January 25, 2013

Rethinking the Long Term

In recent weeks I have been perusing the seminars and written works of The Long Now Foundation, whose stated mission is “to provide a counterpoint to today’s accelerating culture and help make long-term thinking more common.” This certainly seems an admirable goal, and the foundation’s projects do a superb job of melding science and engineering together with artistic and cultural sensibilities – a prime example of which is the 10,000 Year Clock, a 200-foot-tall multi-million-dollar monument being built inside a cave in a remote western Texas mountain that, as the name implies, is designed to mark the passage of time for the next ten millennia.

The Long Now Foundation emphasizes the importance of our perception of the passage of time, and indeed our cultural conceptions of the passage of time. (J. Stephen Lansing, for example, shares insights into the role that language plays in shaping our perception and conception of time by discussing the case of Polynesian and Austranesian languages that do not have tenses but instead construe time in “multiple concurrent cycles”). More specifically, The Long Now Foundation asserts that long-term thinking is in short supply, and that in the face of accelerating technological change our culture needs more rather than less of it if we are to avoid both imperiling and impoverishing future generations.

During my masters program at the University of Michigan’s School of Natural Resources and Environment I received two years of formal scientific training in how to understand the sustainability and resilience of complex systems, both in theoretical and empirical terms. This training placed much the same importance on long-term thinking as The Long Now Foundation, and for similar reasons: both are predicated on an underlying set of assumptions about the finitude and fragility of our world.

In this essay I am going to revisit three of these assumptions, and ask which if any are likely to continue to hold true for the indefinite future – say, for the next 10,000 years. I hope to show that they actually reduce to a single assumption that will inevitably – and rather quickly – prove to be false: that biology, whether human or nonhuman, is immutable. I should hasten to add, however, that this emphasizes rather than diminishes the importance of long-term thinking and environmental conservation.

Assumption 1: we will continue to value and depend upon the same resources indefinitely.

It is an incontrovertible fact that the Earth’s natural resources are finite. As Carl Sagan so famously observed in Pale Blue Dot, our entire planet is but a mote of dust in an almost inconceivably vast cosmic ocean. It is also incontrovertible that the currently dominant economic paradigm fails to adequately account for this fact, and that nascent fields such ecological economics do a far better job of grounding our understanding of human consumptive activity in terms of real-world biophysical limits.

However, the natural resources that we consume today are not the same as those we consumed in the past. Prior to the 20th Century, for example, humanity consumed virtually no rare earth elements. Today yttrium, neodymium and some 15 other rare earth elements are in high demand because of the role they play in the manufacture of modern computers and electronic devices. Prior to the 19th Century humanity consumed virtually no fossil fuel. Today our entire global civilization is utterly dependent upon coal, oil and natural gas.

Of course, some natural resources have been crucial throughout human history. The most obvious examples are breathable air, potable water, fertile soil, healthy ecosystems capable of producing food, and the full range of ecological services. Although the relative value of these resources compared to other goods and services has changed dramatically over the last several centuries, there is no question that human beings still need air to breathe, water to drink, food to eat, and the means to clothe and shelter themselves.

Here we must ask two questions: 1) Will we continue to value these same resources indefinitely? 2) Will we continue to need these same resources indefinitely? At some point in the future, likely before the end of the 21st Century, the answer to both of these question will be no.

Regarding the first question, some futurists envision technologies that will render all matter fungible. An advanced 3D printer capable of constructing objects from their atomic constituents, for example, could manufacture air and water and food on demand, so long as the necessary inputs of raw elements and energy were provided. An even more advanced 3D printer similar to Star Trek’s replicator, capable of transmuting matter and energy and thereby constructing objects from subatomic constituents, could manufacture elements themselves and would therefore be subject to little or no input constraints at all. The advent of such technologies would end resource scarcity entirely, thereby upending the foundation of economics as we know it, but more to the present point it would fundamentally alter our relationship with the natural world.

It might seem as though this technology must lie centuries away, but 3D printing is already a billion-dollar industry and physical objects are downloadable today from the Pirate Bay. The more optimistic estimates by futurists such as Ray Kurzweil forecast the arrival of molecular-resolution printing of macroscopic objects (including all of the components required to replicate the printer itself) within 50 years.

