Like children opening their eyes to a new day, having gone to bed the previous night as tufts of snow began falling, we dash to the window and lift ourselves to the tips of our toes to behold a landscape transformed: a winter wonderland glittering with possibilities for discovery and play. Even the tree branches, before so boringly bare, have been changed into something beautiful and magical. We feel we are inhabiting a storybook or a gameworld, and we want very much to put on our boots and mittens immediately and run outside to see it, touch it, experience it, and to play, play, play …
Nick Bostrom, Deep utopia
1. Introduction
Nick Bostrom’s Deep utopia is, by all accounts, a strange book. After its opening words about playful children encountering a winter wonderland, the book is structured around a series of six hypothetical lectures delivered by Bostrom himself on six successive days.
The prose is narrative, though often reminiscent of the dense style of Bostrom’s Superintelligence. The discussion is wide-ranging, though perhaps not as loosely organized as it may appear.
Like many readers, I am not quite sure what to make of the book. This series is an attempt to find out. My aim is to cover one lecture per post, beginning with the first (Monday’s lecture) today. The lectures are quite long, so that pace may not be sustainable, but given the looser organization of this book, it seems best to respect the lecture structure that Bostrom has imposed.
Because the book is at times difficult to understand, I will begin with my own attempt to summarize key themes from each chapter before discussing them.
2. Narrative setting
Monday finds Bostrom delivering a lecture in the Enron auditorium. Three students, Tessius, Firafix, and Kelvin enter midway through the lecture.
The announced theme of the lecture series is “the problem of utopia: the problem we will face after we have solved all the other problems.” In such a solved world, Bostrom asks: what would give us meaning and purpose? And what would we do with our time?
From there, the lecture is loosely structured around two topics: work and income, and population dynamics.
3. Work and income
Keynes predicted that the past century would see a 4-8x increase in productivity, accompanied by a rapid decrease in average working hours. Keynes was right to expect rapid productivity growth, but his predicted fifteen-hour workweek did not come to pass. For the most part, increased productivity led not to less work but instead to more production and consumption.
In principle, Bostrom notes, this trend could continue indefinitely. Even in much more productive economies, people may work to obtain new and valuable types of consumption goods; to pursue resource-intensive social projects such as promoting wild animal welfare and creating digital minds; and to secure positional goods in the form of a stronger relative economic position compared to their peers. Even if nearly all sectors of the economy were to be cheaply automated, it might remain possible for humans to demand large wages in incompletely automated sectors of the economy.
What would happen if all sectors of the economy could be cheaply automated? Humans might lose our ability to earn significant income through labor. However, we need not lose our ability to earn income through other means.
Nevertheless, Bostrom considers a three-factor model of the economy where output is driven by labor, capital and land. Here, the category of “land” covers a broad range of non-labor, non-capital inputs including not only physical land, but also resources such as minerals and fuels which cannot be easily replicated. Here, Bostrom notes, even if capital and labor lost their scarcity, humans could continue to earn income through control over land.
However, there is a catch. Because land inputs cannot be easily replicated, there are only so many of them to go around. This means that the ability to earn income through this land depends on the size of the human population. This leads to the second main topic of Monday’s lecture.
4. Population
Bostrom’s discussion of human population is structured around a background Malthusian dynamic. Under Malthusian dynamics, resource scarcity is the dominant constraint on population size, so that human populations continue reproducing until, in equilibrium, no more resources are available and, at a first pass, average living standards approach subsistence level.
After noting some complications with the equilibrium interpretation of the Malthusian model, Bostrom asks under this model what becomes of typical intuitions about the kinds of policies that would promote welfare, such as improved technology or social stability. Again, Bostrom notes, the answer is complicated.
In the short term, these policies promote welfare by making those alive better off. In the medium term, Malthusian dynamics may react to shrink at least average welfare back where it began. In the long term, they may enable the kinds of technological and social progress that took humanity out of the Malthusian trap. And in the even longer term? Bostrom does not know.
5. Feodor the Fox
After the lecture, the students glance at the evening’s assigned reading. The reading is a parable about Feodor the Fox. Puzzled by many philosophical questions, including questions about the best way to improve the world, Feodor the Fox seeks out advisors.
After many candidate advisors prove unavailable or uninterested, Feddor encounters Pignolius the Pig. Pignolius is generally skeptical that much can be done to improve the world. Readers are told that Pignolius’ skepticism will be relevant to later lectures.
