Social Science Critics

Many critics of Age of Em are critics of social science; they suggest that even though we might be able to use today’s physics or computer science to guess at futures, social science is far less useful.

For example At Crooked Timber Henry Farrell was “a lot more skeptical that social science can help you make predictions”, though he was more skeptical about thinking in terms of markets than in terms of “vast and distributed hierarchies of exploitation”, as these “generate complexities” instead of “ breaking them down.”

At Science Fact & Science Fiction Concatenation, Jonathan Cowie suggests social science only applies to biological creatures:

While Hanson’s treatise is engaging and interesting, I confess that personally I simply do not buy into it. Not only have I read too much SF to think that em life will be as prescriptive as Hanson portrays, but coming from the biological sciences, I am acutely aware of the frailties of the human brain hence mind (on a psychobiological basis). Furthermore, I am uncomfortable in the way that the social science works Hanson draws upon to support his em conclusions: it is an apples and oranges thing, I do not think that they can readily translate from one to the other; from real life sociobiological constructs to, in effect, machine code. There is much we simply do not know about this, as yet, untrodden land glimpsed from afar.

At Ricochet, John Walker suggests we can’t do social science if we don’t know detail stories of specific lives:

The book is simultaneously breathtaking and tedious. The author tries to work out every aspect of em society: the structure of cities, economics, law, social structure, love, trust, governance, religion, customs, and more. Much of this strikes me as highly speculative, especially since we don’t know anything about the actual experience of living as an em or how we will make the transition from our present society to one dominated by ems.

At his blog, Lance Fortnow suggests my social science assumes too much rationality:

I don’t agree with all of Hanson’s conclusions, in particular he expects a certain rationality from ems that we don’t often see in humans, and if ems are just human emulations, they may not want a short life and long retirement. Perhaps this book isn’t about ems and robots at all, but about Hanson’s vision of human-like creatures as true economic beings as he espouses in his blog. Not sure it is a world I’d like to be a part of, but it’s a fascinating world nevertheless.

At Entropy Chat List, Rafal Smigrodzki suggests social science doesn’t apply if ems adjust their brain design:

My second major objection: Your pervasive assumption that em will remain largely static in their overall structure and function. I think this assumption is at least as unlikely as the em-before-AI assumption. Imagine .. you have the detailed knowledge of your own mind, the tools to modify it, and the ability to generate millions of copies to try out various modifications. .. you do analyze this possibility, you consider some options but in the end you still assume ems will be just like us. Of course, if ems are not like us, then a lot of the detailed sociological research produced on humans would not be very applicable to their world and the book would have to be shorter, but then it might be a better one. In one chapter you mention that lesbian women make more money and therefore lesbian ems might make money as well. This comes at the end of many levels of suspension of disbelief, making the sociology/gender/psychology chapters quite exhausting.

At his blog, J Storrs Hall said something similar:

Robin’s scenario precludes some of these concerns by being very specific to a single possibility: that we have the technology to copy off any single particular human brain, we don’t understand them well enough to modify them arbitrarily. Thus they have to operated in a virtual reality that is reasonably close to a simulated physical world. There is a good reason for doing it this way, of course: that’s the only uploading scenario in which all the social science studies and papers and results and so forth can be assumed to still apply.

Most social scientists, and especially most economists, don’t see what they have learned as being quite so fragile. Yes it is nice to check abstract theories against concrete anecdotes, but in fact most who publish papers do little such checking, and their results only suffer modestly from the lack. Yes being non-biological, or messing a bit with brain design, may make some modest differences. But most social science theory just isn’t that sensitive to such details. As I say in the book:

Our economic theories apply reasonably well not only to other classes and regions within rich nations today, but also to other very different nations today and to people and places thousands of years ago. Furthermore, formal economic models apply widely even though quite alien creatures usually populate them, that is, selfish rational strategic agents who never forget or make mistakes. If economic theory built using such agents can apply to us today, it can plausibly apply to future ems.

The human brain is a very large complex legacy system whose designer did not put a priority on making it easy to understand, modify, or redesign. That should greatly limit the rate at which big useful redesign is possible.

GD Star Rating
Tagged as: ,
Trackback URL:
  • Matt M

    What do you think is the most compelling criticism of your thesis? That is, what is the most likely way in which you are wrong?

    • The most likely way the book is irrelevant is that ems don’t in fact come before other AI, or another enormous social change.

  • “But most social science theory just isn’t that sensitive to such details.”

    That would explain why when psychology papers fail to replicate, the original authors often complain that some small detail of the original study not present in the replication was critical to produce the effect.

    I recommend readers check out this recent essay by Andrew Gelman which explains how common shoddy social science research is, as a counterpoint to Robin: Note that the replication crisis is not just in psychology–it’s also happening in experimental economics, for instance.

    Overall your critics seem on point to me. It’s common wisdom in psychology that research done on WEIRD subjects should be generalized with care. If you think social science is reliable, we should trust it when it says it’s unreliable! And WEIRDness will be nothing compared to the differences between meat humans and ems. Modeling ems as rational agents is probably a good first guess, but saying that lesbian ems will earn more because lesbian women earn more feels pretty tenuous.

    • Robin’s solution to the replication crisis involves prediction markets. But I don’t think he’s addressed whether it might justify discounting social-science findinfgs.

    • Distinguish a topic and how well insights on it would generalize from researchers and how sloppy they are. One can do sloppy research on a topic that generalizes very well. I’ve tried to stick to less sloppy work, and to flag my uncertainty. I flag the guess about lesbians as a weak guess.