Regarding the second question, it is easy to assume that we will always need air and food and water and shelter. And nearly everyone you might care to ask does assume so. But what this really means is that we are assuming that humans, no matter how far into the future we look, we still be humans, meaning biologically identical to ourselves today.

Again, optimistic futurists forecast the arrival technologies that enable full biological transcendence within 50 years or so. If we accept these forecasts, technology will allow us to upload our minds into non-biological substrates, namely to the computing cloud of the future. With our conscious minds safely decoupled from their fragile biological origins (and hence indestructible and immortal), we will be able to temporarily download ourselves into whatever physical bodies we require at any given time. These bodies might take the form of good-old-fashioned Homo sapiens, or they might be something completely different – a bald eagle or a spacecraft, for example.

Perhaps even more radically, futurists imagine that the conscious minds of the future (whether of human or artificial origin) will spend more of their time in virtual reality than actual brick-and-mortar reality. Many of our more sophisticated human needs, such as entertainment and creative expression, may be far more fully realized in virtual realities than in actual reality, since virtual worlds will have far fewer constraints. Again, this is a revolution that has already begun: millions of people today fulfill some of their needs for entertainment, creative expression and social interaction online. This is a trend whose trajectory will lead to the creation of entire universes where the only limit is our imagination.

(I should note here that while the timescale of these predictions may seem impossibly short, the arguments stand whether the actual number proves to be 50 or 500 years).

Assumption 2: we are doing irreparable damage to our world.

Many ecosystems are fragile, while others are quite robust. Either way, as a direct result of human activity virtually every ecosystem on the planet is in decline. Only a handful of ecosystems are resilient enough to absorb our blows and retain their general configuration. Some number of others cannot retain their original configuration but nonetheless have the adaptive capacity to shift to a new configuration, albeit usually one of less complex functioning. But far too many, sadly, are so fragile that they are collapsing or have already collapsed.

Ultimately, our reckless degradation of the world’s biosphere is self-destructive. As E.O. Wilson has so elegantly articulated, the biodiversity and complexity of ecosystems represents an enormous reservoir of natural capital that we liquidate at our peril. The question is, how much of the damage that we are doing is irreparable?

It is politically incorrect to suggest that technological advances may one day allow us to repair the ecological destruction we have wrought on our world. There is an unspoken fear that if we admit today’s problems might be soluble in the future, we will be less inclined to mitigate our presently destructive behavior. This is certainly a justifiable fear. However, as a scientist I am obligated to eschew political correctness in favor of honesty, and my honest opinion is that in the long term (and here I mean decades, not millennia) only a small fraction of the ecological damage we are doing will prove truly irreparable.

Biology is a form of molecular technology based upon a digital code: DNA. Today the sophistication of that technology remains daunting. Although we have recently acquired the ability to read and write DNA code base-pair by base-pair and to print entire genomes, we do not yet have the power to modify living organisms at the molecular level, or to resurrect entire organisms based upon their DNA code alone. But optimistic futurists foresee these capacities arriving in around 50 years, not coincidentally in conjunction with 3D molecular printing and computers trillions of times more powerful than those of today. And it is conceivable that we could achieve this ability even without the assistance of superhuman intelligence, which, again not coincidentally, is also expected to arrive within 50 years or so.

However, even if we do acquire the power to resurrect extinct species, we will likely need the original DNA to work from. Without a sample of Siberian tiger DNA we might be able to recreate an organism very much like a Siberian tiger, possessing all of the attributes we knew it to have, but in an important sense it wouldn’t be a Siberian tiger; it would be a facsimile. In a very real sense, DNA samples are like seeds or eggs, just waiting for the right set of circumstances under which to develop. Thankfully there are already concerted efforts underway to preserve the DNA of endangered species.

However, a larger question about the importance of authenticity arises here. Is an original always more valuable than a recreation? Even if we were to recreate the Mona Lisa atom by atom, it still wouldn’t be the Mona Lisa. A copy, no matter how perfect, is still a copy. Does the same objection apply to living things? Perhaps, but it is useful to note one important feature that distinguishes living organisms from an object like the Mona Lisa: the matter of which living things are composed changes over time. Unlike Da Vinci’s masterpiece, the atoms you and I are comprised of today are not the same as the atoms we were comprised of a year ago. Virtually every bit of our bodies is replaced at regular intervals. I am not quite the same stuff I was even yesterday, and yet here I am. Like a river or a forest, what I consist of over time is not a particular clump of matter, but a recognizably distinct pattern. I offer no conclusive answers here; I merely point out an important and open question.