6. Some points of concern
Many of Bostrom’s readers have expressed concern about the quasi-academic style of his prose, seemingly obsessed with chasing down each trivial detail for many pages. Scott Alexander complains that the prose at times involves:
Typical analytic philosophy “let’s spend five pages analyzing the subtle differences between ‘meaning’ and ‘purpose’”.
and the best-known rationalist review of Deep utopia is concerned that the book:
Begins as if addressing a broad audience, then drifts into philosophy that seems obscure, leading me to wonder if it’s intended as a parody of aimless academic philosophy.
Both concerns contain a grain of truth, in the sense that some detours may not be earning their keep. But in many places, I have the opposite impression: crucial moves are being made quite quickly in a way that would benefit from significantly more discussion. Here are two examples.
6.1. Beyond the three-factor model
All models make assumptions. But the three-factor economic model makes some very strong assumptions. In particular, it assumes no technological progress and no increase in land.
On the face of it, both assumptions are surprising. As for technological progress, the purpose of Bostrom’s book is to consider a society that has made incredible amounts of technological progress, so one might naturally expect them to make more. And as for land, those of Bostrom’s ilk tend to expect large amounts of growth in physical land (from space colonization) as well as other “land” resources that come along with it.
Bostrom’s reply to these concerns is rather terse. He writes:
The assumptions that there is no technological progress and no increase in land are, I think, less rickety than might initially appear. I expect that the rate of economically relevant technological progress will eventually asymptote to zero (once most useful inventions have already been made). Land growth (from space colonization) will asymptote to a polynomial rate, since the volume of the sphere reachable from Earth by a given time is bounded by the speed of light. In the very long run, land growth will asymptote to zero, since the expansion of space means that sufficiently remote galaxies are forever unreachable from our starting point. But even during the long period in which a polynomial rate of land growth could be sustained, a decline of average income to subsistence can easily occur, since a population is able to grow at an exponential rate.
Let’s begin with Bostrom’s remarks about economically relevant technological progress. Bostrom claims, without argument, that humanity will eventually reach a point where most useful inventions have already been made, at which point technological progress will asymptote to zero. This is a very strong claim on any timescale: there is no clear reason why there need be any endpoint to the space of useful inventions, rather than a continual increase in humanity’s ability to understand and make use of the world around us. Nor is there any clear reason why humanity need be expected to reach this point in any appreciable amount of time.
Here it is especially important to dwell on a point that will recur throughout this series: many of the argumentative moves that Bostrom is making are most plausible, to the extent that they are plausible at all, when the envisioned deep utopia lies very, very far in the human future. Even if we grant that there may be some point in human history (say, a million or a billion years from now) at which most useful technological inventions have been made, the importance of thinking about the meaning of life at this point will be radically reduced the further back in history that the point is pushed.
Turn next to land growth. Here, Bostrom predicts a polynomial rate of land growth and suggests the plausibility of an exponential rate of population growth. Putting the assumptions together leads to falling per-capita shares of land, since exponential growth eventually outpaces polynomial growth.
The motivation for polynomial land growth is the idea that humanity will settle Earth in an ever-expanding sphere, so that land growth will be cubic in the radius of that sphere. The motivation for exponential population growth is an underlying Malthusian population dynamics. As we saw in Part 4 of my series on Mistakes in the Moral Mathematics of Existential Risk, three challenges should be raised here.
First (benefitting Bostrom), cubic land growth is enormously ambitious, reflecting the assumption that humanity will have the technological means and desire to settle space in all directions at a significant rate. This is not merely the assumption that we might, for example, expand in one dimension to the east and the west, but rather in three directions, pushing outwards from the surface of a sphere.
Second (contra Bostrom), Malthusian population dynamics are extremely out of fashion. No demographers believe that humanity is currently experiencing anything like Malthusian population dynamics, and all signs point to further departures from Malthusian dynamics in the future. Even if, against this trend, Malthusian dynamics play a larger role in future population change, it is quite a large step to imagine that humanity would once again begin reproducing as quickly as we were able, limited only by resources and do so for a very long time.
Finally (contra Bostrom), the contrast between polynomial land growth and exponential population growth exhibits a curious decoupling between population growth and land growth, two factors that are usually thought to be intimately linked. One of the primary drivers of territorial expansion is typically thought to be population growth, as growing populations seek room to expand. If this is right, then models of space settlement should relate the rate of land growth to the size of the existing population, rather than the size of humanity’s spherical frontier. Many natural ways of doing this will have land and population growing at comparable rates, restoring compatibility.
