For explaining the concept of complexity, professor Ralph Stacey developed a very convenient tool, the Agreement Certainty Matrix. It situated complexity between the concept of order and the concept of chaos. The order assumes high agreement and certainty about what is going on. Chaos comprises situations where uncertainty rules with agreement and certainty both low. Stacey presents complexity as a hybrid concept, which capacitates it for bridging between order and chaos, modern and postmodern understanding of the world, as well as between facts and values.

The mainstream schools define complexity in incompatible frames (Castelani). The first understands it in a disorganised state, located ‘on the edge of chaos’. It addresses the lowest level of description of a larger system with random behavior of individual parts. Global structures emerge unpredictably exclusively by means of the local interactions of its individual components. Disorganised complexity is typically studied with formal language and quantitative means such as in theories of informational entropy, or with theories of computational complexity.

The opposite is the school of organised complexity that locates its object of concern ‘on the edge of order’. In this case, complexity emerges from a moderate number of macroscopic variables related to one another in interdependent ways, such as between billiard balls or between domains of sustainable development: economic, social, and environmental. The concept of organised complexity found its place in the philosophy of science and in qualitative methods of social research, among others.

Each school develops as exclusive and often neglects or ignores the other. Polarisation contradicts the very meaning of complexity as a bridging principle (Malaina). Complexity cannot be properly understood with binary means but with mesoscopic ones. This means as partly ordered and partly disordered, applying partly systemic and partly antisystemic logic. In turn, systems and complexity do not belong to the same level of generality. The theory of complex systems contributes very relevant and indeed indispensable insights into many aspects of social complexity but only as a special case. Complex systems cannot explain what is genuinely complex: fundamental coalescence between the rational and the irrational.

If so, evaluation of complex matters needs to be theorised as a mesoscopic undertaking, integrating between systemic and chaotic reasoning at the middle level. On the one hand, evaluation is undoubtedly a rationally systemic effort with a distinctive logical frame, methodology, and standard procedures. On the other, it is also value-based, ‘irrational’, and so chaotic when it seeks to include incommensurable stakeholders in participatory evaluation in search of coherent understanding from incompatible grounds. As well as the opposite, mesoscopic methodology is integrative precisely because it is evaluative: it is based on scientific evidence without being dominated by scientific (systemic) way of reasoning (Radej). Evaluative and mesoscopic reasoning are identical.

So, what can research in complexity advise the policy impact evaluator about his or her practices? It seems it advises integration of systemic and participative approaches to evaluation by applying a mesoscopic methodology that relies on both:

–      Systemic step: The evaluated matter is first defined as complex at the meso level, which is at the intersection between foundational contradictions, vertical and horizontal that emerged a given complex matter.

–      Participatory step: Stakeholders then convene to identify horizontally three main domains of a given complex matter on the meso level, such as physical, economic, and social domains of territorial cohesion (EU); or education, research, and innovation in the triangle of knowledge (European Institute of Innovation and Technology), etc. This is a constructivist part of the evaluation that invites creative design thinking, for instance with the three-part Venn diagram.

–      Systemic step: Evaluator continues by rigorous analysis of the evaluated matter’s impacts on three domains and their cross-sectional, secondary links in the meso-matrix – a tool that locates synergies between domains in their correlations (Radej).

–      Participatory step: Evaluation concludes with stakeholders discussing synthesised findings and evaluatively interpreting them in search of shared understanding.

It seems that the evaluation of complex matters does not demand very sophisticated interventions in standard evaluation procedures. Nothing really new is required except applying the consistent approach to connecting several compatible features of conceptually opposite approaches to evaluation at the meso level where order and chaos can naturally overlap. Complexity is after all only a specific ordering principle. It produces simplicity, order from disorder, only in characteristically uncharacteristic, and unpredictable but nevertheless surprisingly reasonable way.

(not language edited)

Bojan Radej, Slovenian Evaluation Society, September 2020


Castellani B., F. W. Hafferty. 2009. Sociology and Complexity Science. Berlin, Springer.

Malaina A. Two complexities: The need to link complex thinking and complex adaptive systems science. Litchfield Park, Emergence: Complexity and Organization, 17/1(March 2015).

Radej B., M. Golobič. 2021. Complex Society In the Middle of a Middle World. Vernon Press, forthcoming. Presentation:

Stacey R.D. 2011. Strategic management and organisational dynamics. The challenge of complexity to ways of thinking about organisations. Sixth Edition. Harlow, Pearson.