Malthus, Romer, Moore, AI, Degrowth, And What It All Means

Brief Historical Review ‍

To start, a summary of where we’ve been and where we are today:

‍

In 1826, philosopher-economist Thomas Malthus rose to fame for his “Essay on the Principle of Population”, in which he claimed that Britain’s population would soon outstrip its food supply, leading to famine, political and social instability, and widespread suffering.

‍

Roughly 165 years later, Stanford economist Paul Romer developed a growth model that, in part, attempted to explain why Malthus’ predictions never came to fruition. His model’s most important contribution is the idea that changes to technology are endogenous to growth, and thus advancements in technology improve economic output. When Malthus wrote his essay, he only saw the effects of improvements in public health represented by quickly ballooning population. He did not foresee complementary technological advancements in agriculture, transportation, etc that in sum contributed to the ability to feed such a historically large population. Romer's model of economic growth accounts for all of these technological advancements taken together, helping to reduce the delta between the observed effect and modeled effect.

‍

Between Romer and Malthus, in 1965, Intel cofounder Gordon Moore famously observed that the number of transistors in an integrated circuit was doubling about every two years. This observation held approximately true for so long that it became known as Moore’s Law, which can be derivatively simplified to represent the notion that total compute power will double approximately every two years.

‍

4A. As of this writing, we are living in the midst of unprecedented advancements in both microchip and machine learning development. In microchips, Nvidia has pioneered accelerated computing, a type of computing where extremely powerful microchips can be networked together to tackle exceptionally complex computational tasks across the algorithmic, network, and data center levels at significantly reduced cost. On the back of these microchip advancements, machine learning researchers have made tremendous headway in domains like reinforcement learning and natural language processing, among others. The result is that, for the first time in history, advanced machine learning algorithms are not only commercially accessible to governments and corporations, but to consumers as well, and they are becoming both more powerful and more efficient at breakneck speed.

‍

4B. Yet, as our technology becomes more sophisticated, we are nonetheless living in the Malthusian specter of an exponential rise in greenhouse gas emissions, largely due to increased economic growth, that has led to a congruous rise in global temperatures and wrought havoc on environmental ecosystems. Most experts agree that without some change in parameters, our current economic path is untenable for human existence on this planet.

‍

Finally, confusing things even more, the unprecedented economic growth that is ransacking Earth’s resources and environment has had bifurcated social effects. In rich countries, profits have been inequitably distributed due to skepticism of social safety-net programs, leaving many people working jobs that will soon be at-risk of automation with very little protection. Developing countries, in contrast, have seen a historical number of people lifted out of poverty, with many living lives better than they ever could have imagined as children.

‍

So what does this all mean? How are we to make sense of all this?

‍

Economic Growth ‍

Market-Based Pricing ‍

Perhaps the first place to start is economic growth and how we measure it.

‍

Global growth, as you likely know, is typically measured by aggregating Gross Domestic Product for every country in the world. GDP (for short) is in turn an aggregation of the value of all the goods and services exchanged in a country within a certain timeframe. So growth, whether nationally or globally, is fairly straightforward: prices staying constant, more goods and services exchanged equals more growth. More is better, a mantra that should feel, if not sound, quite familiar to many of us.

‍

The problem, if you want to look at it that way — Hello Joseph! Hello Mao! — is that in market-based economies, prices don’t stay constant: they go up and down based on how much people want things (demand) and how many of those things there are (supply).

‍

Demand is fairly easy to understand. Organizations mostly want things out of necessity, whereas people mostly want things to make themselves appear more fuckable. Organization needs something? Price goes up. Something makes you a whole lot more fuckable? Price goes (way) up.

‍

If you’re a liberal arts student, supply is much more boring; if you’re a STEM student, much more soothing.

‍

Historically on the supply side, pricing has been driven primarily by the Cost of Goods Sold (COGS, for my accounting nerds ✊🏻). That is to say, the material, labor, production, transportation, and packaging costs that go into putting a good on the market for sale. Any time you’re on Amazon and you think to yourself “How the *#&@ does this cost $12?!” it’s your intuition telling you you’re paying way higher than COGS for your item. There are lots of ways supply can drive pricing, but for the sake of simplicity and coherence, let’s just talk about COGS for now.

