with the intervention proper as they form their teams, carry out the tasks they have been set, and report on their accomplishments. The challenge designers, on the other hand, take up responsibilities before, during, and after the event. There is no implied hierarchy between challenge takers and challenge designers. For instance, challenge takers can include executive MBA students who take up challenges that were set to them by high school students. Regularly the challenge designer team also includes temporary ‘hackers-in-residence’ who enrich the team with their expertise, be it artistic, cultural, physical, spiritual, technical, or other.

Before the intervention the challenge designers set up the workflow (that is, a set of behavioral protocols and sequenced activities), often using GIT to expedite data sharing. In effect, they design a micro-society that is built around the social forces of architecture, law, the market, and norms. In order to support the learning process, new software technologies are introduced and integrated in the XLP architecture. The challenge designers develop the tasks (or ‘challenges’) for the teams against the backdrop of a coherent story line which draws on real situations, addresses genuine needs, and is told with an authentic voice. During the intervention the challenge designers constantly collect feedback in the form of operational data. They may gather field notes through participant observation, for instance while assuming the role of technical consultant for the challenge takers. To the extent that the feedback is analyzed and acted upon in real time, the challenge designers can modify aspects of the workflow during the intervention. Otherwise, the challenge designers refine the workflow afterwards in view of future interventions. Throughout the process the challenge designers communicate with each other through a web-based project management and collaboration tool (e.g. Teambition).

the 4 stages The four stages alternately bring the challenge takers inside and outside their comfort zone. In the case of an 80 hours’ orientation event each stage would correspond to one day of the experience. In the first stage participants start off in a winning mood and experience early success to boost motivation and increase self-confidence. In stage 2, nicknamed “fail early, fail safe”, expectations are set so high that team members experience some frustration and realize that they will need to step up their learning and break across boundaries. Stage 3 emphasizes convergence as each team discovers that they completed just part of the tasks and that it takes collaboration across teams to finish the job adequately. Stage 4 is about demonstration: everybody presents their stories, deliverables and outcomes based on the XLP experience and extending into a personal enterprise plan.

the 4 forces The XLP workflow is defined as a set of formal protocols or ecorithms along the axes of four forces that enable, encourage, and constrain participants’ behavior during the event. These forces are architecture, law, the market, and norms. The force of architecture refers to the pre-given elements of the environment in the widest possible sense, including the spatial-material environment, the technological environment (including natural and artificial ‘languages’), and the organizational environment (e.g. interaction mechanisms). The force of law regulates how participants may deal with any infringements they observe. In XLP this force takes the shape of a court with procedures for arbitration and sanctioning violation of the rules of the game. A currency is introduced to make the force of the market more tangible, with specifications of the practices of buying and selling goods and services in the marketplace. Finally, the force of norms refers to explicit and implicit ways of instilling of what is good and bad in a particular XLP through the media or other communications. These four forces do not operate in separation; rather, it is the dynamic interaction between the forces that can be said to define the given culture of a particular XLP.

Figure 3. Logic chain for the market force in XLP

Figure 3 represents the ecorithm for the market force in XLP as a logic chain. The ‘activity’ of trading on the market is made possible because of various market-related ‘inputs’: the provision and availability of currency, the definition and installment of trading mechanisms, the distributed knowledge and skills required for trading, etcetera. The activity of trading results in certain ‘outputs’, above all the digital recording and collection of all transactions in Git. The anticipated ‘outcomes’ include that teams can demonstrate they have created added value as a result of their trading. On the part of the challenge designers, the optimization of market protocols would be considered a desired outcome of the event. In addition, the figure also illustrates how the market force is interdependent on the three other forces in both inputs and outcomes. In this way it would be possible to develop a coherent and formal logic chain vocabulary across all four forces, resulting in a further refinement of the XLP workflow.

For more information, please contact benkoo@tsinghua.edu.cn or jan.vanmaele@kuleuven.be