In 2023, OpenAI changed the world by releasing ChatGPT, the first AI that’s actually intelligent. Technology aside, ChatGPT shifted attention in the tech world away from crypto and NFTs to AI, resulting in capital and talent flight. The prices of cryptocurrencies going down and increased regulatory hostility from the SEC were the icing on the cake.
In this historical context, it’s easy for people still working in crypto (like yours truly) to start asking themselves: why not switch to working in AI? Is what we’re working on really valuable? Is working on crypto the best use of our time, despite all the amazing breakthroughs in AI? This isn’t a new phenomenon. When I first got into crypto back in 2017, AI was a hot topic much like today, and all the college kids were aspiring to become machine learning researchers at Google/DeepMind (remember them?) and HFT traders at Jane Street. I still vividly remember talking to an upperclassman who interned at Jane Street. When I mentioned that blockchain was the field to get into, his response was an unironic “You must be kidding, right?”. AI has always been more high-status than crypto, and ChatGPT has changed nothing in this respect.
Nevertheless, we need a good answer to the question: why work on crypto? More specifically, I want to answer the question: assuming your goal is to push humanity as a whole forward, how do you rationalize working in crypto as the best way to spend your time and effort out of the universe of possibilities? The following response is what I’ve believed in ever since I started working in crypto, and I hope it can help you, my dear reader, gain conviction in the value of crypto as well.
1. Technology increases authoritarianism
The more advanced technologies you have, the more powerful you become.
By power, I mean the capability of an individual or organization to modify the external environment to their will. It encompasses violent power, productive power, political power, and many other kinds of power. As you acquire better technologies, your power increases. For example, if you’re a farmer with a basic scythe, acquiring a combine harvester is going to dramatically improve the rate at which you can harvest your produce. If you’re a soldier with a spear, acquiring an assault rifle is going to significantly increase your ability to kill your enemies. More tech = more power.
Technological advancements increase the power of governments more than they increase the power of individuals. This is caused by multiple factors. First of all, governments enjoy the power of scale. When people come together and combine their capabilities into a coherent whole, the resulting organization is more capable than the sum of the individuals. This is not surprising, as the power of scale was one of the main reasons that governments formed in the first place. Important tasks like building canals for irrigation, fighting external threats, and providing relief for floods and droughts were beyond what an individual or a group of disorganized individuals could do, but could be accomplished by a large group of individuals coordinated with each other. Here’s a historical example: as knowledge of gunpowder spread to Europe, European powers began to include guns and artillery in their armies. Producing swords and lances was relatively simple: a village blacksmith could make them by himself, armed with only what’s in his workshop and knowledge passed down from the older generation. Producing guns and cannons, on the other hand, was much more demanding: you needed expensive precision tools wielded by engineers educated in math, physics, and chemistry, as well as artillerymen educated in the same subjects. This required schools and universities that could provide years of education to such people, researchers who could come up with new techniques and improve your weapons, not to mention enough capital to jumpstart and maintain an arms industry. This means that in the age of gunpowder, only governments had enough resources and coordinated individuals to field infantrymen armed with guns and cannons. An individual could go buy a gun and arm himself, but the power multiplier gained is going to be less than what an army could gain by acquiring thousands of guns and cannons and adopting more appropriate tactics.
(This is an oversimplification that does not apply to all technologies. Some technologies enhance individual power more than governmental power. Crypto is one such technology, and we will talk about it later.)
Secondly, governments often maintain monopolies on technologies, such that even if an individual knew everything about a technology, there’s nothing they could realistically do with that knowledge. Nuclear weapons are a good example. Given the knowledge of nuclear physics, it’s not hard for a sufficiently motivated individual to figure out the specifications for building a basic nuke. However, there’s a reason random people are not building nukes in their backyards: individuals can’t just go to the mall and buy some enriched Uranium/Plutonium, since such materials are under the tight control of governments. A corollary is that governments are often the only ones with the knowledge of certain technologies, allowing them to maintain a technological advantage over regular individuals. For example, the British government invented public-key cryptography three years before Diffie and Hellman published their algorithm.
