5. Tokens
Technologies that change society are technologies that change interactions between people.1
Single-Player and Multiplayer Technologies
If you were stuck on a desert island, alone in the world, money wouldn’t be very useful. Lacking connectivity, neither would computer networks. On the other hand, a hammer, box of matches, or food supplies would come in handy. So might a stand-alone computer, if you had a power source.
Context matters. Some technologies are social, and some are not. Money and computer networks are social technologies. They help people interact with other people. Sometimes, borrowing from video games, people call technologies that are useful alone single player. Social technologies are, by analogy, multiplayer.
Blockchains are multiplayer. They let you write code that makes strong commitments. Individuals and organizations don’t have much need to make commitments to themselves. That’s why attempts to create “enterprise blockchains,” which function exclusively inside existing corporate organizations, haven’t been successful. Blockchains are useful for enabling coordination among people who don’t have preexisting relationships. They are most useful when they are not just multiplayer but massively multiplayer—in broad use across the internet.
Any social technology that tries to scale to billions of people needs simplifying assumptions. Software, where every line in a code base is a logical statement, can be complicated. At the scale of the internet, which five billion people use today, it gets even more complicated. Each overlapping logical interdependency introduces a greater likelihood for mistakes. More code means more bugs.
A powerful way to address this complexity is through a software technique called encapsulation. Encapsulation reins in complexity by circumscribing units of code within well-defined interfaces, making the code easier to use. If that sounds unfamiliar, it might help to think about an example from the physical world, a device that’s so simple people rarely give it much thought: the electrical outlet.
Anyone can plug into an outlet to access electricity and run any number of appliances: light fixtures and laptops, alarms and air conditioners, coffee makers and cameras, blenders and blow-dryers, Xboxes and Model Xs, and so on. Outlets unlock the electrical grid and give humans superpowers without anyone having to understand what’s happening on either side of the socket. Outlets abstract away the details. The interface—the encapsulation—is all that matters.
Because software is so flexible, encapsulated code has another benefit: it can easily be reused. Encapsulated code is like Lego bricks. The bricks can be combined into sets of bricks that create much larger and more impressive structures. Encapsulation is especially helpful when large groups of people are developing software, as is the case for most modern software. One developer can create a few Lego bricks—basic pieces of programs that can, for example, store, retrieve, or manipulate data, or access various services, like email or payments. Other developers can then take those components and reuse them, without either party having to understand the details of what the other is doing. The bricks just snap into place.
When it comes to blockchains, a key simplifying concept is units of ownership called tokens. While people often think of tokens as digital assets or currencies, a more accurate technical definition would describe them as data structures that can track quantities, permissions, and other metadata for users on a blockchain. If that sounds abstract, that’s because tokens are an abstraction. This abstraction makes them easy to use and simple to program. Tokens encapsulate complicated code into an uncomplicated wrapper, just like an electrical outlet.
Tokens Represent Ownership
What tokens are matters less than what they do.
Tokens can represent the ownership of anything digital, including money, art, photos, music, text, code, game items, voting power, access, or whatever people come up with next. Using some additional building blocks, they can also represent real-world things, like physical goods, real estate, or dollars in a bank account. Anything that can be represented in code can be wrapped inside a token to be bought, sold, used, stored, embedded, transferred, or whatever else a person might want to do with it. If that sounds so simple as to seem trivial, that’s by design. Simplicity is a virtue.
Tokens enable ownership, and ownership means control. Tokens that run on traditional computers, like the hypothetical Google Coin example from earlier, can be taken away or changed at will, undermining user control. Tokens that run on computers that can make strong commitments about future behavior—namely, blockchains—unlock the technology’s true potential.
Take games, for instance. Digital objects and virtual goods have existed for a long time in computer worlds. Popular games like Fortnite and League of Legends make billions of dollars per year selling virtual goods2 such as cosmetic items for players’ avatars. These kinds of digital goods aren’t bought; they’re borrowed. Users are renters. The company behind a game can remove or change the terms at any time. Users can’t transfer the goods outside the game or resell them or do any of the things people associate with ownership. The real owner—the platform—calls the shots. If the value of an item goes up, the user doesn’t reap the reward. Almost invariably, games eventually fade away or shut down, and along with them their virtual goods blink out of existence.
