TL;DR This is a partial look at broad system of information with many nuances and differing schools of thought from a first principles analysis.

Intelligent Life as a Ziggurat

Just recently came across a discussion of two geopolitical points of view: Peter Zeihan & Francis Fukuyama. Definitely some interesting points and nuanced discussion, though my devil’s advocate criticism is that some of the comparison is more conclusory than descriptive.

I’ve had some thoughts about geopolitics (who hasn’t, amirite?). After all, in the present world, there’s a sense of tragedy of the commons, perhaps some banality of groupthink, and an increasing awareness of corruption or spiteful inefficiency. A lot of these thoughts are shaped by preconceived values and it’s not difficult to introduce the inheritance of values as a negative factor in geopolitical discussion. So perhaps it’s worth analyzing this from first principles.

One thing that Zeihan accurately focuses on is geography, and the first principle is easy enough: when one strips away abstractions like society and goods, ecology is the preexisting architecture. And Fukuyama’s “End of History” plays out similarly, when countries become less ecology-dependent, they tend to have more continuity & less costly evolution of institutions. Broadly speaking, we also see ecology as a fundamental global consequence in the black swan scenario of two nuclear powers exchanging a certain megatonnage of ordinance, or a supervolcanic eruption. In both cases a certain amount of particulates is dispersed into the upper environment, less sunlight yields fewer crops, and therein is a downstream cascade of problems for all the exposed economies.

I’m going to go out on a limb and describe economy as an upper layer of a metaphorical ziggurat where ecology forms the foundation. For simplicity’s sake, let’s treat a ziggurat as a truncated pyramid.

Pyramids happen to be early superstructures because the geometry optimizes for the most stable height. In order to best describe ecology and economy in this metaphor, one could also use the Drake Equation:

We can see that some ziggurat layers and some fractions of astral bodies are less specified, therefore cover more metaphorical “space”. One can also see each layer as its own ziggurat:

What does this mean? For me, the direct conclusion is that there can be elements that broadly overlay a preexisting substrate with little depth, but there can also be specific, nuanced elements that are broadly incompatible but are deeply impactful in exception. Ironically, a lot of geopolitics could be described as a simple misconception of a small pyramid building itself from the upper layer of a ziggurat. Some concepts might just be obelisks instead of pyramids.

Tangibly, where is this going? I alluded to some competing philosophies, like Zeihan and Fukuyama, but they share common ground. Same goes for starkly different ideological schools like religion and secularism, or political schools like capitalism and communism. Both contain people that gotta eat, and for the most part, sleep indoors. Moses & Gilgamesh involve chiseled stone. Adam Smith and Karl Marx both appreciated the printing press and the contemporaneous economic trends that inspired them.

The error in forming this ziggurat is committing to forming a higher structure with the preconception that it is as universal as its foundation. In the aforementioned podcast, Gell-Mann amnesia was briefly mentioned. As Michael Crichton describes it:

Briefly stated, the Gell-Mann Amnesia effect is as follows. You open the newspaper to an article on some subject you know well. In Murray’s case, physics. In mine, show business. You read the article and see the journalist has absolutely no understanding of either the facts or the issues. Often, the article is so wrong it actually presents the story backward—reversing cause and effect. I call these the “wet streets cause rain” stories. Paper’s full of them.

In any case, you read with exasperation or amusement the multiple errors in a story, and then turn the page to national or international affairs, and read as if the rest of the newspaper was somehow more accurate about Palestine than the baloney you just read. You turn the page, and forget what you know.

So, I think that there are probably first principles that establish a set of facts:

• Human activity is not just ecological, but an intelligently moderated system of exchanging different forms of capital. At scale this becomes either industry or the state, in both cases for the necessity of self-regulating increasingly denser human activity.

• Where industry or the state is dependent on raw resources, there is an ecological cost (sustainable or not)

• Insulation from ecological cost is bottom-up; a mercantile government can import scarce resources so long as it can trade unmolested. In other words, a broader hegemonic sphere of influence supports more specific instances of international fungibility; it is more sustained as a universally appreciated & less demanding set of social principles.

• Inventions are hegemonic devices of a universal human quality: using tools to reduce ecological costs. Inventions are also formed from ecological parameters, e.g. crop rotation & agroforestry for maximal yield, aqueducts & modern plumbing for sanitation, electronics from voltage, flight from aerodynamics, rocketry & extraterrestrial travel from combustion & gravity, etc.

A lot of this is perhaps far-reaching extrapolation, but one major conclusion I take from this thought process is that certain kinds of ecological resources, especially intellectual capital, broadly benefit our economic goals. The largest hegemony to date are societies that can support unfettered/evasive access to a globalized digital information network (The Internet). It’s not 100% as broad as all human society in aggregate, but it’s not difficult to see the economic impact of that network effect. A lot more cultural capital is exchanged globally today, rather than 60 years ago. The same goes for academic research, globalized freight, and monetary investment. One can argue that the freedom of information allows for this scale of capital exchange.

