Teleology & Simulation

(—This essay is a draft, undergoing feedback and revision.—)

In 1802, William Paley famously set out the "Watchmaker Analogy", a teleological argument, or argument from design for the existence of God.

The argument was as simple as it was intuitively compelling. Suppose you are walking alongside a river, and suddenly, you look down and see a pocket watch. In that moment, you know without a shadow of a doubt that this watch was deliberately made. There was a Watchmaker who created it.

This intuitive leap is bolstered by the fact that you can look at a thousand other objects lying on the banks of the river, and not come away with a similar impression. Design is something we recognize intuitively, and in recognizing it, also recognize that there had to be a designer.

This example seems relatively uncontroversial. But of course Paley was trying to make a bigger point: When we look at the natural world, with creatures who see, swim, fly, and dig, we intuitively recognize the presence of design. We understand that these things show the existence of purpose. Teeth are made for biting, wings are made for flying, and something must have intended that it be so.

Thus, there must be a Watchmaker for the natural world as well.

This argument has come up against strong criticism. Paley was writing in 1802, but by 1859, Darwin had published his theory on the origin of species, forever changing the nature of the discussion. Darwin seemed to provide an answer to Paley, and in fact, Richard Dawkins, in his book "The Blind Watchmaker", famously claimed that this is exactly what had happened. Natural Selection had shown that non-purposive processes can lead to seemingly purposive results.

The argument seems to have reached a stalemate. It's still intuitively compelling to many people, and yet, it is largely unconvincing to anyone educated in the natural sciences.

Let's see if there's a way to move forward.

What the Argument from Design actually means

Why is it that we recognize a pocket watch as designed? What are we actually recognizing when we do so?

A number of people have tried to spell out an answer in terms of complexity. But complexity is notoriously hard to quantify, and this path has produced few, if any, real results. I want to suggest a different approach.

I want to suggest we're actually identifying two things:

This object is improbable
This object is useful to a being like myself

How do we know this object is improbable? Almost certainly, it's because we've tried to create similar things before. We understand just how difficult it is to shape metal in this way, to wind springs, to construct gears. It's no coincidence Paley formulated his analogy in the context of a society that was very good at making pocket watches. His readers intuitively understood what kinds of difficulties were involved.

So with the criteria of usefulness. How do we know this object is useful? It's almost certainly because we've used something like it ourselves. We understand the utility of watches in an industrial society, we understand the value of making them portable, we understand the value of using compact symbols and dials to represent the passage of time.

Our Bronze Age ancestors probably would be able to recognize something similar. Shaped metal, glass, writing, numbers—all point to something they would have understood in terms of usefulness-to-humans-like-them. They probably would not immediately recognize the function of the device, but they would be able to understand it as the kind of things humans do.

Our hunter-gatherer ancestors? This is harder to picture. I guess that they would recognize some aspects of human design, and some aspects of something bizarre and unknown. In so doing, they would recognize the presence of something like themselves—and something very unlike themselves.

The point is, this intuition is not something intrinsic to certain kinds of objects in the natural world. It's an intuition that comes from within us, and our role as creators of objects.

But why would the recognition of usefulness, indicate that beings-like-ourselves were involved? "A million monkeys pounding on typewriters will eventually produce the works of Shakespeare", so the saying goes, so how do we know that some random series of events didn't just throw up a pocket watch onto the shore?

Here, we've come to a probabilistic argument.

Presumably, a series of random collisions in a river, happening over trillions of years, would throw up a pocket watch onto the shore. But you'd be waiting an awfully long time.

On the other hand, if there is a society of beings-like-us around, and they are inventing and manufacturing things that they find useful, it's likely that trillions of such objects are going to be spread throughout the world.

To put it more precisely: Intelligent beings will construct things that are difficult to make, but useful to them, at a rate unimaginably higher than natural processes alone.

So if you're walking along one day, and discover an object that seems difficult to make, and would be useful for a being-like-yourself—then chances are trillions to one that you've discovered something made by a being-like-yourself.