The same is true of most popular social networks. As we’ve covered, users don’t own their names and followers. Platforms do. Some recent examples of Big Tech’s bigfooting: When Facebook rebranded itself as Meta in October 2021, the company a few days later revoked the Instagram handle of an artist,3 @metaverse. (After an outcry and a New York Times article, Meta reinstated her account.) Similarly, when Twitter rebranded itself X in 2023,4 it commandeered the @x handle from a longtime user. Oustings like this happen all the time. You don’t have to look far for examples of political figures, activists, scientists, researchers, celebrities, community leaders, and other users getting suspended by corporate networks.5 Companies that control networks have complete control of accounts, ratings, social relationships, and more. User ownership in corporate networks is an illusion.
Blockchains shift control to software governed by immutable code, not people, and thereby make ownership real. Through the building block of tokens, they give the concept of ownership teeth.
In the early web, the concept of a website played a similar role as a building block. The web’s founding idea was to have a sea of information, connected by links, controlled by many different people. It was a profound and ambitious vision, one that could have gotten mired in complexity. But websites were designed to be simple units that could provide a foundation for more complex constructions—building blocks that could, at scale, create the digital equivalent of city blocks.
The read era of the internet was defined by the website, which encapsulated information. The read-write era was defined by the post, which encapsulated publishing, making it easy for anyone, not just web developers, to reach broad audiences. The internet’s latest phase—the read-write-own era—is defined by a new simplifying concept: tokens, which encapsulate ownership.
The Uses of Tokens
Tokens, while simple seeming, are not simplistic. They are an expansive technology that comes in two overarching types:6 fungible tokens, like bitcoin and ether, and non-fungible tokens, also known as NFTs.
Fungible tokens are interchangeable. One token in a set of fungible tokens can be swapped for any other token in the same set. It’s apples to apples. Money is similarly fungible. If someone has $10, they don’t care which $10 bill they have, just that they have $10.
With NFTs, each token is unique, in the same way many objects in the physical world are unique. I have a set of books—different titles, different authors—on my bookshelf that are distinct and, despite all being books, are not interchangeable with one another. These are non-fungible.
Fungible tokens have many uses, the most prominent being as a way for software to hold and control money. Traditional financial applications don’t hold money. They hold references to money, but the money itself resides somewhere else, like a bank. Money that is held and controlled by software is a new idea that didn’t exist before blockchains.
The best-known example of a fungible token is a cryptocurrency like Bitcoin. Many public discussions assume that cryptocurrencies are the main use for blockchains. Prominent voices who promote Bitcoin as an alternative to government-controlled money exacerbate the confusion. As a result, many people incorrectly associate blockchains and tokens with libertarian politics even though these technologies are, in fact, politically neutral.
Cryptocurrency, as in new systems of money, is only one of many uses of blockchains and tokens. Fungible tokens can also be used to represent national currencies. People call currency-pegged tokens7 stablecoins, since they tend to be less volatile than other tokens. One common misconception is that stablecoins pose a threat to the status of the U.S. dollar as the world’s reserve currency. In fact, the opposite appears to be true. Demand for internet-native dollars is so strong that most stablecoin issuers have opted to peg their stablecoins to the U.S. dollar. U.S. congressman Ritchie Torres (D-N.Y.), a member of the House Financial Services Committee, which monitors stablecoin adoption, has argued8 that the technology “reinforces rather than challenges the supremacy of the U.S. dollar” and has “enabled the U.S. to outcompete countries like China in the realm of digital currencies even without a CBDC,” an abbreviation for “central bank digital currency.” (So far, the U.S. government has no CBDC, whereas the People’s Bank of China mints a digital renminbi.)9
In the absence of a U.S.-government-backed stablecoin, the private sector has produced a number of stablecoins that vary in the way they maintain their pegs. Some stablecoin issuers back their tokens, one for one, with fiat money held in a bank. USD Coin (USDC) is a popular fiat-backed stablecoin10 that’s managed by a financial tech firm called Circle. The system is designed so that one token can be redeemed for one U.S. dollar. When people trust that the tokens can be redeemed for dollars, they value the tokens that way, even if they rarely redeem them. Many applications use USDC tokens for programmatic money transfers, including decentralized finance (DeFi) applications.