Biodiversity loss is not the only type of environmental damage caused by human activity. What about pollution? One thing to note here is that we define the term relative to biology: a pollutant is anything that cannot be accommodated by normal ecosystem functioning. In the future, synthetic biology together with inorganic molecular machines are likely to be given the task of processing our waste. Research is already underway to create microorganisms capable of processing oil spills, for example. In the near term these technologies will be primitive and potentially dangerous – we certainly don’t want to unleash a microbe that mindlessly turns all plastic in the world into dirt, since so many of our products and machines contain plastics. But in the long term we will be able to direct swarms of semi-intelligent nanobots the size of mites or even ants to the task of environmental decontamination. And just as ants will happily collect every grain of sugar spilled on your kitchen floor, these machines could cleanse entire landscapes of everything from snickers wrappers to radioactive dust. The predicted arrival time of nanobot swarms? You guessed it: around 50 years.

It therefore seems to me more important to ask exactly when and how we will repair the ecological damage we have done to the Earth, rather than to lament the damage as irreversible. Under what future circumstances might we restore large portions of the planet’s surface to a pre-industrial state, and how can we precipitate the arrival of such circumstances? What is the optimal state to which we should turn back the dials of time? Many of the world’s terrestrial ecosystems, for example, have been significantly modified by human activity for at least 15,000 years. Should we restore North America to its pre-Columbian configuration, or to its pre-Homo sapiens configuration? If the latter, should we strive to resurrect the wooly mammoths, saber-toothed tigers and other megafauna that disappeared after the arrival of early humans? Again, I don’t pretend to offer answers; I am merely identifying the types of questions we need to begin asking. And note that these questions are not just technological or ecological but also political, socioeconomic, ethical, and philosophical.

Assumption 3: the pace at which we experience the world is fixed.

What does long term actually mean? It certainly seems as though 10,000 years is a long time, even relative to the longest human lifespan of around a century. But this begs the question, what is a short time? Although we may have cultural and linguistic differences, all human beings must mark the passage of time relative to natural rhythms. The pace of our lives is intimately tuned to the biological rhythms of the other living things on which we depend, as well as to the rhythms of the Earth itself. Our minds evolved in synchrony to these rhythms, and so we are built to acquire and value life experiences according to the passing of the days, the rise and false of the tides, and the pulse of the seasons.

The Long Now Foundation expresses concern that as the pace of activity in our lives accelerates due to technological advancement, we are coming unmoored from the natural rhythms to which we evolved. In the process, we are losing our already feeble ability to properly situate the events of our own lives within the context of the broader sweep of human history. Next quarter’s profits become more important than the next generation’s health, and the integrity of election cycles becomes more important than the integrity of the ecological cycles that sustain us.

I am certainly sympathetic to this view, but I would also caution against unwarranted anthropocentrism. With a steering wheel in one hand and a smartphone in the other, our accelerating lives may seem frenetic. But relative to ants or houseflies, our moment-to-moment experiences are rather leisurely affairs. And compared to microorganisms our lives are positively glacial. At the other end of the spectrum, our lives are indeed frenetic compared to the stately procession of bristlecone pines and Antarctic lichen.

Today there is a compelling reason to privilege human time over ant or lichen time: the pace of both our experiences and our needs is determined by our biology. Our bodies – and that includes our brains – function at a speed that is governed by how often we need to breathe, drink, and eat. But will this always be the case?

The same optimistic futurists who foresee 3D molecular printing and mind uploading also foresee the technological enhancement of human cognition. This means, among other things, a greater speed of thought. A few decades hence, our minds may operate much more rapidly than they do today. A nanomachine-enhanced cerebral cortex might operate twice as fast as an unenhanced one, the most obvious consequence of which would be that the subjective passage of time would seem to slow by half. There is a reason why slow motion film and video has such a compelling effect upon us: it gives us more time to process what we are seeing. To a faster brain, the whole world would appear to move in slow motion. How much faster might future brains become? Hundreds of times faster? Thousands? Millions?

