Blog‎ > ‎

Collective Technologies

posted Dec 23, 2014, 4:59 AM by Kevin Esvelt   [ updated Oct 25, 2016, 9:31 AM ]
Our society assumes that all new technologies are intended to benefit individuals. Any effects on groups of people arise from combined effects of mass adoption. To have an impact, new advances must be commercialized and purchased by consumers or individual corporations. This is true even if the benefits (or costs) of the technology are partly shared.

Consider vaccination, which protects people who are not themselves vaccinated via “herd immunity”: if enough of the population is vaccinated, pathogen replication will be too low to affect those who remain vulnerable. The smallpox vaccine enabled us to completely eradicate the virus (outside of a few controversial laboratory stocks), a remarkable achievement that continues to benefit all of humanity even though hardly anyone is vaccinated today. Yet without most of the world agreeing to be vaccinated at one point, that tremendous public health success would never have occurred.

This focus on individual adoption shouldn't be surprising, as our society is clearly built around personal freedom and choice. But it does raise the question of how we should approach advances that could provide great benefits without requiring any individual to voluntarily utilize them. How might we decide to deploy – or forgo – inherently collective technologies?

Let's define a collective technology as one that could benefit many people without ever requiring them to adopt it. For example, it might alter the “global commons”, those aspects of the world that are shared by many people. Any technology intended to alter the environment falls into this category, because everyone shares the resources produced by natural ecosystems. The air that we breathe, the water that we drink, and indeed most living things down to the pollinators of our crops are all part of the global commons.

Because we all depend on these shared resources, any technology that can unilaterally affect the global commons is morally hazardous. We have environmental laws to prevent people from destroying the commons for private gain. But we don't have a framework to govern technologies intended to improve the commons for people or for the other organisms that share our planet. That may be because there hasn't previously been a need. There just aren't that many ways of unilaterally making large changes to the environment.

Until recently.

The best-known example is geoengineering. In this case, our individual decisions have added up to a seemingly inexorable process of climate change. The problem is exacerbated by the fact that some countries have much to lose and little to gain, while the reverse is true for others. Seeing the writing on the wall, numerous scientists have proposed a variety of technologies intended to mitigate or directly counteract the effects. Some of these interventions might conceivably be enacted by a single nation, or even a single billionaire. Yet the potential effects would be distributed unevenly across people around the world. Who should determine whether these technologies are deployed, or even tested? We still don't have a good answer.

Less well known are gene drives, which can alter the characteristics of entire populations of wild organisms. Thanks to the development of new genome editing methods based on Cas9, it should be possible to build gene drives capable of editing the genomes of entire populations of sexually reproducing organisms. That is, as long as one is willing to wait for 15+ generations of the target species for the results. The gene drive itself merely allows the change to spread over generations; the effects will depend on the nature of the change. Because there are so many potential alterations to different species with varying ecological roles and geographic distributions, talking about gene drives in the aggregate is difficult; the risks and benefits would depend almost entirely on the particular species and alteration in question. But no matter the specifics, the benefits and risks posed by any given gene drive will likely be distributed unequally among people and nations. Unlike geoengineering, a gene drive wouldn't even require a billionaire backer; a single cutting-edge laboratory with adequate funding could decide to release a drive they developed for a specific purpose.

However, the genome edits spread by a gene drive could be undone using a “reversal drive”, which would overwrite earlier changes and immunize still-unaffected genomes as it spread through the population. That means the rest of the world could always decide to reverse any change made unilaterally by a small group of people, which isn't necessarily the case for geoengineering. But reversal would itself require us to alter the entire targeted population of organisms. And any ecological changes that occurred in the interim would not necessarily be reversed.

So who gets to decide whether we should release a given gene drive?

Let's focus on a concrete example: altering mosquito populations so they can't transmit malaria. As discussed in another post, there are two basic ways to do this.

Plan A involves altering the mosquito populations to reduce their ability to transmit the parasite. One version, would simply spread every naturally occurring genetic variant known to increase the mosquito's resistance to malaria. This “natural variation” drive would be unlikely to cause ecological side-effects because all of these variants are already present in the environment; we would simply be ensuring that all mosquitoes possess them. We might also build “synthetic resistance” drive to spread genes designed to boost resistance in the laboratory. These changes are also unlikely to cause ecological effects because they would primarily affect the mosquito's immune system or attack the parasite within the mosquito, neither of which is likely to affect the insect's interactions with non-pathogen species. The problem is that the malaria parasite has a tremendously large population size and is likely to evolve resistance to whatever combination of resistance mutations we choose, whether they're natural or synthetic. Plan A might succeed in dropping infections tenfold for a few years, saving hundreds of thousand or millions of lives, but malaria would return with time.

