You are here

Research logbook

This page contains a yearly review on my research. I give a general view on the activities during that academically year (starting October). The page is a way for me to reflect on what has happening over larger time period.
2014-2015-- GBI fellow: PhD defense & consulting
2013-2014-- GBI fellow: Wrapping up the PhD
2012-2013-- GBI fellow: Burnout and delays
2011-2012-- MOSI assistant: First major Collective Intelligence test
2010-2011-- MOSI assistant: The unexpected outcome of Drupal
2009-2010-- MOSI assistant: Failing ÆIP in existing organizations and reorientation by WSDB
2008-2009-- MOSI assistant: The concept of Agile-Enterprise Innovation Planning (ÆIP)
2007-2008-- MOSI assistant: Novelty theory & Web service development for business (WSDB)
2006-2007-- MOSI assistant: Teaching & exploring management (IT & innovation)
2005-2006-- Project developer KnoSoS: knowledge sharing over social software (for MOSI)
2004-2005-- Project developer CRAB: Cartography of Research Actors Brussels (for DISC)
2003-2004-- Independent Research: ECCO (Evolution Complexity and COgnition) and DISC
2002-2003-- Student: Creative agent, Master thesis
2000-2002-- Student: Experiments during Master studies


2012-ongoing-- Global Brain Institute
2006-2012-- Drupal community
2005-2012-- MOSI department
2004-ongoing-- ECCO research group
2003-ongoing-- Novelty model


Introduction

In 1998 during my summer/student job at Starlab nv, the blue sky research facility, I've encountered an atmosphere where being creative with science and technology was the key to create innovation. I was only helping out to organize a conference, but experienced first had what it means to working in high-tech and creativity stimulating environment . It became clear that Starlab was a state-of-the-art experiment way ahead of its time and it would become a major influence for my research focus.

The reason to take an academic studies in informatics was related to that summer job, but even more to the program language scheme. Scheme got me interested in dynamics of information processing and I found it interesting to understand how a program is related to Artificial Intelligence. The courses on AI however was in strong contrast to my experience at Starlab. With a focus on expert systems and simplified simulations that in general was all about statistical mining. My interest is however more to architectures and how they create coordination. Simulations are good tools to investigate an some of the coordination. I consider them stepping stones to then use the insight to actually support people and make them coordinate more fluently around innovation by self-organizing distributed means. My research would start with agent simulations to research how intelligence and creatively/innovation relate to coordination and feedback. It became the research on novelty.

Master in Computer Science

To make the ideas explicit simulations where created. I was able to code 8h to 15h a day during 2000-2003, a luxury if you think of it. In the summer of 2000 I started my first project and would work on it for more then a year. It was an agent simulation representing animals in a habitat. The simulation was inspired on games where you had a map and a mini-map to scroll a 2D-world. The agent had some variables that would change by every step they made and when the values hits a specific value, a concept would get triggered like hungry thirst, etc. This concept would get evaluated by a logic program that represented the animal cognitive structure. The experiment was a first attempt to understand the relation between embodiment and rational reasoning. Reflecting on the results, the question arise how the logic got their in it in the first place. By expanding the project with visual recognition and self-organizing of lexicon more questions than answers where created.
In my second project in the Fall of 2001 I became creative with the knowledge of expert systems. The new agent's cognition contained a a mixture of reinforcement learning and bayesian logics. The reinforcement learning was used to track back the reward in a specific context. The bayesian logic analyze the context of the reward, thus creating several context to do reinforcement learning in. The interesting part of the solution is to see how an interplay of the two quite statical methods can resolve a highly dynamic interaction. Several problems exist with the method. First it seems impossible to explain the origin of the method in an evolutionarily plausible way. The second problem is that when we look at the history of innovations we notice agents create their environment. Thus our created solution is only good enough to find the rules. The rules would not change the environment. I wanted to investigate creative agents. Lucky I found a mentor in the same line of thoughts, Walter Van de Velde. During the summer of 2002 we discussed an search for agents who learn by doing and do by learning. The idea of craftsmanship was the prior inspiration. For example, how does a painter improve herself by doing?