To an uploaded conscious mind whose brain-substrate is a computer a billion times faster than the human brain, the subjective passage of time in the real world would all but cease. In the ten minutes it would take to experience a cup of coffee in the real world, this non-biological mind of the future could subjectively acquire over 200 lifetimes of experience in virtual reality. Moreover, the computer-based minds of the future will possess total recall as well as the ability to share the memories of others. Any appreciation of the long term is inherently contingent upon positioning new experiences within the context of previous ones, and it is the various mechanisms of memory – individual, cultural, institutional – that make the ongoing creation of such a context possible. It therefore seems likely that appreciating the long term will become a more feasible prospect when our memories are both flawless and massively shared.

If current exponential trends continue as expected, computers of such power are forecast to arrive – once again – in 50 years or so. In the face of such developments, how likely is it that human time as we know it today will retain its primacy for the indefinite future?

Rethinking the Long Term

It is certainly not my intention to discourage either long-term thinking or environmental conservation. Quite the opposite: I think these things are both crucially important not only for human wellbeing but for the wellbeing of all the other forms of life with which we share our precious planet. It is precisely because I care so deeply about these issues that I think we should insist upon a view of the future that is as honest and realistic as possible. And given what we now know about the technologies that lie in the decades and centuries ahead, it is dishonest and irresponsible to assume that people in AD 12,013 will be biologically no different than the people of 7,987 BC.

In the 21st Century, we cannot responsibly measure the long term in centuries and millennia, nor can our metric be constant. We must first measure the long term in no more than decades, and then in years, months, weeks – and eventually days and perhaps even minutes.

Furthermore, in the 21st Century we cannot continue to argue that environmental conservation is important primarily because it is necessary for our survival, since that may not hold true for long. Instead, we must insist upon genuine ecological justice, and upon the rights of other living things independent of the utility they represent to us.

Given what we now know about the exponential progression of technology and the truly fantastic developments that lie not millennia but mere decades ahead, I believe we have a very real responsibility not only to think seriously about but to actively prepare for these developments today. We can no longer in good conscience defer that responsibility to future generations.


  1. “Furthermore, in the 21st Century we cannot continue to argue that environmental conservation is important primarily because it is necessary for our survival, since that may not hold true for long. Instead, we must insist upon genuine ecological justice, and upon the rights of other living things independent of the utility they represent to us.”

    I remember thinking along these lines after listening to one of the Long Now podcasts. It is probably inevitable that more and more ecosystems will fall under human micro-control, whether because they collapse and need to be restored or because they provide some valuable eco-services whose failure we will not accept… in any case, the constitution and organisation of many ecosystems will come to reflect human _aesthetics_. e.g. what counts as an invasive species? What counts as “original” forest in North America? I find the term “justice” a bit problematic, since what justice is there in deciding that a given plant, expanding its range due to climate change, is a pest, while the pest that settled 100,000 years ago is OK?

    It seems to me that the decision to go with laissez-faire or “maximising ecosystem fitness” or beauty or some other metric is an aesthetic one, broadly interpreted.

    – You write about the possibility of “restoring” extinct animals by saving their DNA and reconstructing them later on. Obviously we agree that it would be better to avoid this in the first place… but it also seems like a fascinating technical problem.

    On the one hand the objection that this is technically difficult seems to me correct but short-sighted; I don’t know of any fundamental obstructions to, say, restoring wooly mammoths using elephants (did they try that yet?). And the future is long. But restoring functioning ecosystems seems an order of magnitude harder…

    One problem I see is that of serialisation ( (and on a more general point, one of the main issues I have with Kurzweil is that just because you can reduce all problems to information theoretic ones, doesn’t mean everything is easy… what if your meatspace problem reduces to an NP complete information problem? tough).

    I don’t know if you program, but it’s common to take structures in computer memory and “serialise” them to a format that can be stored to disk, transferred over the wire, etc. But imagine you have a data structure X, some part of which is numbers or strings that are easily serialised, but which also contains references, i.e. addresses in memory of _another_ data structure Y. To serialise X you would have to first serialise Y, which perhaps references Z, and so on…

    OK fine, so you could just pause the system, enumerate all these links, and serialise everything all at once…

    Except maybe there is not enough storage for you to serialise everything at once, and in any case it’s probably impossible to pause the system.

