From genes to genome of collective intelligence

Individual genes can be combined in various ways to create genomes of collective intelligence systems. Mapping of the genome makes it easier to see the underlying structure and to think of new ways of arranging the genes to form new solutions.

Combining genes to form new genomes requires careful assessment, as the usability of different genes depends strongly on the situation. For example the crowd gene is suitable when useful resources are distributed widely or their location is not known in advance. In addition it must be possible to divide the activities in smaller pieces satisfactorily. Often a crowd is used for creation and intermediate decisions while the final decision is left for a specialized group. (Malone & al. 2009)

The choice of motivational factors in a collective intelligence system can be clarified by two rules of thumb. Appealing to love or glory can help to reduce costs, while using money and glory can make the crowd work faster. Motivation is a difficult

issue but still an extremely important one. Getting the motivational factors wrong guarantees the failure of the whole system. (Malone & al. 2009)

Collection gene can be applied when the conditions for crowd gene are met and activities can be done mostly independently. Competition, the subtype of Collection gene is suitable when only a few best solutions are needed. It should be noted that for competition to work the incentives must be strong enough to ensure participation without guaranteed rewards. Collaboration gene is usable when a satisfactory way to divide the task into independent pieces does not exist and it is possible to manage the dependencies between the individual contributions. (Malone & al. 2009)

Group and individual decision genes both require that the conditions for the crowds are met. Group decision should be used when the whole group has to be bound by the decision. For example everyone in a product development team should agree on product specifications. When an agreement is not necessary the decisions can be individual. (Malone & al. 2009) A more detailed presentation of required conditions for each gene can be found in appendix 1 and an example of a complete genome in Table 3.

Table 3. Genome of development process of Linux operating system (Malone & al. 2009)

Example What Who Why How

Linux Create New

software modules

Crowd Money Love

Glory

Collaboration

Decide Which modules warrant inclusion in next release

Torwalds and

lieutenants

Love Glory

Hierarchy

6 BUILDING THE CONSTRUCT

Producing innovations comes down to creating knowledge and transforming it to value trough two overlapping processes. Knowledge is created in knowledge creation process as described in rye bread model and innovation process is then used to transform the newly created knowledge to value. Interactions between the processes are complex and dynamic with multiple feedback loops. Simplified relationships between the processes in the context of idea development are presented in Table 4.

Table 4. Simplified relationships between knowledge creation process and innovation process in the context of idea development

Phase of knowledge

Visualization Idea generation

Socialization Idea generation

Externalization Search Idea collection

Combination Search Idea evaluation

Internalization Select/Implement/Learn Idea selection &

implementation

Potentialization Learn Formation of basis for new

ideas

The reason for this simplified approach is to provide an accessible model for the purposes of the study. It is assumed that ideas are mostly generated in visualization and socialization phases by embodying from the abstract to mental models and by forming new combinations of shared tacit knowledge. Defining a clear starting point of innovation process is difficult because of the fuzziness of the front end. For the sake of simplicity here the innovation process is considered to begin when ideas are made explicit in externalization phase of knowledge creation process; documenting

ideas generated in previous phases transforms knowledge from tacit to explicit form making the communication easier. Evaluation takes place in combination phase;

knowledge about the quality and feasibility of ideas is combined to ideas in explicit form. In internalization phase the ideas are implemented. First a selection is made about which ideas are developed further and then they are put into practice. Large part of learning involved in innovation process also takes place in internalization phase, when developed ideas are tested in the real world. Learning continues in potentialization phase, where the experiences gained in internalization phase are transformed to self-transcending knowledge, forming the basis for new ideas.

Focus of this study is on the search phase, the interface between knowledge creation process and beginning of innovation process. A smooth transition over this interface requires effective documentation and evaluation of ideas generated during the knowledge creation process. In STI mode of innovation the transition over the interface from research to development is relatively straightforward and simple.

Increasing emphasis on DUI mode, open innovation paradigm and shift towards the fifth generation innovation process complicate the matters significantly. Large amount of information from multiple sources and weaker signal-to-noise ratio increase the strain on idea processing mechanisms. Both knowledge creation and innovation processes are becoming networked activities making the traditional hierarchical management difficult. Collaborative Innovation Network (COIN), being an integrated networking model and as such an example of the fifth generation innovation processes, appears to be a more promising approach to managing networks. COINs aim at flexibility, robustness and self-organization. Processes involved in innovation networks should be compatible with these demands. Capable of satisfying the requirements of COINs, collective intelligence offers a promising basis for idea evaluation tool development.

Collective intelligence has been successfully utilized in business context to gain outreach, additive aggregation and self-organization. In order to facilitate collective

intelligence and to avoid common pitfalls of decision making the system should ensure diversity, independence and decentralization in decision making and motivate participation. Other desirable features are modularity and self-organizing properties.

An evaluation method relying on these features could prove to be useful in crossing the interface between knowledge creation and innovation processes in the changed innovation environment.