Robin Hanson, The Age of Em [Review]

Originally published 14 December 2016 but lost in the Great Server Mistake of 2017.

Attention Conservation Notice: Robin Hanson has written a provocative book illuminating the contributions social science can make to understanding the future, even if the details are (amply) debatable.

What will the future look like? Social scientists pay a great deal less attention to this question than they should. There are departments of History, but no departments of Future–nor even a great number of academics whose primary research program explores the future.

There are good reasons for this. All of our evidence about how humans and human societies behave exists in the past (yes, literally all of it). The future, by contrast, may not even occur. Theorizing about the future, then, appears from one direction to look like nothing so much as a modern version of medieval Scholasticism–or, as the demotic version goes, like playing tennis without a net.

But the optimal level of effort to be invested in thinking seriously about the future is not zero, or close to zero. Indeed, it’s interesting that businesses and governments are more likely, not less, to invest resources in trying to estimate at least the parameters within which the near future will take place — for instance, as reflected in the US intelligence community’s Global Trends reports or Bill Gates’s now-forgotten The Road Ahead. More to the point, even though all of our evidence comes from the past, none of us will live there–and (almost) all of us will live in the future.

Robin Hanson’s Age of Em represents a significant intervention in debates about what kind of futures to envision, how to envision those futures, and why we should do so in the first place. Hanson’s future concerns a post-human society within the next 100 to 1,000 years in which whole-brain-scanned humans exist as emulations (or “ems”) within a joint virtual-physical world. Within this civilization, the limits of physics trump the limits of biological life, and ems are able to work and live in a population that expands at rates closer to insects or bacteria than to homo sapiens. As a consequence, the GDP of em-world increases rapidly—doubling every month or so—even as the population heads towards the trillions and wage rates plummet toward subsistence (which, Hanson notes, is the historical regularity within human societies over time and for most life-forms generally).

If this seems strange, it is: the Guardian‘s review was on the mark when it termed this a “strange, yet serious” book. Part of the strangeness derives from the fact that any world whose economic and consequently cultural milieu will be so different in form from our own will necessarily seem strange to us: reading seriously in another culture’s history or daily life eventually forces us to confront how contingent and bizarre both their lives and our own are. (As an illustration, consider Brad DeLong’s gloss on Cosma Shalizi’s argument that the Singularity is in our past–that the world we live in is as incomprehensible to a savannah-dwelling anthropoid as we imagine the Matrix would be to us.)

Another part of the strangeness of the book derives from its simultaneously playful but earnest argument to show that extrapolations from the present to the future prominent in STEM imaginations neglect the insights, and cautions, that the social sciences provide. Although Hanson makes detailed reference to the physics of the em world, from computing cycles to the ways that wind, not gravity, will affect the physical manifestations of em cities, his relentless focus is on how em society will work and understand itself.

In a sense, then, this is a book of “exotic social science” in the same way that (some) physicists look for “exotic physics“. It is speculative both in the sense that Hanson’s predictions are unfalsifiable on any (current) human lifespan; it is also speculative in the sense that Hanson, a pioneer in prediction markets, thinks that he can assign a weight to his subjective beliefs.

But it is not speculative in a more important sense: that social sciences should be involved in answering those what, why, and how questions about the future. Hanson places a big wager: that much of what social science theory has demonstrated about the ways that individuals interact under institutions is as universal as certain concepts in the physical sciences. To put it another way, commodity markets in the Klingon Empire will work the same way as markets in the Babylonian empire. From that insight, Hanson is able to show that many common problems will continue to recur: ems will have to worry about crime, love, property rights, and a whole host of other problems.

My biggest concern about the book, in fact, is that it is altogether too optimistic. Hanson does not seriously treat the ways that coordination failures, agency failures, and negotiation failures can lead to warfare, predation, and rent-extraction in ways that could make em futures nastier, more brutish, and shorter than most of Hanson’s book supposes (the longest description of wars in em world takes place over 2.5 pages, on pp. 250-252). What will ems fight for? If they worry over status, group together in clans, and seek to protect their intellectual property assets, then one would think that the degree of indivisibility involved in those issues of esteem, loyalty, and property would make war more, rather than less, likely. In other words, even though ems may look science-fictional in some regards, they might also well wage organized violence like the Greeks did against the Trojans. The possibility of war, moreover, leaves the door open to the emergence of central authorities with coercive powers designed to coordinate collective (violent) action–and at that point we need an account of the state, not of the more generally anarcho-libertarian state that Hanson has envisioned.

Of course, the same process that creates ems might also enable personality modification to endorse strongly pro-social behaviors to avoid these problems. But at that point, the chance emergence of a predatory em, even a single one with the ability to reproduce, might destabilize the pro-social equilibrium in the same way that a single lion among a herd of sheep can induce massive changes. Adaptations to resist such outcomes might lead to a differentiation of em types, from doves to lions to hawks to vultures, and thus an em society differentiated not just by their computing cycles’ speed or specific intellectual capital but also for their propensity to engage in coercion, deception, and cooperation.

Yet even though I am (after one reading) less sanguine than Hanson about the ways in which em society might be realized, I am nevertheless grateful for the book for serving as a serious, worthy intervention in a debate that social scientists aren’t even having. Too much work is driven by contemporary standards of what is and what is not acceptable. The social sciences would be better if we stood more orthogonally to contemporary issues (not just at the policy relevance level but even at the Zeitgeist-y one): doing so would allow us to improve not only our theories but, by doing so, the sorts of evaluations we could make to society. (You may not think relativity matters to you, but if you use GPS, you’re indebted to it.) Engaging in disciplined debates about what deep factors drive change in society would prove one way to do so.

Indeed, the biggest argument that social sciences should have is about the relationship between our object of study–human societies–and the fact that human beings are just another biological lifeform. Robust and sustained engagement among anthropological social sciences and ethologists could lead to understandings about what parts of human society derives from factors that would be common to any complex agglomeration of agents and which derive from factors that originate in the limitations common to DNA-based agents that evolved on the sunlit earth. This question may seem strange: do we really need to investigate whether humans’ status as humans leads them to have different societies than non-human intelligent life would have? But as one thinks through it, this is a central if unrecognized theme in debates between “rationalists” and “interpretivists” within and among different social sciences. Work like Hanson’s should stimulate large-scale rethinking of our ideas on these and other subjects.