Nassim N. Taleb book “Antifragile: Things That Gain from Disorder” is notoriously divisive: you either love it or hate it. In my case, it’s both: I loved reading it as much as I found it infuriating. I will start my discussion with a short description of the concept of antifragility and its obvious merits. I will then dive directly into some criticism: not because it proposes wrong or worthless ideas, but because it consistently fails to complete the journey it begins. The book feels as if it was written in haste, following an urge more than a plan, with little or no attention at how it could have been improved. On the other hand, at least one key concept connected to antifragility has implanted itself in my own mind, and it feels like it will not go away. With this series of posts I will try to expand Taleb’s ideas in the direction that suits the aims of this blog: at the very least, I’ll explore the tight link between antifragility and evolution and will look at Taleb’s use of cognitive heuristics.
But first, I’ll briefly look at the basic antifragility concept. Taleb sees everything under the sun as either fragile, robust or antifragile.
- Fragile things require a stable environment: a china teacup can survive for centuries, provided it never receives a strong enough shock. Shake it above a given threshold and it will break and either become less functional or cease to exist.
- Robust things are the same, but they can survive greater shocks (I’m not using the word “object” because the same concepts can be applied to ideas and non-material entities of any kind). Taleb stresses this point many times: robust is not the opposite of fragile, the opposite of fragile is a very different animal.
- Antifragile things thrive in an unstable environment. Up to a certain extent, they get better if they are exposed to random events, stressors or forces. At first sight, one would think that such things are very rare, but this isn’t the case: (this is my addition) the ubiquity of antifragile entities is the direct result of Natural Selection.
To some extent, every entity that has been shaped by Natural Selection is antifragile. Even the simpler organisms require a source of energy, and if you feed some energy into a system, entropy will rise. This means that at some fundamental levels, all living things, no matter how simple, require some amount of chaos. More: one can see that more complex creatures show increasing degrees of antifragility, to the extent that they become truly the opposite of fragile: give a human too much stability (for example, close her in a dark room) and all sorts of bad things will happen. The same applies to other entities that go beyond the directly biological: philosophical or cultural constructs, human societies, organisational systems and so forth. They can all be looked through Taleb’s lens, and be classified among the spectrum of fragility, from fragile to antifragile, passing through robustness. [For more introductory information, assuming you don’t want to read the book, Google will help you, but you may find some guidance also on my previous post on the subject.]
So far, so good. I have exactly zero issues with the idea of antifragile as expressed above. In fact, I love it, but please do note that I’ve already tweaked it a little: the emphasis on Natural Selection and living organisms is not my own doing, but the way I’ve used is. In fact, this little twist is the direct result of the main issue I have with the book: Taleb rarely (if ever) mentions Natural Selection, instead he talks of Mother Nature. This repeatedly got to my nerves: Natural Selection operates on information, allowing the sort of information that promotes its own replication to get copied over space and time; whatever promotes its own replication better will (tautologically) spread more. As a result, the environment changes, at least because more copies of successful replicators are present, and therefore replicators that are excessively fragile will be consistently selected out. To say it in another way, fragile replicators disappear almost immediately: Natural Selection can be seen as a generator of antifragile replicators.
On the other hand, any degree of antifragility will increase the success (fitness) of a particular replicator: their existence implies that the environment will keep changing, but because Natural Selection has no hindsight, the direction of this change is effectively random. The result is that all organisms are somewhat antifragile, and more importantly, all reproduction strategies of such organisms are antifragile. Hence, from the perspective of a biologist, it is possible to say that Natural Selection is the source of all antifragility. I’ve checked this claim: Taleb provides a long list of antifragile “things”, and all of them rely on replicable information. With no exception. Whenever replicable information is present, Natural Selection will be at play, it may not act through the familiar DNA-based replication mechanism, but it still regulates what information is replicated and what else is “forgotten”. The tautological property of Natural Selection guarantees that this is the case: what is selected for is what intrinsically favours its own replication.
For me, the beauty of this argument is that it allows to clearly see a pattern at play: antifragility requires complexity, so Natural Selection promotes the emergence of ever more complex structures of information. This is true for organisms, but also super-structures such as social organisations, social norms, laws and regulations, economic structures (from hunter-gatherers to super-specialised economies of the present), but also stories, pictures, writings, books, printing press, typewriters, computers and the Internet. The same fundamental process is at play: Natural Selection favours the antifragile, and antifragility favours complex structures. Of course, not all complex structures are antifragile, and this consideration is important, but I will tackle this distinction in a later post.