    I think you see the analogy with ecosystems so I won’t belabour it, but the point I’m trying to make: while it might be feasible to restore an individual species, it might still be impossible to boot up an entire ecosystem from a backup.

    – Regarding Assumption 1. I’m not so sure that the technologies you describe necessarily imply an end to resource scarcity. If you believe that, I feel it’s incumbent on you to explain to, say, Thomas More why the technologies of today haven’t put an end to conflict over resources. I can easily imagine a society in which the only constrained quantity (and hence the only quantity worth giving the name “resource”) is computation cycles. But while the available computational power has increased exponentially in the last few decades, I would guess that the percentage of that computational power _in use_ has remained close to fixed.

    Why wouldn’t a society with immersive virtual reality and computation cycles as currency be plagued by the same inequalities we see today? The rich get higher fidelity. And if people can “upload” why wouldn’t the same principles that lead to genetic evolution necessarily imply brutal competition for computational resources to generate copies or descendants?

    Probably the need to organise large groups of people for agricultural and manufacturing societies was important in us reaching a consensus about being reasonably civilised; if technology takes away this predicator couldn’t it be just as easy for society to be violent anarchy or despotism rather than happy socialism? If not, why not?

    To echo what’s in your last paragraph, I would suggest that developing the kind of culture which avoids these “bad” endings (aesthetics again) is one of the kinds of long-term thinking that the Long Now is about (or should be about). That is, the same habits which make someone good at preserving photos of San Francisco or knowledge of amazonian tribe culture, or figuring out how Easter island statues work, or paying attention to long-term questions about ecosystem preservation, are likely to be useful in identifying the traits that make our society work – to the extent that it does – and preserving those during whatever transitions take place in <= 50 years.

    • Thanks for the comment, lots of great points there!

      With respect to ecosystem restoration, “aesthetics” is exactly the right word to use, since there can be no value-free determinations of which ecosystems should be restored. As for ecological justice, that is a formal term from the justice literature that denotes the special category into which “non-rational” actors fall. More specifically, ecological justice theory cannot be part of contractarian theories of justice – since nonhuman life cannot enter into a contract with humans! Since contractarian theories of justice (which began with Hobbes and are perhaps most fully realized in the work of John Rawls) cannot apply to the rest of the biosphere, we need to think differently about what rights, entitlements, privileges, obligations, equity, and equality mean for nonhuman life. That’s where ecological justice theory comes in.

      Regarding the capture of large and massively complex systems, I agree this is likely to be a formidable challenge even once computational capacity reaches astronomic levels. I think the response from folks like Ray Kurzweil is that what we are really after are only the “salient” features of the system. Kurzweil assumes, for example, that we do not need to capture the structure of the brain down to the atomic level in order to acquire all of the structure that gives rise to conscious minds. He is betting that quite a lot of the fine-grained structure of the brain simply serves as passive substrate – as scaffolding upon which the important features are built. If he is correct, then the amount of data that represents all of the salient features of a human mind will be many orders of magnitude smaller (and more manageable) than the total amount of data that would be required to accurately represent every atom in the brain. He is also betting that biology is, in his words, a “suboptimal substrate”, meaning that other computational substrates that can be designed from the ground up without having to be contingent upon the constraints of 3.5 billion years of prior evolution will be far more efficient.

      There are folks who disagree with Kurzweil on this front. A number of quite respectable scientists, for example, have written about the possibility that quantum effects are crucial to how consciousness arises in the human brain, and that because of this the “salient” features of the brain go right down to the atomic level. So far, though, my impression is that the evidence for quantum effects in the brain is limited and unpersuasive. The example of general anesthesia would seem to discredit the claim that a special quantum state is somehow required to give rise to our conscious minds: general anesthetic (and to a lesser extent, ordinary sleep) seems to deactivate large parts of the brain, and yet the brain is able to reboot our consciousness quite successfully after these perturbations. This example also gets to your point about the challenges of “pausing” a complex system. There is no question that such pauses present enormous practical challenges. Can an entire ecosystem reboot the way a brain can after general anesthetic? It’s difficult to say. And even with a rebooted mind, there is a very real sense in which I, for example, am not quite the same person after general anesthetic (or a good night’s sleep) as I was before. Certainly the microstructure of my brain is markedly changed. Kurzweil has suggested a sort of personalized Turing Test should be applied here: if I can convince all other relevant parties that I am functionally the same person when I regain consciousness, including myself, then that should in principle be satisfactory.