Plan B involves extirpating the most efficient mosquito vectors. Full stop. No vector, no disease, and no possibility of resistance by the parasite. It's not impossible that the mosquito population might evolve ways to resist a gene drive, but it's very unlikely given our newfound ability to target many different sites with Cas9. Unless resistance interferes with the basic mechanism of all gene drives (not impossible; some species may be intrinsically resistant due to stringent germline expression licensing, but probably not common either), we could readily engineer several genetic load drives in any species in which we've already worked out how to make one. While there are dozens of species that can transmit malaria, most of them do so very inefficiently. If we were to remove the top 6-9 vectors in Africa and Asia, the worldwide burden of malaria would collapse, allowing us to utilize more conventional measures to mop it up and drive the parasite extinct forever. At that point, we could reintroduce the extirpated mosquitoes if we wanted, although there is no guarantee that they would quickly take over their former ranges and ecological niches.

The question remains: should we do it? Should we deliberately extirpate the 6-9 species primarily responsible for malaria? Ecologically speaking, the total removal of a species from the environment is likely to have many more consequences than altering its level of resistance to a parasite. There's a reason we oppose extinction on principle (save of course for smallpox, malaria, and other terrible diseases). Some of the side-effects might even harm ecosystem services that people rely on. While these are highly unlikely to be worse than malaria, they're still something to consider. That said, there are ~3500 species of mosquito, ~430 of which are of the genus Anopheles that includes the 6-9 major targets. It's likely that the ecological niche of the extirpated species would be filled by a competitor that transmits malaria much less efficiently if at all. The sole existing study of the possible effects of malaria mosquito extirpation suggests that no single mosquito species is critically important to any other species. At least in the area examined by the study, no flower relied primarily on malarial mosquitoes for pollination, nor did any predator rely primarily on any particular mosquito species as a food source. This lack of specialization combined with the sheer diversity of mosquito species in the area and worldwide – there are 3500 mosquito species, including 430 of the genus Anopheles that harbors the 6-9 primary targets – suggests that local effects would probably be minor. This is still a single study that would need to be replicated for all candidate species in areas throughout their respective habitats. But it does suggest that we should at least consider the possibility of removing the mosquitoes for long enough to eliminate malaria.

Does that mean we should use gene drives to spread antimalarial genes through wild mosquitoes? What about extirpating the malarial vectors entirely? I don't know. To decide, we'll need to conduct broadly inclusive, fully transparent, and well-informed discussions of the potential benefits and risks. Even if the final decision should arguably be made exclusively by those who will be affected one way or another, including a variety of perspectives will ensure that all possibilities are considered. Transparency is vital to build trust in the process and the outcome, whatever that may be. And making complex decisions without considering the facts is a recipe for disaster.

Nor is there time to lose. Scientific progress should never outpace public understanding and acceptance, especially for advances relevant to collective technologies. Thanks to CRISPR/Cas9 genome editing, we should soon be able to build evolutionarily robust gene drives in many if not all species that reproduce sexually. I will be extremely surprised if we don't have viable gene drives capable of initiating Plan A in at least one if not all three major African vectors within the next five years. Can public discussions and awareness lead to a collective decision in that time?

I don't know the answer, but now is the time to begin. Any day we spend carefully testing gene drives in the laboratory before releasing them is arguably a worthwhile delay. The same is true of a day spent investigating potential ecological consequences, and certainly a day engaged in earnest society-wide discussion of whether the potential benefits are worth the risks.

But a day lost because we didn't think it necessary to begin those investigations or discussions promptly? If we ultimately decide to employ gene drives against malaria and they prove effective, each wasted day will equate to the lives of more than a thousand children. And if we don't decide to use gene drives, we'll still have helped public understanding catch up with the science and established guidelines for how to collectively manage the global commons. So let's begin.

Tags: gene drives, collective technologies, philosophy, malaria, geoengineering