Open Source Capital

If there is a center to the ziggurat, it is more extensive distribution methods of information. Money was made possible by ledgers, a temporally broader multiparty record. The introduction of the printing press changed the comprehension of religious doctrine beyond an exclusive lingua franca, which led to cultural events like the Reformation. The telegraph changed the diplomatic landscape and dramatically reduced the fog of war. These technologies do not 100% cover all societal circumstances, but there seems to be a common benefit if and when they’re adopted.

This is especially true for public goods like bodies of knowledge. When one intellect demonstrates novel gain of knowledge, everyone within the scope of their broadcast method has the capacity to learn that knowledge if they already possess the preexisting intellectual capital. Information is, broadly speaking, less scarce than other ecological resources. Certain classes of information can be treated as more exclusionary, like intelligence in the context of espionage, but even that methodology is agnostic to most versions of a societal state.

At the beginning of this year, the most valuable capital is machine learning methodology & the body of knowledge it consumes. The relationship follows other distributions of information; the limiting reagent of procuring a ML model is the energy and equipment it needs for training, but the cost of consumption for the trained model is considerably less scarce in energy & equipment. Besides, the trained model is intellectual capital that can be globally broadcasted in abundance.

Unfortunately, some firms build pyramids & obelisks in the process of distributing information, instead of broadcasting them as wide as possible for maximal economic benefit. With bookmaking, we had abridged editions. With ML models, we have “alignment & safety”. The concern is that while the basic cultural hegemony of academic exploration is being recognized, it is also being conflated with a more specific, societal hegemony one layer up.

Consider the technology of the atomic bomb. When a single nation-state could exclusively deploy it as military capital, they went ahead and deployed it twice in an active campaign. By the time several nation-states developed atomic bombs, the global understanding of mutually assured destruction made actual use nonviable. Fortunately, the risk of M.A.D. is complemented by our dependence on transparently & globally broadcasted information. It is much more difficult to accidentally assume hostile intent now than it was in the Cuban Missile Crisis, simply because battlefield awareness traverses all physical boundaries at the speed of light.

To connect this to the concept of “optimal economics”, we’re probably going to benefit best as increasing improvements in machine learning, i.e. self-informing to inform further, are widely opensourced in their unabridged forms. Innovations in other fields compound with and for this enterprise. It can become less ecologically costly to optimize the “economy of atoms” because a non-negligible fraction of that cost is in the production & consumption of the “economy of bits”. As a corollary to this, many economic gains are limited by the implementation cost in social & cultural capital, and a fraction of those gains are what many might call environmental conservationism. As we are more capable of articulating an unsustainable ecological cost, integrating that into a global body of account (keeping that at a lower layer of the ziggurat), and abstracting away the social & cultural cost (a superficial pyramid/obelisk), the economy (or the entire upper layer of the ziggurat) can be more productive with sustainable ecological cost.

Unified Church-Turing Thesis

This conclusion is what I think Zeihan drastically underestimates and what Fukuyama is incapable of articulating. Yes, I think that our species has learned extensively about the hegemonic globalization of democratic capitalism, and the agnostic globalization of knowledge & invention as a public good. Even if there’s a black swan following the recession of a global economy and demographic upheaval, there’s too much continuity of self-awareness & inventiveness that we are unable to adapt.

Perhaps this is a good time to point out the unavoidable flaw of Zeihan’s & Fukuyama’s points of view. To say that a pyramid or ziggurat is stable height is to literally describe an architecture that acts proportionally against gravity. One facet of actual society & economy is time/perception arbitrage, where one can figuratively comprehend how to consume the sustaining layers below them at a rate that can be described as sustainable as long as they live to consume. To state it differently, sometimes politicians and corporate officers are aware that they are increasing the global ecological cost but are not accountable to it. Another way to put it is that many institutions & enterprises are a grift. Both Zeihan & Fukuyama entertain the idea of political decay, when a cascade of realized ecological costs become an inescapable downward spiral. But do they entertain a permanent gain of sustainable function from a unsustainable form of consumption?

Except for the very unlikely possibility that a cosmic event overwhelms our ability to store information, self-reinforcing decay is not possible. Simply put, there’s enough of an open-sourced corpus to replace many layers of the ziggurat, down to its ecological base. Much like an ecological disaster can be a near-extinction event, yet proves the tenacity of life on Earth; in the same respect we are tenacious as broadcasters & archivists of knowledge in the face of critical operational failure.

Which brings us to the Church-Turing thesis: the equivalence between Turing machines and λ-calculus, which is the computability of a given function. Simply put, it’s more notable to find something uncomputable like the halting problem (i.e. calculate the last digit of π that a given compute cluster will find before it runs out of resources). Another way to see this is in scientific progress, a progressive body of experimentation leading to testable theories, which conclusively lead to scientific laws (that have yet to be fully computed). It is most applied in computer science, leading to the possibility of atomically interpreting levels of ecology into more succinct informatics. This can be easily applied to financial markets, where we know that actors are unpredictable enough, and yet we can also model such markets stochastically & dialectically.

