Antifragile: Build Open Resilient Systems

Nicholas Taleb Nassim’s latest book, Antifragile: Things that Gain from Disorder (2012) is a nice continuation and development of his oeuvre. The main point in his first mainstream book, Fooled by Randomness (2001), was that humans are not good at thinking statistically, and therefore to improve our lives and ability to act in the world, we need stories or heuristics that package accurate underlying statistical information. Black Swan (2007) made us aware that black swans (seemingly rare events (if you have never seen a black swan, you incorrectly think that they do not exist)) can happen much more frequently than we can estimate. Therefore, we should organize our lives to minimize exposure to negative black swans (events with unlimited downside risk like stock market crashes) and maximize our exposure to positive black swans (events with unlimited upside like investing in startups (with a small portion of total assets)). In the future, there could be a Black Swan App to help us respond to life’s events in real-time with bias correction, heightened rationality, and statistical accuracy.

In Antifragile, Taleb continues to articulate his unique world view and winds it into a proposal for how to better lead our lives as individuals and societies. Fragility is organic systems that aim for stability and avoid change, thereby becoming brittle, weak, and breakable as a result. Antifragility (like Derrida’s autoimmunity) on the other hand, describes systems that are open to mistakes and quickly learn from and incorporate errors, thus becoming resilient and vibrant with the ability to adapt and survive (like Silicon Valley’s mantra to ‘fail early and learn fast’). For better vigor and survival, organic systems (like living organisms, humans, and societies) should develop their antifragility.

Another way of understanding antifragility is that when you have a capability, it means that you are able to handle new situations that arise in the same domain, effectively handling situations that arise that are up to 10% outside the bounds of situations you have seen before in that domain. For Taleb, success is determined more by tinkering and harnessing the disorder and chaos in a system (the variance or antifragility) than applying pure intellect. This is how the industrial revolution happened, and how technologies develop that drive science. Antifragile systems are those that gain from randomness or uncertainty (statistically, pulling a probability distribution’s mean higher with more upside long-tail instances).

Fragility/antifragility applies only in the case of organic systems, not inorganic systems (like our computers (at present)). Organic systems need stressors to grow, thrive, and survive. Taleb’s colorful example distinguishes between a cat and a washing machine.

Systems-level Thinking Helps to Address Protein Folding

Deciphering protein folding is critical to a fundamental understanding of biology as proteins conduct most cellular operations, and since misfolded proteins are often causally implicated in disease.

The status of protein folding (describing proteins as folded into their final 3D shape) is that as of January 2014, the main resource, the Protein Data Bank, has 96,000 listed known protein structures. There has been much technological advance in determining the static and dynamic features of protein structures, including in X-ray crystallography, NMR (nuclear magnetic resonance spectroscopy), cryo-electron microscopy, small-angle X-ray scattering, and other spectroscopic techniques.

This sounds like good progress, however, taking the human example, only 24,000 of the Protein Data Bank's 96,000 listed proteins are human, and this is of an estimated total of 2 million human proteins. Further, all of the protein conformations have been determined empirically (e.g.; manually) rather than with prediction (e.g.; digitally). It was proposed that given the amino acid string, it should be possible to predict the 3D conformation of the protein as finally folded, but this has proved elusive. Both scientific and crowdsourced efforts are looking at the problem. Crowdsourced projects include games like fold.it, community projects like Protein Folding@Home, and prediction contests like CASP (Critical Assessment of Protein Structure Prediction), a community-wide, worldwide experiment for protein structure prediction, the next one taking place May – August 2014. All of these projects focus on the unitary folding of one target, such as TM019, as opposed to more universal system dynamics.

Complexity scientist Sandra D. Mitchell presented work at Stanford on January 15, 2014 suggesting that we need a plurality of conceptual representations and models (any model is only partial in some way), and that any complex problem should be addressed with an integrated multiplicity of approaches. Many (and possibly most) complex biological issues such as cancer and aging are now understood as deeply dynamic and systemic phenomena. Similarly, proteins do not fold in isolation, the local environment is highly involved with protein chaperones and other signaling processes (one example is the intricate behavior of toxic amyloid HSPs (heat shock proteins)).