‍

Now the trick to all of this is that when demand is high, price initially goes up, but then supply increases and prices go back down. So there is a kind of price equilibrium. That’s why USB-C cables used to cost $20 and now they cost $.25.

‍

So, in summary, when we’re talking about economic growth, we’re really talking about the cumulative value of exchanged goods and services at prices that themselves change based on necessity, utility, and the cost of inputs for bringing a product or service to market.

‍

Machine Learning & Supply-Side Growth ‍

With a basic understanding of how economic growth is measured and how its variables change relative to one another, the next place to look for some sense in where we are today and where we might be heading is technology's affect on supply.

‍

Recall Paul Romer’s contribution to the study of Economics, which is to model technological advancement as a coefficient to growth, where the effect of improved technology is modeled as a full-blown upward shift in supply curves. A technological breakthrough that affects any of the inputs to Cost of Goods Sold would theoretically shift the supply curve.

‍

Next consider the effects of Moore’s Law as it relates to accelerating advancements to microchips and machine learning algorithms becoming adopted across global supply chains. These advancements won’t cause a single shift due to a small improvement to material extraction or processing, or to cheaper labor or production. They will cause a cascading and compounding series of shifts - many of them astoundingly large - as new technology displaces antiquated technology and the expensive human labor that supports it.

‍

The economic effect will be deflationary - consistent and high downward pressure on prices of both goods and services. For example, as the service of marketing a piece of software (i.e. my job) becomes easier due to technological advancement, the price of marketing services will drop, fewer people will market software, and the ones who do will on average make less money. The world economy, in theory, should shrink, which is one of the state goals of the Degrowth movement.

‍

But just because prices drop due to the displacement of expensive labor, it doesn't mean that our overall draw on natural resources or the emissions from economic production that are driving climate change will also fall. In fact, holding other parameters constant, widespread artificial intelligence adoption will likely accelerate humanity’s deleterious effect on our environment in the short term. Things will (potentially) get much worse before they get better.

‍

So what’s to be done? Is there a way out? Much like in Malthus’ time, there is hope in both expected and unexpected places.

‍

For one, like in the 1800s, our current economic and climate models likely underestimate complementary shifts in technology. The potential downstream effects of advancements in energy production, distribution, and consumption, specifically, are likely not well understood. (Commercial electricity as we know it is only ~150 years old. The idea that we are anywhere near optimal at anything related to electricity seems unlikely.) The same can be said for advancements in agriculture. There will likely be major breakthroughs in these sectors that allow us to significantly reduce greenhouse gas emissions while maintaining economic productivity.

For another, like with the economic growth of the Baby Boomer generation, the largest marginal gains from technological advancements throughout global supply chains will likely be made by the world’s poor. As goods become cheaper, more people should be able to afford the basics and be lifted out of poverty, and as quality of life improves, the global population will likely shrink since children will no longer be directly tied to economic productivity in the poorest countries. As a whole, there will be fewer people, but people will live longer, better lives than ever before.

‍

While the idea of fewer people, and thus fewer consumers, seems promising, thinking through the effects on the demand-side presents a more complex picture.

‍

Knock-On Effects: Demand ‍

If we presume that the previous section's thesis is accurate - that widespread machine learning adoption throughout global supply chains will drive efficiencies and thus lower prices - then our next task is to understand the response in demand.

‍

Put succinctly: without the discovery of new, less resource-intensive ways of producing the goods and services we consume today, the primary benefits of advanced machine learning - reduced COGS via more efficient supply chains - will result in lower prices, more consumption, and more depletion to Earth’s natural resources.

‍

Given that there is only so much surface area on Earth, there are likely only so many trees from which to make paper, rubber bands, etc from; there is only so much fresh water; there is only so much arable farmland, crude oil, iron, lithium, phosphorous, and so on.

‍

One could argue that the reduction in COGS will primarily come from subtracting human labor from an already shrinking global population, and thus wages will drop and consumption won’t increase as dramatically as it otherwise might. But this would rely on the assumption that we can’t update the social contract with something like Universal Basic Income (UBI) to afford everyone a basic standard of living.