Creative Agents

After several investigations related to creative/innovative agent a challenge was created. The goal was to get a clear vision by making incremental iterative simulations that where used to adapt our view. In February 2003 a specific experiment arose. The agent would start with a knowledge base to "survive" in its environment, but its actions would change the surrounding, forcing his knowledge-system to adapt along the way. The simulation was inspired by a documentary of apes using primitive tools. In my simulation the principle was to figure out how an agent could learn from "eating the fruit of nuts" to "breaking the nut with a stone" to get the fruit out of it. In the setup the agent had knowledge of stones and knew that stones break other stones. The whole solution seams amazingly rational if you look at the process afterwards. First the agent picks up fruit from open shells and it would make the relation with shelter and fruit. By getting some fruits out of half-open shells it would notice the spontaneous breaking of the shell. By eating fruit, open shells would become rear. So the agent has to get fruits out of the shell. From this moment on the agent could find nuts that are to hard to break. Giving him a problem of something that did work earlier on. To solve this the agent search is knowledge system and tries to recombine it to find a solution that may work. So the agent thinks of using a stone to break the nut. After trying it out, the positive feedback would add it as new option.
By the incremental simulations each of the rational steps just explained would become more concrete. Simultaneous the design would change to make the actions possible. The chalange was to shift away from hard coded "knowledge" to a system that learns knowledge. The method used allowed us to get over our own restrictions. For example, in AI you learn the trouble of ambiguity in visual recognition: how do you recognize the object without failures? It turns out we actually need the ambiguity and that such failures are natural and the source for new opportunities. For example, failing to separate the nut from the fruit allowed learning. In the simulation a value, representing the hardness of the nut, would stimulate the ambiguity. When the harness was very low no distinct recognition was made between nut or fruit. In case of low harness the nut could break spontaneous and become the fruit. The agent could observe the phenomena of breaking and use it later on to do it intentionally. A similar ambiguity relates breaking and stones This, than, would lead to an agents regulating its actions to create from understanding out of ambiguity.

Novelty model

By putting so much time to develop the simulations with my Master thesis, it did not put time on other parts, like understanding the existing literature in greater depth. This is where my interdisciplinary research started. I'm connecting the dots in embodied-extended cognition, with studies on the emerging of science, with literature on innovation management and with knowledge on software development. The creative agent simulation was an embodied-extended cognition experiment, many of the literature gave me a better understanding how it fits with current research on cognition. However, the focus on emerging of knowledge is much more related to research on science studies. Interestingly the two domains have each a publication of a design of a control-system that is similar to my creative agent, thus came the generalization to "novelty". While the two domains help to understanding the problem it is innovation management that really requires the insights. The research renews the interest in IT-management to see how software development can help people to conduct innovation in a more structured way.

Evolution Complexity and Cognition

My first contact wit Francis Heylighen was with the course given by my first mentor Walter Van de Velde in 2002. As of that time I would go to seminars organized by Heylighen. In 2004 Heylighen became my mentor to continue the experiments on creativity and innovation from a complex system and evolutionary cybernetic perspective. In 2004 Heylighen creates the Evolution Complexity and Cognition research group. This was mostly a virtual group of individuals working with Heylighen. Meeting over coffee and seminars where a moment when the group members discuss the research. This all changed in the summer of 2011 when a sponsor (who still likes to be anonymous) founded with Heylighen the Global Brain Institute (GBI). The official (but quiet) launch was January 2012. The dynamic changed a lot with weekly research meeting and a team getting more collaborative over time.


MOSI group

The workshop Tanguy organized at the VUB in 2005 was may first contact with the MOSI group. Later that year I would become project developer for the KnoSoS project and so become an member of the department. One year later, when Tanguy finished his PhD, I took over the position as teaching assistant and PhD student. I'll be in that position until September 2012.

Drupal community

We met Dries Buytaert end 2005 to see if Drupal would be a good tool for our KnoSoS project. While coding in Drupal most of 2006 I only became active in the community by co-organizing Drupalcon 2006. However, my activities have always been in the marge of the community.