6.2. Even longer-term impacts
Bostrom claims that on a very long timescale, we have little idea what developments today would make the world a better place.
Little is known about these matters. We are still remarkably in the dark about the basic macrostrategic directionality of things. Truly, I wonder whether we can even tell up from down.
I have some sympathy for these concerns. For example, my paper and blog series “The scope of longtermism” expresses concern about the potential for long-term forecasting, a concern also expressed by forecasting experts deeply familiar with the effective altruism movement, such as Eva Vivalt.
But what exactly is Bostrom’s argument here? Here it is:
I think you can make a case that wisdom and wide-scope cooperativeness are the two qualities currently most needful to secure a great future for our Earth-sprouted civilization. I also think wealth, stability, security and peace are better for wisdom and global cooperation than are their opposites. And so we should welcome advancements in these directions, not only because they are good for us now, but also because they are good for humanity’s future.
This doesn’t imply that earlier progress in these directions would have been good for humanity’s future. Perhaps if my species had lingered longer in the “poor, nasty, brutish” conditions in which my forebears evolved into humans, before matriculating into the Industrial era, we would have evolved, genetically or culturally, to become “more human” than we now actually are? Perhaps we came out of the kiln a little too soon? Maybe we would have been better conditioned for the final vault into the machine intelligence era if we had spent another few hundred thousand years throwing spears and telling tales around campfires?
In the first paragraph, Bostrom expresses the view that wisdom and wide-scoped cooperativeness are the two most important qualities for a good future. He suggests that these are so important that goods such as wealth, stability, security and peace should be valued, if at all, not as direct causes of a better future, but as indirect causes through improving wisdom and wide-scoped cooperativeness.
This view comes out of nowhere, with nothing in the way of argument. One can, of course, plausibly rattle off a few reasons why wisdom and cooperativeness might be helpful for the future. But one can also do the same for many other virtues. And for that matter, we could do the same for many states, such as stability, security and peace. Perhaps we could do the same for institutions, such as democracy and systems of international law, or for forces such as technological progress (or lack thereof). Given this, we really would like to see some argument for taking wisdom and cooperativeness to be so important.
The firmness of Bostrom’s opinions about the importance of wisdom and cooperativeness meshes especially badly with the next paragraph, which casually tosses out any number of questions that might be raised about whether it would be better or worse to achieve wisdom and cooperativeness earlier rather than later. Bostrom might very well have fired off a similar paragraph asking pointed questions about whether it would be better to achieve wisdom and cooperativeness at all. And without any of those questions being pressed or evidenced in any detail, it is hard to see why one set of questions is taken as a strong argument and the second is never pressed.
There may be ample room for skepticism about predicting the future. There may even be ample room for confidence about the virtues needed to navigate the future. But it is very difficult to hold these two views together without turning the same causes for skepticism about predicting the future against purported grounds for confidence about the virtues needed to navigate the future. Certainly, if there are ways to combine these views, Bostrom owes us much more of a story about how he wants to do so.
7. Conclusion
Today’s post introduced Bostrom’s book, Deep utopia, and considered the first (Monday’s) lecture. This lecture explored two inter-linked sets of questions: whether humans might work and earn incomes in the future; and what shape future human population dynamics will take.
In this lecture, I have argued, Bostrom has been twice unfairly maligned: once to his benefit, and once, I think, to his detriment.
The good news is that accusations of poor organization and thematic unity may have been exaggerated. There is a relatively clear through-line to Monday’s lecture, and while the organization is not what one would expect of a more typical argumentative work, it really is not that bad.
The bad news is that Bostrom may have been unfairly accused of detail obsession. At some of the crucial points of the argument, there is hardly any argument at all, and most of the argument that is given is lacking. We discussed two instances where this happens in Monday’s lecture: in the defense of three-factor economic models, and in the discussion of very long-term impacts.
This is not, of course, an exhaustive discussion of the themes and questions raised by Monday’s lecture. Given the length and structure of the book, I don’t think it will be possible to discuss any lecture in the detail that I would ordinarily prefer. But hopefully this post can shed a little light on some of the things that are going on, going well, and going less well in Monday’s lecture.
Leave a Reply