You can never be too safe.

Photographer: Sion Touhig/Getty Images

Black Swans, Frankenfoods and Disaster Fairy Tales

Noah Smith is a Bloomberg View columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.
Read More.
a | A

The 17th-century French mathematician Blaise Pascal had what he thought was a great reason to believe in God -- it’s all about expected utility. If you believe in God and God turns out not to be real, it’s not so bad. But if you deny God and it turns out that God does exist, you get sent to Hell for all eternity. Even if the chance that God exists is infinitesimally small, the downside is so huge that you would be a fool not to believe.

Nassim Taleb, author of "The Black Swan" and other books, believes that there are other such wagers around. The idea of “black swans” combines the concept of tail events -- things that are so rare that you can’t estimate their frequency from past data -- with the concept of special cause, the idea that the future won’t necessarily look like the past. Normal risk management doesn’t work if you can’t use the past as a quantitative guide to the future. So to combat the threat of black swans, Taleb and his co-authors on a recent working paper suggest that we adopt the “precautionary principle” -- a close cousin to the strategy used by Pascal:

The precautionary principle (PP) states that if an action or policy has a suspected risk of causing severe harm to the public domain (affecting general health or the environment globally), the action should not be taken in the absence of scientific near-certainty about its safety. Under these conditions,

the burden of proof about absence of harm falls on those proposing an action, not those opposing it.

Taleb et al. apply the PP to the problem of genetically modified food, or GMOs, and conclude that the technology shouldn’t be used. My Bloomberg View colleague, physicist Mark Buchanan, is persuaded by their arguments.

GMOs and Engineered Food

I, however, am not. Certainly, the PP makes sense in the abstract, at least in certain situations. But should it really be applied to GMO crops? Maybe, but Taleb et al. have failed to convince me.

The key question is whether GMOs, whatever their advantages, really have the potential to cause truly ruinous harm. This question has nothing to do with probability or decision theory, and everything to do with biology. Taleb et al. don’t spin a convincing story for how GMOs might destroy us.

The basic GMO disaster story goes something like this: Genes from GMO crops diffuse into the general population, as GMO plants interbreed with surrounding non-GMO plants. This creates a global monoculture, which will then put us at risk of a pandemic disease that causes a catastrophic crop failure. That would then put humanity on the brink of starvation.

I see problems with every link in this narrative chain. First of all, there doesn't seem to be evidence that natural spreading of modified genes creates global monocultures. The natural spreading of modified genes, via accidental interbreeding, is a slow process, and is itself subject to evolutionary modification as it occurs. Crops, meanwhile, are grown all over the world -- for a modified gene to spread so rapidly and completely that it creates a global monoculture on a short time scale would be utterly different, both quantitatively and qualitatively, from any phenomenon ever observed or theorized in any scientific theory.

Of course, humans ourselves can foster monocultures -- we did this during the Green Revolution, and we may be doing it with some GMO crops now. But since the 1960s we have learned to blunt some of the risks of monocultures by using precautions like seed banks. Those will come in just as handy with GMO monocultures as they did with the earlier cases.

Next, the threat of a global pandemic plague decimating GMO crops doesn't seem credible. Crop epidemics and pandemics do occur, of course, and monocultures really are more vulnerable. But it is very difficult for pandemics to spread across national borders. Remember how on customs forms, you always have to report whether you’re bringing any plants into the country? That’s why.

Of course, that doesn’t make it impossible that pandemics will spread from country to country, but would a pandemic spread to all countries that produce a given crop within a single growing season, causing a global harvest to fail? No. And if a pandemic takes more than one growing season to spread, then it will never be a pandemic, because after the harvest, a whole new crop will be planted, and that crop will use seeds that weren't hit by the pandemic, thus neutralizing the threat.

Finally, there is the question of whether a crop pandemic, even if it could happen, would put humanity at risk of ruin. The answer is no. Even the staple crops -- corn, soybeans, rice, and wheat -- each supply less than a fifth of humanity’s calories. A total global loss of one year’s harvest of a single staple crop would be a terrible event, of course, and there would be localized famines, but humanity wouldn't die off. This takes us out of the Pascal’s Wager territory of the precautionary principle, and back into the realm of simple caution.

I’m not arguing that GMOs are completely safe; as with any new technology, there are bound to be risks we haven’t foreseen, and could never foresee. But they don’t seem to constitute the ruinous risk that Taleb et al. are worried about. The worst-case scenario -- a global loss of 20 percent of humanity’s calories in one year -- would be bad, but not bad enough to make Pascal get religion.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.

To contact the author on this story:
Noah Smith at

To contact the editor on this story:
James Greiff at