This is what our intuition is identifying. The usefulness of this object to beings-like-ourselves indicates something about the choices beings-like-ourselves will likely make. This in turn indicates what they will turn their intelligence to. And where intelligence is, natural processes become statistically irrelevant.

This is actually a good short-hand definition for "intelligence": Intelligence dramatically accelerates the creation of useful things.

So let's call this the Expanded Argument from Design:

Expanded Argument from Design

Our own experience (namely, what is difficult to create, and what is useful to create) gives us insight into both the natural world, and the psychology of beings-like-ourselves.
In identifying an object as Difficult to Create, we are suggesting that the natural rate of occurrence of this object in nature is low.
In identifying an object as Useful to Create, we are suggesting that the rate of creation of this object among beings-like-ourself is high.
Thus, in identifying an object as something we would find Difficult to Create and Useful to Create, we are identifying that object as something produced at an unimaginably higher rate by beings-like-ourselves, than by natural processes alone.
Thus, in identifying an object as something we would find Difficult to Create and Useful to Create, we are identifying the likely action of beings-like-ourselves.

Note that at no point are we suggesting the impossibility of the natural occurrence of apparent design. Our argument here is simply about rates of occurrence. If we wanted to bring this into engagement with the phenomenon of natural selection, for example, we would want to compare the expected rate of what natural selection is producing, vs the expected rate of what intelligent beings might produce.

The Simulation Argument

Having fleshed out the intuition that the Argument from Design is actually trading in, and having attempted to offer an expanded version of this argument, I want to now turn to something much more modern: The Simulation Argument.

For those who are unfamiliar, the Simulation Argument is based on the likelihood that, given somewhere between a few decades to a few more millennia, human civilization will start creating simulated (virtual, holographic, etc) worlds that are indistinguishable from the real thing. And, based on the way we already deal with simulated worlds (video games, virtual realities, etc), they will likely run trillions of these simulated worlds over time.

Presuming that many of these simulated worlds are like the one we live in now (after all, our descendants might want to visit our time in their equivalent of "Oregon Trail 3D"), then we reach a stunning conclusion:

It is likely that there are vastly more simulated worlds like our own, than there are "natural" worlds like our own.

And that leads us to the Simulation Argument Trilemma:

Almost all civilizations go extinct before simulating worlds like our own.
Almost all civilizations are uninterested in simulating worlds like our own.
You are almost certainly living in a simulation.

I have dubbed this trilemma "Suicide, Censorship, or Simulation". The first two are possible reasons for why no civilizations ever do this. The third is the consequence, assuming they do.

Assuming that some civilizations simulate worlds-like-our-own, then it is extremely likely that any given world-like-our-own exists inside a simulation.

Note that, contrary to many popular presentations, this is the hopeful option.

The first option is undesirable for obvious reasons. The second option is hard to imagine without picturing a very dire scenario—perhaps all simulations are forbidden by a totalitarian regime, or maybe worse, perhaps our descendants have no interest in us whatsoever.

The third option, however, is scariest for most people. What would it mean if we were living inside a simulation?

That's another discussion. For now, let's look at the underlying nature of this argument.

First, we observe that beings like ourselves have gone to great difficulties to simulate worlds, and that they show every indication of continuing to do this at an accelerating rate in the future.
In identifying worlds-like-our-own as Difficult to Create, we are suggesting that the natural rate of occurrence of worlds-like-our-own is quite low.
In identifying worlds-like-our-own as Useful to Create, we are suggesting that the rate of simulation of worlds-like-our-own by beings-like-ourself is quite high.
Thus, in identifying worlds-like-our-own as Difficult to Create and Useful to Create, we are suggesting that worlds-like-our-own are produced at an unimaginably higher rate by beings-like-ourselves than by natural processes alone.
Thus, in identifying worlds-like-our-own as Difficult to Create and Useful to Create, we are identifying our world as the likely product of beings-like-ourselves.

This is the same basic argument as the Expanded Argument from Design. In both cases, we are extrapolating from our own psychology to the psychology of large numbers of beings like us. In both cases, we are looking at their likely behaviors. And in both cases, we are concluding from the simple probability of production numbers that we are likely witnessing the action of beings like ourselves.