‍

The counter argument has two parts:

‍

We should and will need to update the social contract with something like UBI so that the benefits of society’s advancement are widely distributed Without any other changes to production, the increase in demand that is spurred by UBI will tax the planet’s natural resources in unprecedented ways. As resources dwindle, prices will once again rise and COGS will go back up. ‍

Assuming this happens, the social contract will already be extended and no amount of additional income will be able to stop the fact that there simply aren’t enough barrels of oil and that’s why your bottled water costs $10 - the bottle alone costs $7. Perhaps an exaggeration, but such a thought experiment clarifies what a world with depleted resources might look like.

‍

This is why the demand side is more complicated: there isn’t a clear economic solution to the human desire for a better life. In many ways, in spite of our technological advancements on the supply-side, we failed to escape the Malthusian dilemma's resource constraint: without unforeseen technological advancement, demand for resources will outstrip not only our ability to produce necessary goods and services, but will outstrip Earth’s ability to yield the raw material inputs necessary to produce those goods and services.

‍

Unlike other Malthusian dilemmas, the problem here is that our tendency to seek happiness in the un-winnable struggle to keep up with the Joneses, Kardashians, or anyone else is the actual culprit, and without some kind of revolution of the soul, we will not escape our own self destruction.

‍

And so it is in our own human nature that hope springs eternal.

‍

One Way Out ‍

Anyone who visits an art museum and looks at the paragons of male and female beauty from the Renaissance knows that human desire is inconsistent. Food is short? We find beauty in fat. Outdoor labor signals poverty? We desire pale skin. Whatever will differentiate us into something more likely to be desired for reproduction (i.e. fuckable), that’s what we want.

‍

And so what if we were to turn having a smaller resource footprint into a symbol of status and wealth? Rather than having more and bigger and the most expensive, we turn our standards on their head and the way to keep up with the Jones’s becomes about quality, artistic intention, and labor and environmental equity throughout a given product's supply chain?

‍

You can already see the seeds of this movement starting to sprout in the fashion industry. Monikers once synonymous with quality craftsmanship like Louis Vuitton, Gucci, Versace, Apple, etc have all been co-opted into very expensive mass-produced status symbols. The next generation won’t desire these brands for the very reasons they have become successful: they are mass produced and thus overly abundant and non-differentiating, and they are expensive without the quality and craftsmanship deserving of the price tag.

Instead, young people who grow into the next generation of consumers will, ironically, likely want to be more like Apple co-founder Steve Jobs, who famously wore a $400 black mock turtleneck designed by Issey Miyake throughout Apple’s resurrection from nearly-dead-has-been to the most powerful company in the world. For someone of his stature and wealth, Jobs opted for a modestly priced, plain, comfortable, well-made garment designed in an artist’s personal studio. (Jobs, according to his biography, was also particularly considerate about the household items with which he furnished his suburban Palo Alto home, famously contemplating “What is a washing machine?” while considering between models. No one ever said Steve wasn’t extreme.)

‍

Jobs' example points one path forward, and it seems to be the one gaining the most traction today. In garments, we are already seeing a new generation of slow-fashion brands that are focused on selling the value of their items’ craftsmanship and environmental and social impact by educating their customers about the various supply chain inputs required to create their products. The opposite of SHEIN, which sells cheap products meant to be discarded after one fad ends and another begins, these brands typically embody a “Buy Once, Cry Once” fashion ethos where they make their money upfront on a lower volume of purchases rather than attempting to sell you more and more lower quality items that you don’t really need or, if you really think about it, want.

‍

Extrapolating such an ethos across consumption of all types of products, there is a chance we would reduce total resource consumption globally in the short-term, giving the Earth, and thus ourselves, a chance to recover while we allow technology to discover novel, sustainable ways to allow us to consumer more without such harmful effects.

‍

Of course, tying this all together is extremely complex and won't just happen overnight.

You need advancements in Machine Learning to bring down the cost of goods sold so that more profits can go to creators rather than redundant middlemen.
You need a renegotiated social contract with something like Universal Basic Income so that wealth and thus disposable income is broadly distributed despite profits becoming more concentrated as labor becomes less important to producing goods and services.
You need status to shift from the fast fashion and mass produced luxury brands ethos that dominate our social media feeds today to smaller producers focused on more socially and environmentally sustainable supply chains.

‍

It’s a lot to ask for much less expect, and there is likely a lot of pain in our short-term futures before we arrive at a solution that is workable for both society and the environment, but there are ways out of our current situation. We just need to seek them out, and when we find them, pursue them.‍

Further Reading