Taleb’s description of antifragility suggests all of the above, but never actually nails it, and in many ways it obfuscates the pattern that I’ve tried to show, and I can’t help to perceive this obfuscation as a very unfortunate missed opportunity. First of all, whenever he suggests what I report above, he always mentions Mother Nature, sometimes evolution. This hides the core concept because Mother Nature is used as a sort of quasi-divine super-intelligence, somewhat reminiscent of Deism. Second: the mechanism that makes Natural Selection promote antifragility is never explained, it took me a few hundred words, but while reading the book it seemed to me that Taleb didn’t really see it in clear terms. Third, and most importantly, the rhetoric of Mother Nature consistently goes too far, and clearly brings Taleb off-track, soliciting a catastrophic mistake (at least, from my own perspective). The wrong idea is that Mother Nature knows better, the infinite wisdom accumulated in aeons of trial and error makes Mother Nature incommensurately wiser than men; small stupid humans can’t hope to know any better. There is a lot of truth in this view, but also the seed of a gigantic mistake. What easily follows is the idea that what Mother Nature creates is Good (yes, our old friend, the absolute Good) and this is a dangerous concept, or worse, a suicidal misconception. First, Mother Nature includes a vast universe, and life (the major vessel of antifragility) inhabits only small corners of it (or just one corner, quite possibly). Second, Mother Nature is a human-generated concept, it doesn’t really exist and is scarcely meaningful. The meaningful concept, Natural Selection, operates only in one or few oddly shaped corners of the Universe and only there does it generate antifragile constructs. And in doing so, it generated morality, and even before that, it generated good and bad, as subjective concepts that sustain the antifragile properties of conscious creatures (not necessarily humans!). But everyone knows that Natural Selection is cruel, it is notoriously a-moral, and so is its main product: antifragility. Taleb consistently follows his mistake one further step: he seem to suggest that antifragility is inherently good. And it is not. It’s a good thing if antifragile qualities are shown by useful constructs, it’s good that our bodies are somewhat antifragile, for example. But whenever an antifragile superstructure appears on top of human bodies, this will be at best neutral, but frequently also dangerous and/or harmful.
I must add that Taleb does make a related point: he frequently states that obtaining some degree of antifragility is usually done by fragilising something else. When this transfer of fragility is done at the expense of other human beings, then Taleb is keen to label it as unethical. He is right, but misses the opportunity to show us that superstructures that are built on/with multiple humans, will generally operate this transfer: the obvious example here are societies that maintain organised armies. Such societies are somewhat antifragile because a reliable army will allow them to seize more opportunities, but they will do so at the expense of less organised societies and of the mere humans that are in the army itself. One could ask himself: is this ethical? Taleb’s take is unclear: maybe it isn’t, because the transfer of fragility is done at the expense of other human beings, but maybe it is, because an antifragile society is good…
My problem here is that never in the book Taleb finds a way to highlight the following tension: on one side Mother Nature (Natural Selection!) produces more and more antifragility, and we (human beings) like it, because in the process it made us, and when we find something that we like, it is certainly better if it is antifragile. On the other hand, the antifragility of a system is frequently obtained by making the single components of the system fragile. And Natural Selection, being an a-moral, mindless mechanism that “just happens”, keeps producing more and more complex systems, eventually fragilising human beings (think of the Great War for example)! And we don’t like that.
In summary, Taleb’s book is confusing with respect to the concepts I’ve tried to explain: it conflates Natural Selection with a seeming superior “Mother Nature”, it fails to recognise that Natural Selection creates all antifragility, it repeatedly praises the wisdom of Mother Nature (where it deserves none, because it is a-moral), and muddles the waters by stating when antifragility is unethical, but without providing a solid explanation on how to make the distinction.
So what is there to be liked? There is plenty: the first thing is what I’ve been writing today. One can see and construct a powerful picture of Universal Darwinism with the help of antifragility. Information that favours its own replication will tend to spread; and in the process, antifragile vessels will be generated. These can be biological, but can also hijack existing organisms and become super-biological, using new (biologically generated) substrates to store the replicating information (minds, books, courthouses, computers…), this is how culture was born!
There are plenty of other valuable insights that are made possible by the antifragility concept; to know what they are you’ll have to wait for the next posts.