The Simulation Argument Trilemma takes this as a starting point, and then offers two ways that our psychology might not reflect the psychology of future beings. First, those future beings might not exist. Second, they may have lost all interest in what interests us.

We can put this in the context of the Watchmaker Analogy. If we find a pocket watch, we assume it is the product of beings like ourselves.

But if we found out that the inhabitants of this planet had never developed the technology for shaping metal and glass, then we would be forced to conclude this apparent design was an illusion of some sort. Similarly, if we found out that the inhabitants of this planet had no interest in time whatsoever, then we would be forced to conclude that some other phenomenon was going on.

If both cases are hard to imagine, it's because it is hard to imagine a psychology so different from ours. We know that in actual history, humans have produced millions and millions of pocket watches, and it's almost impossible to think of it differently.

The Simulation Argument, on the other hand, is not yet as intuitive, because our simulations have not yet reached the precision and quality of the world we actually live in.

But if we keep creating as we have been, then soon, this intuition will change.

The Hole in the Argument from Design

There's a crucial difference between the Argument from Design, and the Simulation Argument.

In the Watchmaker Analogy, we're reasoning about the behavior of beings-like-us, in relationship to the production of pocket watches. We then attempt to turn our "pocket watch reasoning" onto the natural world, and reason about the behavior of beings-like-us in relation to it.

But this introduces a problem. We are unable to create a natural world. We cannot design wings or bones or teeth or claws from scratch. We thus find ourselves in the place of our hunter-gatherer ancestors: we can recognize something familiar here, but we cannot fully identify it.

We are trying to extrapolate from our own psychology—which can produce pocket watches—to the kind of superhuman psychology which can produce worlds.

Most devastatingly, we have not recognized that this is what we are doing. We assumed that the recognition of design was straight-forward, when it was actually an extrapolation of our own behavior and psychology.

This means that there's a hole in the standard Argument from Design. We don't intuitively understand the purposes of a superhuman psychology that can create a world. We thus don't know what to look for, to identify its presence.

This is the hole that Darwin drove a train through.

Darwin's picture did not look like what someone with our kind of psychology would come up with, and so it appeared that the argument from design had been overthrown.

The approach modeled in the Simulation Argument patches this hole. In the Simulation Argument, we explicitly identify the psychology of the beings we are trying to model, tease out the ways that psychology may differ from our own, and then extrapolate from their expected behavior.

This moves us from intuitive recognition to deliberative reasoning. Instead of asking whether we intuitively recognize the presence of design, we ask what purposes might drive beings who are greater than us.

Perhaps we conclude that no purposes of such beings would lead to a world like ours. Thus, we recognize no such design.

Or perhaps we recognize that there are purposes of superhuman beings that intersect deeply with the world we find ourselves in. And thus, we identify and confirm the design woven deeply and pervasively throughout our world.

An Example

I just want to give one example of the way this kind of reasoning might work, to take this from the abstract to something slightly concrete.

All life seems to be built on reproduction. Cells produce more cells, and when cells join together into multicellular organisms, those organisms produce new versions of themselves as well.

Interestingly, as life scales up, life seems to need to lean harder and harder into diversification. Single-celled-organisms can mutate, and so introduce some diversity, but sexual reproduction allows larger life-forms to embrace dramatic genetic changes from one generation to the next.

Arguably, humans have taken this to another level. Where horses and other creatures are born with almost all the skills they need to survive, human babies take a long time to develop those same skills—seeing, walking, feeding themselves, etc.

The fact that human babies are born in a more infantile state than other creatures, means that our primary abilities are all learned. And this means that we experience more dramatic changes from one generation to another than any other creature.

This is the secret of our success. We adapt, we invent, and we introduce dramatically new things all the time. We are born deeply dependent, and this is paradoxically how we become deeply independent.

This suggests a deep necessity at work. The more complex the organism, the more important it is to introduce diversification into its procreative process.

