Blog

Taleb on GMOs

Russ Roberts, the host of one of my favorite podcats, Econ Talk, recently had Nassim Taleb on to talk about Taleb's recent paper urging the use of the precautionary principle with regard to GMOs. I've previously commented on Taleb's paper but thought I'd add some thoughts about this recent interview.

First, Taleb's argument.  He argues that GMOs are risky and that the risks are not well understood. The risk are of the sort, he argues, that could (eventually) lead to catastrophic outcomes - i.e., they are not localized risks.  Moreover, he argues, the risks are asymmetric: the possible bad outcomes are much worse than the possible good.  He argues that with each additional new trait, there is only a small perceived increase in risk,  but that eventually - because of the "fat tails" of the risk distribution - it is inevitable that a really bad outcome will occur.  As a result, he argues, we shouldn't use GMOs  It's the cautious, wise thing to do in his assessment.

Taleb is obviously a smart guy and he makes some valid points.  In particular, I think one of the key take-aways for those of us who see value in using biotechnology - is that we want to make sure that risks are of the sort that they don't result in epidemics; that is, how can we diversify or put a stop-gap on our risk exposure?

Despite these insights, I think there are a number of challenges with Taleb's take on this topic.  First, if you listen to the interview Taleb takes great lengths to make an appeal to authority.  He establishes a hierarchy with geneticits/biologists at the bottom, statisticians above them, and then risk experts at the top.  In his assessment, the biologist is not in a position to judge risks or causal claims, it is the statistician who knows how to do this (forget the fact not all statisticians agree about the burden of proof - the whole frequentist vs. Bayesian approach is the most obvious example), and then the risk theorist is at the top (Taleb himself being the risk expert).  As such, he essentially tries to inoculate himself from any criticism, saying that a biologist can't criticize the risk theorist (Russ Roberts calls him out on this at one point, but Taleb quickly dismissed the point).  

To make his point, Taleb invokes the carpenter fallacy.  The carpenter makes the roulette wheel, but if we wanted to know our chances of hitting on on the number 22 after 150 plays, would we ask the carpenter?  No, according to Taleb, we'd ask the probability expert.  But, there is a huge assumption being made here: that the probability expert knows how the wheel is made - how many possible numbers there are, whether the wheel is balanced, whether the slots in the wheel are of equal size, etc., etc.  In short, a good probability theorist needs to know everything the carpenter knows and more.

Unfortunately, I don't see much evidence that Taleb has spent much time trying to understand GMOs or modern agriculture, and as such it is hard to take his probability judgments in this domain seriously.  For example, one of his examples is that the Irish potato famine, which he argues was caused by a lack genetic diversity.  However, there are good reasons to believe that British politics were the key contributing factor to the famine.  Moreover, what he doesn't seem to get with regard to modern GMOs is that a GMO isn't a variety.  A particular trait - say herbicide resistance - is introduced into many, many varieties in different parts of the country and the world. Moreover, not all herbicide-resistant crops are resistant to the same herbicide.        

Going further, herbicide residence can be "naturally" bred into plants.  There are rice and wheat varieties on the market that are not GMOs but that are herbicide resistant.   Why are these not risky but the GMOs are?  My original comment about Taleb's paper is that he didn't focus on marginal risks; he treats GMOs as a separate class without looking at how plant breeding is done in real-life agriculture: is he also against hybrids? Mutagenesis? Cisgenics? Marker Assisted Breeding?  These all have risks that are on par with GMOs, so I'm curious why only the focus on one particular technique?

As I've sad many times, a GMO isn't a single thing - it is many, many things, each with different benefits and risks.  A hammer can be used to bash heads or build skyscrapers.  It is just a tool.  Same with biotechnology.  What I would prefer to see from Taleb and others is a careful discussion of a particular trait they find worrying along with a careful articulation of why that particular trait is likely to lead to a particular global harm - and what it is about the tool of biotechnology that is particularly worrisome for that trait vs. other breeding techniques?  If everyone was planting the same variety of the same herbicide resistant crop (and we only had one herbicide that we knew worked), I'd be concerned, but that not what's happening.

As it is, we simply have Taleb making an appeal authority and to unknowability.  We can't know what bad things might happen, so Taleb says we should be cautious.  But no evidence of risk is not evidence of an eventual black swan.  Indeed, it seems to me it is a recipe for stagnation.  And it isn't at all clear to me the downside risks are always greater than the upside.  When penicillin was first discovered, the risks were unknown but the benefits were (ultimately) immense.  A precautionary approach might have shelved antibiotics, but we took the chance and with great effect to our life expectancy.  Life is full of risk.  The answer isn't to hole ourselves up in the closet, but rather to think about ways of taking insurance against possible risks while venturing out into the world.

I'll conclude with one last thought.  Taleb makes reference to the Hayek bottom-up vs. top-down planning.  He says GMOs are the top-down sort.  I'm not so sure.  Real life farmers and people have to be willing to buy varieties that have the GMO traits.  No one is forcing that outcome.  It is true that competition will limit - to some extent - the diversity of plants and genetics that are observed because some plants aren't tasty or aren't high enough yielding.  But most plant breeders keep all kinds of "ancient" varieties precisely for the purpose of trying to breed in new traits to today's varieties (and folks working on synthetic biology are creating their own, new strands of DNA, creating new diversity).  Geography also increases diversity.  Iowa grows a lot of corn, Oklahoma doesn't because it isn't our comparative advantage.  I see little reason to believe that a single GMO variety will perform well in all locations.  So, yes individual companies are planning and creating new varieties, but it is all our local knowledge of what works in our places and conditions that determine whether particular genetics offered by a particular company are used.  We do not have a seed czar or a DNA czar.