2003-2004

End 2003 and beginning of 2004 started disappointing, my funding was rejected and the research lab Walter Van de Velde had hoped to create never happened (see DISC open letter). I understood I was standing alone (again) and knew my research needed to become embedded in the research landscape. At that point my research is about intelligence, cognition and awareness. I made contacts with Erik Myin at our university and we where creating a paper "The primacy of context" for the EELC congress in May. At the EELC Francis Heylighen contacted me with the suggestion to become member of ECCO. I knew Francis from the seminaries in 2002, but than ECCO didn't exist yet. The period with Erik and ASSC8 conference in June started a new aria in my activities: the need to get embedded and connected. I went to several FET & NEST information sessions to get contacts. My paper for EELC got rejected because of ill writing and to vague. I still didn't had any publication but my activities and the contacts I made convinced me of going on. Walter Van de Velde had suggested me to go to the convivio summer school in September, so I did. Convivio was another interesting view on IT and design and it fitted nicely in what Walter had learned me.

2004-2005

In October Walter Van de Velde contacted me with the option to work on the CRAB project. I would start at that project in December, at that time Walter was changing to a new function as one of the EU FET coordinators. The most part of my activities in 2005 was about developing CRAB and my interaction with ECCO. My relation with ECCO would become more imported, with my ECCO-colleges I kept on discussing our research, Francis became my advisor and helped me with writing a working paper and suggested some readings to get me more embedded. With Marko I was having some interesting collaboration and some outlines for papers, but he left in April and our plans got canceled. My own working paper transformed to a paper for EMCSR, thanks to the support of Francis.

In December I made my first contact with Tanguy Coenen on a workshop of Knowledge management and IT. So 2004 started bad, but it ended good. The developing of CRAB would also bring me into contact with several other interesting people and organizations like the cross-talk sessions and Maya Van Leemput who works on future studies. Still inside DISC where I was developing CRAB I was very isolated. In the mean while Tanguy and Dirk Kennis where preparing a IWT-TETRA on "Knowledge sharing" and in the summer they got the good news the project was accepted. They asked me to work on the KnoSoS project.
Next to cognitive research and the KnoSoS projects, I started to learn about entrepreneurs. I got back into contact with Walter De Brouwer, after a terrible accident with he's son we talk a lot in the Hospital. Walter is the nephew of my mother, but having a small family I call him my nephew as well. He was the head of Starlab and the reason I get the summer job (see introduction). Walter De Brouwer defended his PhD in September. We discussed topics several times that year and he gave me some interesting books to read.

2005-2006

I started working for KnoSoS in December. The atmosphere in the team was as good as it can get. I've became a Drupal addict and programmed a lot. Their where two new challenges, the fits is to code in team, the other is to push the code back to the open source community. My biggest talent in coding, creating new generic designs, became challenged with other needs in development. While being pleased with my skills for the generic design the problem of user demand did result in dirty code. Something I had trouble accepting. When Dries asked me tp co-organizing the drupal conference I did not have to think twice. It became an interesting coctail of three smaller events: govcamp, drupalcon and barcamp. ECCO was a bit on a low track during that year and so was my research on cognition. Everything was focused on KnoSoS and Drupal.

2006-2007

I've stopped working on the KnoSoS project as Tanguy finished he's PhD, their was the need for an assistant at MOSI. Of course I wanted to do this as I wanted to research & teaching. With KnoSoS entering a non-development face the timing was right. I became the assistant for programming and databases. Most of my time would go to creating exercises for the courses. Particualr for the programming course, which became a web development course. In the mean time I still was coding a bit jquery. My advisors in MOSI (E. Vandijck & E. Torfs) learned me a lot about IT governance. Next to this I was exploring literature on innovation management and the history of technology. During the summer I gave a presentation relating higher-level cognition to radical innovation. The research on using cognitive research for innovation management was just starting.

2007-2008

Considering that the teaching would stay a significant part of my job I made the best of integrating it with my research, I was developing my own course "Web service development for business" (WSDB) for next year, but the base was created during this and previous year. For my research I was still wrapping up the articulation of the earlier research and testing the new concepts in managerial conferences. In April I got the idea to use the novelty model to design an "Agile-Enterprise Innovation Planning" (ÆIP).