We must assume that any future civilization faces the same concerns. We can imagine a society spreading through space, seeding new worlds—and we must then assume that they are concerned that they do not simply clone an old world, but diversify, encouraging new kinds of life, and new directions in civilizational growth.

How might they do so?

One risky move—perhaps as risky as having offspring who take years to develop even the most basic survival skills—would be to drop human beings on a planet, and allow them to recapitulate the process of technological development.

This would ensure that an old world hadn't simply been cloned, but that the rise of this new world would actually contribute something new and useful to the interstellar civilization. When they developed pervasive space-flight, then we might welcome them into our interstellar civilization.

We could imagine doing this in a few ways.

One way, we give them all of our technological knowledge, and allow them to study it and rebuild their world from scratch over roughly 50-100 years. This minimizes risk, but it maximizes sameness. They don't end up making any real discoveries themselves, and so they don't learn the kind of independence that will make them an asset to an interstellar civilization.

Another way would be to drop them on a planet, with a few basic starter ideas, and let them recapitulate the process of developing from hunter-gatherer society to global civilization. This is much more risky, and might take 10,000 years, but when they join interstellar civilization, they may have a lot more to offer.

We might imagine this process scaling up, as unimaginably powerful societies seed new planets, new star systems, new galaxies, new universes.

At each level, the society is faced with the question of how to grow new worlds, while introducing new kinds of diversification. The greater the scale, the more pressing the question becomes.

At the highest levels we can conceive, we might imagine a society seeding a new universe, watching for it to recapitulate the process of galactic, stellar, biological, intellectual, and technological evolution—eagerly awaiting what new things it will bring about.

Note that we actually have two concerns in these scenarios. We want "child societies" to grow and become independent, but we also want to make sure this isn't abortive. We don't want to invest 10,000 years growing a new planetary civilization, if it's never going to get off the ground. So we have to figure out the sort of "minimum viable starter kit" for each level.

The same dilemma is played out in human biology. It's important that we learn almost everything from scratch. But it's also important we get the tools we need to learn. So both our biology and our culture tries to balance this out. Much of children's initial learning is truly independent, but we try to "seed" their environment in a way that is maximally conducive to what we expect them to need.

So if some vast civilization is out there seeding universes, what might we expect to see?

We might expect to see a universe that seems uniquely fine-tuned for producing life.

We might expect to see a universe that seems uniquely comprehensible to intelligent beings. This might require it to be relatively "smooth", so you can get a similar picture no matter where you are within it. This might require it to be relatively safe (few vast, fast-moving black holes). This might require it have interesting patterns to study at every scale.

We might expect to see a universe where life emerges very early in cosmic history, but has to develop from a single cell all the way to intelligent beings. We might expect to see each significant transition (non-existence to existence, non-life to life, non-intelligence to intelligence) happen almost exactly once.

And we might look at our universe, and notice that these are some of the most unusual and interesting facts about it.

Of course, what we make of this is wholly contingent on what we think superhuman psychology would produce, versus what we think we would expect to find without it.

And that's an area for a lot more discussion.

Conclusion

Paley's reason for introducing the Watchmaker Analogy was to create a case for design from the observation of the natural world. He intuitively recognized that the natural world shows indications of being improbable, and yet meaningful to beings like us. We might assume then, that a being like us had a hand in it.

But the Simulation Argument shows us that this analysis is too limited. Yes, this kind of analysis works with pocket watches, but that doesn't provide us with a linear path to anticipating the behavior of beings who have the power to create worlds.

Thus, in order to really grapple with the situation, we need to think about what those kinds of beings will want to do. Fortunately, our own future gives us a handle on some of those questions. We can tentatively approach them by asking what we will want to do in the future.

In order to determine whether the world we live in is the result of some form of design, we need to ask what we would do once we had the power to create a world. This may then allow us to identify the presence of design, and simultaneously challenge and improve our understanding of the psychology of beings who could create a world like our own.

Thanks to Lincoln Cannon, Giulio Prisco, and Jonathan Gunnell for offering early comments on this essay.