Wednesday, 11 March 2015

A Behavioral Interpretation of Antifragility

I have been recently invited to participate to a Webinar on Antifragility: "Antifragility Webinars: Practice Beyond the Rhetoric!"

Quoting from the referenced site, on March 18, 2015 Dr. Russ Miles and Dr. Si Alhir
"will host a panel of practitioners to explore:
  • How these practitioners have interpreted Taleb’s concept of Antifragility,
  • How these practitioners have translated their interpretation into practice, and
  • The results and impacts of their efforts — Practice Beyond the Rhetoric!"

Here I would like to share with you my answers to the first question. If time allows I will address the other two questions the coming days.

So, how have I interpreted Professor Taleb's antifragility?

Mine is a behavioral interpretation, meaning the focus is not on the way a system is structured and constructed. Rather, it is on the way the system responds to change. [cf. the work of Wiener and others that brought to the concept of cybernetics]
This approach focuses on systems and their output, regardless of the nature of those systems. Therefore it applies to biological systems ("beings"); artificial systems (cyber-physical "things"...); and it also applies to collective systems made of beings and things.
Wiener and others used behavior to characterize all types of systems — to tell how smart a system was when facing change. There's a famous paper, called "Behavior, Purpose and Teleology", where they distinguish systems according to their behaviors; they have
  • systems that react with no concern about the situation (elastic systems)
  • systems that check what's going on and try to adjust to it (adaptive systems)
  • and systems that keep track of what's going on and try to "tell the future" (anticipate conditions that could be black swans or maybe gold swans) (predictive / extrapolatory systems).
Obviously if you consider resilience, the above classification is somewhat in line with Prof. Taleb's vision of fragile, robust, and antifragile systems. I say "somewhat" because Wiener & co. did not take into account the effect over time of facing change: the genetic feedback produced by the experience. In other words, a systems' evolvability.

If we want to extend the behavioral classification with Prof. Taleb's antifragile systems, we have to consider an extra dimension. I call this dimension the one of evolving feedback behaviors (EFB) — behaviors, that is, that leave a trace in the system, and actually modify the system. It is important to understand that such systems do not preserve the "self" — their identity. It is more difficult to make sure that such systems "stay the same" — namely, comply to their specifications; behave as expected; and so forth. Is this a big problem? Yes it is. Quoting Professor Hawking, an artificial system that can evolve

"would take off on its own, and re-design itself at an ever increasing rate. [..] Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded." (cf. my post "What System Is The Most Resilient?")

Antifragile behaviors may be considered as a particular type of EFB: one in which the self-modification improves the system-environment fit — one that makes it more probable for the system to survive in the current (or the hypothesized future) environment. (Note that being able to improve one's system-environment fit has nothing to do with guaranteeing that what the system does is "right". In other words, special care must be taken to make sure that the drifting of system identity associated with antifragile behaviors does not translate into "dangerous" or counterproductive behaviors. Some form of safety enforcing  invariants should probably be embedded into antifragile behaviored systems (cf. Asimov's Laws of Robotics)
Creative Commons License
A Behavioral Interpretation of Antifragility by Vincenzo De Florio is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Permissions beyond the scope of this license may be available at