Alicia Juarrero - Complexity is not complicatedness

Deliberate Complexity Conferences 2022 Recap

This talk was presented at Deliberate Complexity Conferences - Building Successful Platforms and APIs on 19 July. We are honoured to present the video recording and talk summary below. Enjoy!

Visit our talk recaps ToC page for an overview of all presentations!

Alicia Juarrero, PhD

Department of Philosophy, University of Miami (FL)
President and Co-Founder, VectorAnalytica, inc.

Alicia's presentation (video recording)

Alicia's slides


(from 40:00 in the recording)

  • Are interfaces placed on the border of systems, or can we say they define the border of systems?

  • While we can see interfaces, it is much harder to see other constraints in software systems. How can we approach this dilemma?

  • Do you find the notation used by Montevil and Mossio applicable outside of chemistry?

"Complexity requires constraints that generate these complex, multi-level parts-to-whole interdependencies. Once they entrain, once they synchronize, the individual processes change the likelihood of their occurence conditional upon the interdependencies in which they are embedded. At that point the complex wholes become the governing constitutive constraints that maintain that system, so the hypercycle becomes the governing constraint that modulates/regulates and otherwise controls the goings-on in the individual processes."

Complicated or Complex?

Complicated systems:

  • parts have no mutual dependencies
  • totality is additive of the parts and is in principle possible to track and predict (even if difficult), output is smoothly linear to input.
  • no phase transition of parts into a system with qualitatively novel propertiers
  • for example a sand dune

Complex systems:

  • nonlinear systems where the output is discontinuous with the input
  • interdependent constrained interactions among the components
  • elements synchronize and/or the whole begins to oscillate
  • a complex system emerges in discontinuous phase transitions
  • properties of the coherent synchronized whole are qualitatively different from the elements alone
  • for example an ecosystem, organizational culture

Efficient causes (from Aristotle, as energetic transfers) alone do not generate complexity. Constraints: are conditions or factors that raise or lower barriers to energy, matter, and information flow – without themselves directly transferring energy. For example: timing of a kick on a playground swing.

Context-dependent constraints enable complexity

Catalysts and feedback: inherently context-dependent, conditional on one another.

"A coherent interdependence is generated thanks to enabling constraints."

Enabling constraints: weave together relations among parts or parts and their context, so coherent totalities–that have characteristics that are greater than the sum of their parts–emerge as a result. The new coherent dynamics generate novel information.

Complex systems are embedded in their context.


About interlevel relations: parts-to-whole and whole-to-parts.
There can be relations of constraint among components of the same level of organization.

But also, the interdependencies with the qualitatively different component–that is different than the components themselves–do exercise control over their components.
Interlevel relations are also constraints, but usually not energetic transfers.

For example: autocatalysis is a non-linear self-reinforcing process. The metastability of the system is preserved, even if through a different pathway. And the same input can produce qualitatively different outputs. (Unlike mechanical processes.)

"The emergent properties of the whole can be carried out through different component pathways. [...] The overarching system's dynamic equilibrium–given historical and current context–is maintained. The emergent properties of the ecosystem become the variables that matter."

Closure of constraints (Montevil & Mossino, 2015)

A higher level set of constraints that become metastable, in contrast to the components that make it up but which can vary. The emergent properties have relational interdependencies.

The interdependent relational dynamics (the hypercycle, Nszostak 2015) function as governing/constitutive constraints that exercise control top-down from the wholes to the parts.

"[Chess grandmasters] view the relationships in terms of a system-wide parameter, a functionality, call that strategy. [...] Strategy is flexible, it is a view of a constraint-structure, not of the individual pieces. [...] Opportunities and threats are emergent properties of the totality."


  • Interfaces are enabling constraints that generate and maintain this system dynamic, negotiate the inter-level relations.
  • Filter, recode, reformat, re-select, such as to carry out the functionality of the whole.
  • Interface mechanisms determine the selection criteria in view of the satisfaction of function.

"APIs are interfaces that act as enabling constraints that facilitate the transduction of information into the internal dynamics of the system."

Managing complex system

  • Cannot directly create or control a complex system.
  • Have to perceive the ongoing complex system in terms of constrained relations.
  • Can change/manage by adjusting enabling and governing constraints, relations, interdependencies, and context. By doing so, we can expand the adaptive space of the system.

Sign up to the Deliberate Complexity Newsletter so that you never miss out on the latest recaps and the API Resilience podcasts.