This does not mean that Hari Seldon and his Seldon crisis are possible, let alone inevitable. We can falsifiably claim that current distributed computing is redundant enough to preserve gains in “lambda abstraction” despite classic history informing us of critical failures that can force us to regress ecologically. Zeihan thinks we might run out of rare earth metals before we electrify our way out of climate catastrophe. Fukuyama might think that sociopolitical polarization will disrupt our ability to elect competent officials who will moderate the bureaucracy. Both might think that a labor crunch is a critical economic failure that will dismantle our current attempt to automate labor.

It’s important to see the forest for the trees. Church-Turing may not directly apply to how humans govern themselves on an socioeconomic level. We probably can’t perfectly compute cultural, artistic, or philosophical values. We won’t be manifesting Ultron or Vision. Not all jobs will be replaced by robots & AI. Instead, it’s more likely that we will continue to formalize the global ecology into a succinct form. It is likely that the global ziggurat gains broad & thin layers of knowledge, more than any pyramid or obelisk of pop culture, dominating the news cycle, gains any permanence or universality.

Practically speaking, what does this actually mean? Recently, I commented on the recent ChatGPT trend:

Even more recently, more developments have been released in the field of machine learning. Deepmind just released Dreamer v3, which uses a world model to more efficiently train an agent to operate in an environment with given parameters (like Minecraft). Geoffrey Hinton has discussed an alternative to the method of backpropagation called the fast-forward algorithm. Stephen Wolfram has addressed the absence of a preexisting knowledge base in LLM chatbots. Sparrow will likely release in 2023 with such a capability. These are part of the metaphorical ziggurat. Perhaps AI is not a zero-shot pursuit of consciousness, and likewise it is not a bundle of NLP, cultural “safety”, and a megacorporation’s endorsement. Ultimately, what humanity desires is a complex & reusable Church-Turing program that has already computed an expansive set of abstract ecological functions. We don’t just want CRISPR-edit therapy, we want the epigenetic control as well. There’s probably more we don’t know but will inevitably learn.

This is the quintessential response to Ludditism. In order to coexist with an ecosystem to the most intelligent degree, we expand the substrate for scientific discovery so we can compute the ecological functions with the least aggregate cost. A wider ziggurat can afford higher layers, and a falsifiably centered perspective from a higher layer might widen the ziggurat’s foundation. A niche profession like artistry might be off-center and nonfalsifiable, so it is a limited perspective as to how DALL-E, StableDiffusion, and other text-to-image models compute aesthetic value at minimal ecological cost. Stances on consumable GMOs might be a limited perspective on how biotic factors can regenerate an ecosystem at minimal ecological cost. In both cases, the ziggurat expands because some set of facts has been computed and broadcasted.

Try explaining this to a geopolitical expert that thinks of the world as a static pyramid. Try to explain the antifragility of chip production to someone who believes the nexus of <5 nm transistor density will get destroyed in an invasion, or the necessity of nuclear power in decarbonization within a world that has yet to procure sufficient desalination to complement insufficient rainwater. None of this is clear-cut, but as we progress as a civilization, the momentum of progress forces us to converge with the now-viable computation of the economy as a stochastic ecosystem, and the underlying ecology as a formalized ecosystem. While human history could have been viewed as the ultimate halting problem during the Cold War, now it can be transparently summarized to some articulable degree & digitally stored in an arctic vault.

Last year, I thought about the imminent “cambrian explosion” of space colonization:

It sounds like a gambit, but how quickly would shit have to hit the fan in order to prevent us from becoming a multiplanetary, Kardashev Type I species? I would guess that it is not quick enough that we cannot afford to preserve the “stack” of present technological gains. And we already know how to scale to Type I by expanding beyond Earth. In nature, information is preserved to be broadcast when it is viable in the future. We’re already at such a stage that humans will independently strive to broadcast our presence into the void. Perhaps it is just as reasonable to consider that humanity will continue to achieve further ecological formalization. In case it wasn’t clear by now, what I’m talking about is the global potential for near-permanent capitalization. Economics is optimized by reducing costs and increasing the ecological carrying capacity. This is the major distinction between AGI as a practical halting problem, or as a definitively disproven hypothesis of computability.

Just as Jules Verne predicted a trip to the Moon imperfectly, I think of Zeihan and Fukuyama as imperfect oracles of a partly true, familiar-seeming fiction. And so it will remain in the future; a lot of technologies are not even premature in our time, they are wholly undiscovered. We just have to be curious enough with the intellectual capital we already have for those unknown discoveries to manifest. We just need a little further ecological account to permanently reduce the unsustainable cost of human activity. Perhaps the wisest perspective we can afford is that calamity is neither guaranteed nor impossible, and our tenacity in the face of such calamity is neither limitless nor futile.