These factors result in governments gaining relative power over individuals as technology advances, and with great power comes great corruption. As governments become more powerful, existing institutions for restricting their power become less potent, giving governments strong incentives to abuse their power. Technology usually increases the power of the state without simultaneously increasing the power of institutions that restrict the state’s power. In The Origins of Political Order (which everyone should read IMO), Francis Fukuyama provided three dimensions for assessing a political entity: the power of the state, the rule of law, and popular accountability. Western democracies rank high on all three dimensions, authoritarian regimes like China and Russia have powerful states but no rule of law or popular accountability, and certain democracies like India have the rule of law and popular accountability but also have weak states that can’t effectively implement their aspirations. As technology develops, governments get shiny new toys for exerting their wills and improving their internal efficiency, but the rule of law and popular accountability don’t automatically advance in lockstep. The US has essentially the same model for restricting governmental power today as it did back in 1789, with Congress and the Supreme Court keeping the executive in check, but today the US government has computers, nukes, fighter jets, and aircraft carriers. Increasing the rule of law depends on things like improving public education and changing social norms, and increasing popular accountability requires reforming political institutions, all of which often take decades if not centuries to accomplish. Humanity is always going to consist of millions and billions of people with complex preferences and conflicting interests, so calling for them to change their politics in a particular way is always going to be a grueling process.
(Institutions for keeping governments in check do evolve over time, just much more slowly. Before the French Revolution, countries around the world became more absolutist, with governments acquiring better weapons, building bigger armies, and increasing centralization. This can be seen in Louis XIV’s famous (though misattributed) quote: L’état, c’est moi, meaning I am the state. The Enlightenment and the subsequent development of liberal democracy caused a reversal in this trend, a process that took only hundreds of years and millions upon millions of dead people to complete.)
When governmental power relative to individuals increases enough, a kind of phase-shift occurs where the role of government in society becomes fundamentally different. It may seem pedestrian to us that modern governments have their hands in nearly all aspects of life, but this hasn’t always been the case. Governments used to be far less powerful and only performed a limited number of functions in society. For example, in medieval England most courts were under the purview of local lords rather than the king, partly due to the fact that the central government simply didn’t have enough bandwidth to handle all court cases in the country. The king thus played the role of impartial mediator rather than overlord, often traveling around the country in “circuits” to local courts to provide judgments unaffected by parochial interests. This is where modern circuit courts originated. The central government didn’t really have a hand in how most people lived, chiefly because it wasn’t able to.
As modern nation-states developed, however, central governments built increasingly powerful bureaucracies and shifted away from playing the impartial judge to playing parent for the whole nation. This happened to France a few decades prior to the French Revolution when it transformed from a medieval feudal state into the first modern state. In The Old Regime and The Revolution by Alexis de Tocqueville, it was mentioned that after this transformation,
…neither city, nor borough, nor village, nor hamlet, however small, nor hospital, nor church, nor convent, nor college, could exercise a free will in its private affairs, or administer its property as it thought best. Then, as now, the administration was the guardian of the whole French people…
With the speed at which technology is advancing today, it’s likely that we will see another such phase-shift. Which brings us to AI.
2. AI will push us towards totalitarianism
AI is a technology that can significantly enhance the power of governments and thus push them further towards authoritarianism. In fact, governments may go beyond authoritarianism and land squarely in the realm of totalitarianism.
With AI, governments will likely shift away from governing nearly all aspects of public and private life to all parts of life period. One way that technology has enhanced state capabilities is via increasing legibility. Making a polity legible to its government means wrapping it in a model that the government can understand and make use of. For example, a city-state in ancient Mesopotamia would have kept records of who owned which parcels of land and how much agricultural product each parcel produced, and using this information it could calculate how much tax each land-holder owed to the state each year. In other words, the state made land-ownership legible to itself, and thus could collect taxes based on land-ownership. Something that Mesopotamian city-states didn’t keep track of was individual income, because most people didn’t have jobs with monetary wages. This means individual income was illegible to the city-states, which was part of the reason why they didn’t collect income taxes. What is legible to the state can be taxed, judged, and regulated by the state.
Today, most aspects of your life are legible to the state. The state knows where you live, where you work, how much you earn each year, what your net worth is, what car you drive, what stocks you invest in, and taxes you for all of them. There still exist things that are illegible to the state: what you talked about with your friends over dinner the other day, what your private political views are, how happy you’ve been today, etcetera. Armed with AI and modern surveillance infrastructure, however, the state can make everything legible.
With the ubiquity of smartphones and the looseness of privacy laws, at any point in time some organization out there is almost certainly keeping track of where you are, what you’re doing, what you’re saying in real life as well as online, what websites you’re reading, and maybe even what expression you’re making on your face. Currently, most of this data is not used productively because there’s just too much of it and it’s too expensive to comb through it all: your government doesn’t really want to hire someone to monitor what you discussed with your parents on FaceTime last night (unless you’re really important).