      More technically, I believe it is Max Tegmark who has criticized the notion of quantum effects in the brain, saying that quantum states cannot remain coherent at the spatial and temporal scales necessary for neurochemistry. Unfortunately I am not qualified to evaluate that particular criticism.

      But there may be other barriers to the Kurzweilian vision of uploading minds and ongoing exponential gains in computational capacity through until late in this century, such as the issue of computational irreducibility (which is connect to the NP complete information problem, if I am not mistaken), which Stephen Wolfram has suggested may represent a major obstacle. Again, I am not qualified to evaluate this criticism, but Wolfram’s is certainly an opinion to take seriously.

      Regarding competition for computation cycles, I think this a very astute observation with fascinating implications – it seems quite a plausible endgame for the evolution of agent-driven ecosystems. One issue that comes to mind here, however, is the assumption that individuated and uncoordinated agents will still be “free” to compete in such a system. They may well be, but history – though not a perfect guide for the future – shows us that cooperation plays as large a role in the evolution of complex systems as competition does.

      Throughout the evolution of life on Earth, coordination and cooperation have proven to be fantastically effective. The degree of coordination and cooperation that occurs at the molecular, cellular, and subsystem level within any complex organism – from a dust mite to a human being – is absolutely staggering. And this is for mindless, motiveless, unintelligent units. I imagine that because of the advantages on offer, there will be enormous pressure for individuated intelligent actors to cooperate and coordinate their activity. So much so, perhaps, that a collective, singular consciousness might prove a more likely far-future outcome than myriad individual intelligent actors competing for resources.

      Of course we have to take all of this conjecture with a very large grain of salt! One of the key points of the Singularity concept is that these eventualities are impossible to predict because we know too little about what superhuman intelligences will be inclined to do. Still, these are fun ideas to speculate about!

  2. What’s a good place to start reading about ecological justice?

    It’s interesting that you picked up on the “salient features” angle, because this is the context in which serialisation first crossed my mind…

    But I think this personalised Turing test is a terrible idea. The original Turing test makes sense, since it represents a sensible bar that a _technology_ has to surpass. But imagine Beethoven was digitised when he was 30, and his digital copy was able to convince everyone it was completely accurate… so to save on heating costs or something the biological Beethoven was put down.

    But the “interaction surface” that the copy needed to passably simulate is so tiny compared to Beethoven’s inner life, that he never goes on to write his 9th symphony. Oops. What if we all upload and things seem more or less just as vibrant as before, but we notice 200 years later that there were no more Beethovens? Or Feynmans?

    I think you raise an excellent point regarding co-operation and competition throughout the history of life. I’ve been rolling this around in my head now for a week, and I want to make a couple of points.

    The first one is pretty obvious, but I think to aid in clear thinking it’s worth setting down.

    Try to imagine what Darwin would say if we went back and explained superhuman thinking machines. I think he might be fascinated, but observe that it may not be a fundamentally different form of life, since the principles that he identified in natural selection for carbon-based biology, i.e. resource limits, struggle for survival, are in fact very general and apply independently of the substrate life is operating on. Of course there are exceptions, but I think the basic principles are robust enough that one can, for example, sensibly think through examples like his discussion of parasites and sexual vs. asexual reproduction, and have them make sense for computation based life.

    On this basis Darwin might predict that this new kind of life would show similar behaviour to the ordinary kind, just at a different scale.

    Let me now get back to your point about co-operation and competition. I think the question of scale is the relevant one. Individuated and uncoordinated agents at one level of organisation might not be free to compete in such a system, but the principle that I just put in Darwin’s mouth suggests that there will always be a level of organisation (e.g. superagents) at which violent competition can exist.

    So let me finish by echoing your comments, but in the form of a question: even if violent competition _can_ take place, will it? This seems like a bit of a Rorschach test, designed to reveal one’s personal reading of history and geopolitics, and I’m certainly not going to try to answer it…

Leave a Reply