2008-2009

Teaching would still take up most of my time. The WSDB was now given to first year master student, which was a different audience than the master-after-master students I had previous two years. I had to add several new exercises. The new target group was also less familiar with the complexity of projects and needed more support. My research was now focused on working out the ÆIP. A first paper got published for the ISPIM congress (International Society for Professional Innovation Management). Other event to find feedback an possible testing environment was IBM we had 4 workshops and 2 meeting about ÆIP specifically. I also gave three more seminars for MOSI/ECCO during the first semester. An participated in the summer school by EIASM (European Institute for Advanced Studies in Management)

2009-2010

With my advisors in MOSI a plan was made to contact companies and work out experiments on ÆIP:
big companies like IBM, Toyota, medium companies like Recticel and small companies like CREAX and Namahn and bigger research groups at our university like TONA. While most showed interest, only with Recticel an actually planning was intended. The economic crisis and heavy restructuring in Recticel disrupted our plans. While we failed to creating an experiment for ÆIP in existing organizations, the time spend in teaching turned out to create a backup plan. Every year the corrections to the WSDB course had lead to better projects. In 2008-2009 some students sold their project, but that were the mediocre projects. The best projects can be seen as a prototype and one of the students in 2010 actually made a business plan and searched for funding. This made me recognize WSDB as soil for micro-spinoffs. As a consequence we decided to stop trying to have the ÆIP integrated in an existing organization and focused more on understanding how to make ÆIP emerge with WSDB.

2010-2011

While I was promised to get more time in my last two years for research, instead I had to give a new course that I had to learn myself first (Software for Management). This consumed so much of my time it was very frustrating. With the reorientation to WSDB I decided to interview the founders of Drupal companies during the 2010 conference in Copenhagen. The outcome was impressive, it made me construct the concept of self-innovation. The effect was a boost in my involvement with Drupal, going to Chicago and organizing a workshop in London. I created papers of my insight on Drupal for R&D conference and ISPIM.

2011-2012

The project developments with the course Software for Management was unmanageable with 3 teachers last year, this year I would be the only teacher. During the summer I and the professor would design a solution, solving the teaching problem with my research Intelligence Amplification by Collective Intelligence for the ÆIP architecture. In total about 600 students were enrolled. During the first semester 300 students participated (see blog). Next to the huge workload of teaching I found time to create a methodology paper that would get presented in June. I worked over 50h each weak for four months (650 hours in total). Add to this a a conflict with dean and the lack of trust by my advisers about finishing my PhD in time and you have an idea of my mental breakdown. During spring I was truly burned out only weeks before my wife had a difficult childbirth in March. All this left me weakened and burned out.

2012-2013

My weakened condition and a cold winter made me seriously sick (first flue, then pneumonia, later several flues again). The fight with my health creates delays on my PhD the feedback from my adviser also led to many changes, creating again a lot of delay. As a young father I would need to take more responsibilities at home. Just as with the fist child the second came as a surprise, luckily now the birth was without complications. Being a father would have a huge impact on the workload I'm able to handle. All this created delays on the PhD. Next to the PhD my attention was to support GBI team, but my contribution stayed marginal.

2013-2014

My focus was mostly on wrapping up the PhD. Consider that it took me about 10 years to create this interdisciplinary PhD should give some idea of the complexity we faced on wrapping it up. On weekly base I met with my adviser. My dyslexia did not help, but we solved that problem by taking on an editor. Week after week we would finish some pages. Also being a young father with two kids that keep you awake almost every night did create a surrealistic experience of time and of course diminished my abilities to function at normal capabilities. Next to warping up the PhD my attention went to validating the research and find appropriate post-doc opportunities. One path followed was the development of the education platform as a spinoff, which got presented on the 15th of May for Innoviris. I did had meeting with Technolgy Transfer and With IWT to investigate the spinoff opportunities, but they were still too immature for now.

2014-2015

By October 2014 we finally were ready with the PhD. We added an executive summery, which is really needed for a interdisciplinary PhD. Next we needed to approach possible jury members and by November the PhD got submitted. At the moment it is in the hands of the jury members and the defense is set for February 2015. From September onwards a new challenge at home would absorb much of my attention and transform the house we live in, into a construction aria. It had to happen, but it is not an easy thing with yong kids. During November and December I fully committed to the construction at home, creating the proper atmosphere to fully commit to my research in 2015. The plans of the spinoff made me release that my best option is to become a consultant on novelty regulation, which is my main challange for 2015.