Using AI, however, the government can combine all the data it has on you into a coherent profile and analyze it in real time as data streams in. Imagine if whenever you open your mouth and say something, the smartphone of you or somebody near you would pick it up, transcribe it, and use GPT-4 to determine whether you said something the government didn’t like. If you did, the government would be able to punish you accordingly: maybe they’d fine you, maybe they’d suggest to your boss that you should be fired, maybe they’d ban you from leaving the country, or maybe they’d just jail you. Western democracies probably won’t do this anytime soon, but authoritarian regimes like China and Russia are likely already scrambling to implement something along these lines. Whatever is legible to the state can be controlled by the state, and when everything becomes legible, everything can be controlled by the state. AI will make everything legible, which is why it’s incredibly dangerous even if we could keep it under human control.
AI is also a technology that heavily advantages state-level actors over individuals. Let’s make four assumptions:
- The more compute time you use in training an AI, the more intelligent it is.
- An AI with higher intelligence is almost always better than an AI with lower intelligence, even if it was slower and/or more expensive.
- Having a better AI increases your productivity in all areas.
- A government has vastly more resources compared to individuals and even gigacorporations within the borders of the polity it governs.
Using these assumptions, we can make the following argument:
- A government can easily become the owner of the most computing power in its polity, due to the immense resources it commands.
- Therefore, it can become the owner of the most intelligent AI in its polity.
- Therefore, it can become the owner of the best AI in its polity.
- Having the best AI makes the government the most productive at AI development.
- Therefore, the government not only can have the best AI but also can improve its AI the fastest, ensuring that it will maintain its lead.
Creating AI regulations can further guarantee that governments will have the best AIs albeit unnecessary.
AI will make governments significantly more powerful in the near future, and it’s unlikely that we will be able to reform the institutions restraining state power quickly enough in response. It’s possible that AI will be what breaks the camel’s back and make governments too powerful to be contained by such ancient institutions as courts and voters. This is why I personally believe it’s ethically untenable to work on AI development today. Sure, AI has the potential to increase individual welfare by orders of magnitude, but we’ll ultimately be worse off due to the empowerment of authoritarianism and totalitarianism.
3. Crypto is our only hope of fighting back
Both blockchain and cryptography are part of the group of rare technologies that empower individuals more than governments.
Blockchain and cryptocurrencies are pretty much useless to governments. Their utility stems from their resistance to fraud and abuse by a powerful, centralized actor, but governments are such powerful and centralized actors. What use would a government find in a system premised on the distrust of governments? Cryptography is useful to both governments and individuals, but I would argue that it’s especially useful to individuals. The power of most technologies scales with the amount of resources you have: for example, if you had a single assault rifle and I had a thousand of them plus a thousand people to operate them, I would overpower you despite us having the same technology. This doesn’t apply to cryptography. Cryptography puts everyone on a level playing field, regardless of the amount of resources they have. If I encrypted a file using AES-256 and only I knew the password, I would know with near certainty that no government will be able to recover its content, because the time required to brute force through all 2^256 possible keys would be orders of magnitude more than the current age of the universe. I have no adequate words to describe how amazing cryptography is.
In a world descending into authoritarianism, crypto provides perhaps the only tools with which we can fight back. If we want to fight back against authoritarianism, we will likely need to form an organization with a pool of resources and many members. Such an organization requires three things to get off the ground. First, it needs a method for enforcing contracts between members, so that they can collaborate without the fear of broken promises. Contracts are traditionally enforced by governments, but if governments are what we’re fighting against, then we cannot rely on them to enforce our contracts. This is where blockchain comes in: smart contracts are a solution to enforcing contracts without a government. Second, the organization needs a way to raise, store, and transfer funds, because what organization doesn’t? Fighting against the government while using traditional banks and fiat currencies under their control is just begging to be shut down. Cryptocurrency is the obvious solution to this. Finally, members of the organization need to be able to communicate with each other outside of government surveillance, which can be solved via cryptography.
Crypto is both radical and conservative. It’s radical for people already living under authoritarian regimes, since it enables radical change towards a more liberal society. It’s conservative for people living in liberal democracies because it protects their existing way of life against growing authoritarianism. Ordinary people living in the West will one day need to use crypto just to live and breathe as freely as they do today.
This is why we work on crypto. We need to build more financial tools and attract economic value into crypto so that if and when we need to organize against the encroachment of authoritarianism, there will be enough censorship-resistant capital at our disposal. We need more private communication tools to ensure that our private conversations stay private. We need more private blockchain networks to make our smart contracts illegible to governments. The more people participate in an economy illegible to the government, the less control the government will have, the further away we stay from authoritarianism.