      You don’t at all know that “WEIRDness will be nothing compared to the differences between meat humans and ems.” You’d have to apply standard theory to make a reasonable guess on that.

      • Joe

        I don’t remember how you addressed this in the book but – surely lesbian humans earning higher wages just wouldn’t translate into lesbian ems earning higher wages, because of the crucial point that all ems earn roughly the same wage?

      • I draw inferences about the fraction of ems who are gay and lesbian, not about their wages.

      • I guess I would like to see a lot more citations in this post. The fact that you failed to address either the replication crisis or WEIRDness, and chose instead to make forceful and unsupported blanked statements, undermined a lot of my faith in you as someone who can do a good job of representing the state of the debate on a topic.

        You’re falling prey to the same old biases you regularly see in others. In this case, it’s about defending your status as the author of a book you’re trying to get people to buy. (Maybe we can design some clever new publishing institution to solve this incentives problem? :P)

        Since this post isn’t actually about social science (it’s really about defending your status), I’ll help you out. All you have to tell people is that in forecasting, there’s a precision tradeoff: you can keep your error bounds wide and constantly be talking in fuzzy generalities, or you can keep your error bounds narrow and try to make a point estimate of what one particular scenario near the center of the distribution might look like. Say that you chose to try to make a point estimate because it’s more engaging for readers, and because it’s easier to work out all the ramifications of a single scenario (that’s basically your specialty as an economist). Then say that you used the best social science available at the time you wrote your book to inform your point estimate. This excuse future-proofs you against arbitrarily bad replication crises.

      • WEIRDness and the replication crisis aren’t directly related to Hansons thesis, which is that a general skepticism about social science underlies much automatic rejection of his conclusions.

      • Well he also says that skepticism is unjustified. I argue that it is in fact justified to an extent.

      • My critique is that he was impressionistic rather than analytical in his showing that skepticism about social science grounds much criticism of his em hypothesis. I wouldn’t expect him to take up the actual defense of social science in the same piece.

    • What odds would you give that the apparent higher wages for lesbians is actually an error, and they actually have lower wages?

  • Romeo Stevens

    Another point might be that social science has made insights that apply across many brain architectures in animals.

  • Yes, the human brain didn’t evolve to be easy to *subtly* modify.

    However, you predict a huge number of em-instances and admit the motivation and productivity of these instances are hugely important giving a huge economic incentive to develop modification techniques. These techniques can be developed by experimenting on ems (or even just em-pieces) run at high rates of speed. This plus the ability to separately modify each little piece of the simulation would seem to make modification quite plausible.

    Moreover, it is relatively easy to non-subtly modify the brain (consider the various CNS drugs we have) and this has proved remarkably beneficial in areas we might want to apply em-labor. Erdos (perhaps the most productive mathematician of the 20th century) famously did all his math on amphetamines (medical dose not crystal). Of course use of things like amphetamines to aid focus/concentration is limited by issues like tolerance and the fact that it can disrupt other aspects of biological life but these aren’t a problem for ems.

    Given that we have a fair number of drugs which can boost performance on various task while also substantially affecting how people interact and socialized (merely making everyone super shy, focused and standoffish could result in huge social changes) and these are drugs that must cross the BBB and were developed without the kind of microphysical understanding of the brain emulation would require and I’m confident modification would be the norm.

    • Much of economics is robust enough to describe people who are chewing cocoa, taking amphetamines, or castrated.

      • marshall bolton

        I had a client once, who was told that if he ever injected himself in the leg again, he would lose his leg. What was the first thing he did, when he was released from hospital?

      • Economics. Yah I’ll but that. It is the projections about em socialization, values and other sociology/psychology style assumptions that I am concerned about.

        I expect economics to describe extraterrestials when we meet them. I don’t expect them to pair bond, engage in human social structures or leisure. While small molecule drugs won’t totally undo those inclinations it does alter relative the subtle balances that dictate whether we are social or anti-social, looking for long term or short term mate investment etc.. And with simulations we can do way better than small molecule drugs.

        But I doubt this will convince you.

  • marshall bolton

    How can a machine be lesbian? This is surely a ridiculous idea – a confusion of logical types, if you will. This story about Ems and their inevitability seems to be based on the Laws of Economics which themselves seem to be based on a few assumptions about how agents behave: self-interest, competition and prices determine well everything. Thus Ems are inevitable – once the are “conceived”. Economics prides itself in explaining lots of phenomena in lots of fields. But what happens if the agents being explained are no longer homogeneous? Does self-interest, competition and prices still work? The concept of Ems (the Machine / the Computer) is such a bastardization of categories that I see no reason to believe that they will show self-interest and thus Robin’s scenario is just an exercise in rationality. Has an economist ever tried to talk with an unhappy woman who may or may not be lesbian, who may or may not be lots of things….. If they had, then they would no longer believe in the Laws of Anything.

  • Perseus

    A bit of topic – but interesting read – molecular sociology -

    Sociological reseach on interaction in social networks will not be applicaple to this Research stream, but just the fact they use this taxonomy is interesting

  • Pingback: Overcoming Bias : No One Rules The World()

  • There are frequently criticisms of the social sciences out of hand, because it’s “not empirical enough”. It’s a part of the Humanities vs. Science false dichotomy, which assumes that we can only look at problems in one way, and that way is Science. But what we have here are different tools for different needs, and we need to be confident with all of them to properly tackle a complex problem like that described in the article.

    I have a friend who has an even stricter view on this: “The Science part is easy because the data is what it is. You need the Humanities for the hard part- to think about what the data means, and what we should do as a result.”

  • Pingback: Links for October 2016 – foreXiv()