Back in December Robin Hanson asked people if they would trade engagement with him on neglected topics. I couldn’t trade with him because I had to make my ideas clearer first, and I still have a ways to go before I can reasonably ask many others to engage much with my ideas. But he’s written an entire book on the implications of brain emulations, or ems, so it seems reasonable that I may try to engage his neglected topic of ems through my neglected topic of phenomenological complexity, and in doing so may make clearer my own ideas while more widely spreading his. If I’m lucky, I may even get some engagement back. Let’s see what happens.

Robin focuses on a possible future where brain emulation — the ability to run a human brain on artificially constructed hardware like computers — develops before other similarly disruptive technologies like superintelligence. Based on the application of the strongest results in physics, engineering, economics, psychology, and sociology, he works out the implications of such an em scenario. You can read his book, The Age of Em, for all the details, but he gives a short overview in this TED talk.

To summarize, we can imagine em cities coexisting with human cities, only em cities are much denser than human cities in order to balance communication, cooling, and energy needs. Rather than experiencing life directly in these cities, ems will live in a virtual reality running orders of magnitude faster than human subjective time so that years in em subjective time might pass in weeks or days. Although there may be millions or billions of unique ems, most ems will be copies of the few hundred most successful ems, and further still most of those copies will be spurs that are deleted after completing a specific task. Ems will live near subsistence levels and need to constantly work to support the resources they run on or retire to slower speeds.

Importantly, since ems are emulations of human brains with little modification, ems can be understood psychologically as humans living in a very different environment, comparable in difference from our modern world to the difference between humans living in forager and farmer environments. Thus much of the analysis of em behavior comes down to taking humans who evolved for forager life, spent several millennia adapting to farmer life, then spent a few hundred years adapting to industrial life (a process which is still ongoing), and asking them to adapt to em life. Most of that change is likely to be cultural, but since the em age will last for subjective millennia for the fastest running ems, some amount of “biological” adaptation can be considered, similar in level to the kinds of adaptations populations living in farmer societies for millennia have picked up.

One of the biggest changes Robin expects from ems is a return to more farmer-like values, such as more respecting social hierarchy and more carefully planning for the future. Modern humans have moved away from farmer values during the industrial era dreamtime for more forager-like values, I suspect largely due to changing amounts of relative economic excess per capita. As a result ems are likely to be more calculating, fearful, judgmental, nepotistic, self-controlled, conservative, loyal, honest, self-sacrificing, polite, hard-working, and religious than industrial era humans with far reaching effects given their technological advantages.

We see this in Robin’s expectations for child-rearing, education, training, and work habits especially. Young ems and humans who may be uploaded are likely to be educated in more varied ways to increase the probability of big winners, with the best performers moving on to join the workforce. Early adulthood is likely to be filled with much high variance apprentice and journeyman work to prepare ems/humans for later work in a variety of roles, and the most productive adult ems (native and uploaded) will later take on master-level work around a subjective age of 40, assuming industrial era individual productivity trends hold. Without biological deterioration, adult ems will likely work for subjective decades to centuries before becoming insufficiently flexible in their thinking to remain productive, at which point they will likely be forced to retire to slower speeds to make way for younger adult ems to replace them, including possibly by younger versions of themselves.

This presents an environment in which we can analyze the phenomenological complexity of ems through the lens of developmental psychology, a topic Robin has thus far ignored presumably because there is not yet strong consensus on how developmental psychology works or if it’s even a thing at all. I, however, think there are some robust results we can draw from the field, and I’ll explore their implications in order of my confidence that the underlying theory is correct.


But before we continue I want to note a couple things about what Robin has done and what I’m planning to do in response to his work. Robin is not a normal futurist. Most futurists present futures that are meant to reaffirm values they hold dear. For them futurism is less about predicting the future and more about expressing what they would like the future to be like. That’s fine, but Robin is explicitly trying to give an accurate assessment of how he thinks an em scenario would play out without expressing an opinion on what he would like the future to be like. In fact, the things you may find repugnant about an em scenario are possibly similarly dissatisfying to Robin, but that doesn’t change the nature of what we can reasonably predict. If we find the theorized em era in contradiction of our values, then so much the worse for our values.

The other thing to note is that what I’m attempting here is similar in tone to what Robin has done, so I’m going to describe possibilities rather than give you a specific narrative of the future. I don’t know what the future will look like, so at best I can point in the direction of outcomes that are likely given what we can extrapolate about the em scenario now. Thus this is less futurism and more predicting the future, and so my claims are all tentative and couched in possibility and do not necessarily promote my values. Please keep that in mind when considering these ideas.

Growing up Em

The most robust result in developmental psychology is that people mature over time, where I take “maturation” to be a proxy for phenomenological complexity. This is said without explicit reference to notions of developmental stages, just a correlation of complexity with time, or if we’re willing to take a small gamble to get a more precise statement, with variety of life experience. How does this suggest em society will differ from industrial society?

Since most ems will be subjectively middle-aged, the population of ems will skew much older than populations in most industrial societies. This implies that the average em will be more mature and phenomenologically complex than the average industrial human, so as a baseline we can look at how human preferences change with age to predict what most ems will prefer. As humans enter middle-age they remain concerned with personal values of pleasure, sex, power, prestige, and success that they hold when they are younger, but grow more concerned with abstract and social values like beauty, knowledge, health, stability, affection, belonging, support, obedience, religiosity, and tradition. This differs from a weaker abstract and social values focus when people are younger and a weaker personal values focus when people are older. From this we can expect em populations to share more balanced, conservative, and moderate views than industrial era human populations. This agrees with Robin’s idea that ems will hold more farmer-like values.

Middle-aged maturity can be viewed not just as change with time but as a growth in understanding due to compounding complexity of thought. Until the full onset of senescence, older humans become increasingly likely to exhibit more complex behaviors as measured by Michael Commons’ model of hierarchical complexity. I believe this is equivalent to saying they more think with greater phenomenological complexity as measured by Kegan, Kohlberg, and myself along the post-formal stages of thought from the traditional/pre-conventional/subject-level stage to the modern/conventional/subject-relationship stage to the post-modern/post-conventional/holonic stage. Further, although most humans in industrial societies still struggle to move beyond the subject-level stage, the people most successful in complex organizations are generally middle-aged and show signs of more often thinking at the subject-relationship and holonic stages. This suggests more ems will express thoughts and behaviors requiring subject-relationship and holonic complexity than humans do.

This may not be uniformly true of ems, though, because there are sometimes competitive advantages to less complex thought. An em who must do many repetitive, boring tasks will likely be happier if they less often experience complex thoughts because the level of task difficulty necessary for them to experience flow state will be lower. Additionally, greater phenomenological complexity appears to be mostly helpful in management and executive tasks and give little marginal advantage for performing tasks specified by others, so there may not be much selection pressure to encourage greater complexity among ems who function primarily as individual contributors within organizations. However it may still prove useful for individual contributors in em organizations if they more tightly coordinate than humans organizations do because phenomenological complexity enables more cognitive empathy and better communications skills.

So on the whole we should expect a smaller proportion of the em population to think primarily at the subject-level stage than among the human population and a larger proportion of ems to think primarily at the subject-relationship and holonic stages. In addition to further supporting the conclusion that ems will hold more balanced, moderate, and conservative views than humans do, this also suggests they will be better at self-control, more pro-social, and better at coordination than humans are on average. This slightly disagrees with the ems-are-farmer-like notion because it suggests ems will exceed industrial era humans in terms of social technology who themselves exceed farmer era humans in social technology, and unsurprisingly Robin draws a similar conclusion in The Age of Em.

Effects on Em Culture

After so much agreeing with Robin, you might wonder if there are any places where my expectations of higher average phenomenological complexity among ems predicts anything different. In fact, there are, and they come when we look closer at culture and social structures. For brevity, I will look at those topics on which I disagree with Robin. Note, though, that this section is far more speculative and based largely on my own observations and interpretations of capta rather than data. I think it’s well reasoned for what it is, but new information could also easily surprise me and cause me to change my mind. I look forward to your comments.

Because most ems will hold post-conventional moral views, I expect em law to less depend on efficient enforcement of written rules and more depend on ambiguous social enforcement of conventions based on the judgements of respected ems. What written rules there are will be fewer and more strongly enforced when the consensus is that they are useful, but most of what is today “law” will instead become the unwritten rules of polite society. Ems will be free to defect when they so choose, with the main consequence being (temporary?) ostracism or exile, and most ems will choose not to defect because already living at subsistence levels they will not be able to afford to maintain themselves for long in such a situation. However this mostly applies to the “aristocracy” of the few hundred em copy clans that make up most of the population, and more efficient law enforcement may prove necessary for em clans with few or only one copy since among these ems society will likely be even more impersonal than it is in the industrial era.

For similar reasons I suspect em religions will be less farmer-like and continue the industrial era trend towards being less moralizing, more accepting, more encouraging positive behaviors, and more tolerating negative behaviors. Em religions will be more like Unitarian Universalism and modern Chan/Zen Buddhism and less like Abrahamic and Vedic religions because they will need religions that are more amenable to taking a complete stance. Thus em religions will mostly focus on ritual, social cohesion through positive feedback, and advice and less on metaphysics, social cohesion through negative feedback, and ethics.

I can be less certain of this, but I also predict ems will have much weaker notions of personal identity than even Robin suspects. Phenomenological complexity beyond the subject-level, corresponding to post-formal stages of developmental psychology, creates a more integrated understanding of an individual’s place in the universe because the defining feature of the nominal holons of the holonic stage is that they have fluid boundaries that only come into focus by subjects taking particular perspectives. With many physical markers of individual identity made optional for ems, they will likely develop eusocial societies with ems playing roles similar to those of individual ants in a nest or bees in a hive. By late in the em era it seems possible that em society will look little like human society and much like an insect colony, at least among the few hundred copy clans that compose most of the population.


On the whole my analysis of the em scenario via phenomenological complexity agrees with Robin, and in those places where it disagrees the evidence for those disagreements is largely capta rather than data and so amenable only to phenomenological rather than scientific methods and thus outside the scope of what The Age of Em attempts. Taken together this seems to support the strength of Robin’s predictions and partially verify phenomenological complexity theory by seeing it generate results that agree with other theories so long as we restrict ourselves to evidence both may consider. Phenomenological complexity theory has a possible advantage of incorporating evidence unavailable to strictly scientific theories, but this relies on a willingness to employ phenomenological methods and evidence. Additionally, the em scenario is a rather strange theory to verify it against, but that’s what I get for engaging in neglected topics.


New Comment

New to LessWrong?