“Algorithmic” stablecoins are another model. These try to maintain their pegs through automated market-making processes. To stay solvent, they automatically sell collateral, such as tokens held in escrow, when market prices decline. Thanks to careful stewardship of reserves, some algorithmic stablecoins, most notably a system called Maker, have successfully maintained their pegs even during periods of extreme volatility. Other algorithmic stablecoins that played fast and loose with collateral have collapsed, including Terra, which infamously crashed in 2022.11
Tokens are general-purpose software primitives. They can be designed well or poorly. It’s worth noting, by the way, that some people differentiate between the terms “coins,” “cryptocurrencies,” and “tokens.” As you might have noticed, I treat these mostly interchangeably, although I, and many others in the industry, do prefer “tokens” because the term communicates the abstract, generalizable nature of the technology. “Tokens” sounds neutral and so rings truest: it doesn’t over-index on financial aspects, as “coins” does, and it doesn’t have the political connotations of “cryptocurrency.”
Another use for fungible tokens is as fuel for blockchain networks. Ethereum has a native fungible token, called ether, that serves a dual purpose. The first is as a means of payment within Ethereum-based networks, like NFT marketplaces, DeFi services, and other apps. The second is as payment for “gas,” a measure of computational effort, which Ethereum requires to run software on its network. Many other blockchains use the same design, requiring token payments to purchase computing resources. The pay-as-you-go model, a throwback to computing in the 1960s and 1970s when time-sharing on mainframes was popular, has made a comeback.
Non-fungible tokens also have multiple uses. NFTs can represent ownership of physical items, like artwork, real estate, and concert tickets. Some people have bought and sold property, such as apartments, using NFTs (tied to LLCs) to transfer ownership and keep a record of transactions, similar to a deed. NFTs are best known, however, as a way to represent ownership of pieces of digital media. The media can be anything, including art, videos, music, GIFs, games, text, memes, and code. Some of these tokens have code attached that can do things like manage royalties or add interactive features.
Because NFTs are so new, it’s not always clear what buying one means. In the physical world, when you buy a painting, you are buying the object and the right to use it. You are generally not buying the copyright to the art, or the right to prevent others from using its likeness. Similarly, when you buy an NFT representing an artistic image you are generally not buying the copyright (although buying this is possible—it just depends on the design of the token).
Most of today’s NFTs act more like signed copies, analogous to autographed paintings or record albums. The value of an artwork depends on many things, including its scarcity and critical appraisal, but it also depends on some complex mix of social and cultural signals. People assign financial premiums beyond utility value to many things, including art, baseball cards, handbags, sports cars, and sneakers. Similarly, people can assign premiums to tokens that represent objects with cultural or artistic significance. Value is a function of many factors, some objective and some subjective.
NFTs can also have digital utility. One popular use of NFTs is to track transactions such that artists can receive royalties from secondary sales. In games, NFTs can represent objects, skills, and experiences that give players special items and abilities—a warrior’s sword, a wizard’s wand, a new dance. They can provide access to subscriptions, events, or discussions, as they do in some popular token-gated social clubs through which members convene both digitally and physically.
Another use for NFTs is connecting digital and physical objects. Tiffany & Co. and Louis Vuitton created NFTs12 that can be redeemed for jewelry, handbags, and other merchandise. The artist Damien Hirst created a collection where the NFTs represent digital artworks13 but can also be redeemed for physical versions. Other NFTs blur the line between the digital and physical worlds. Nike created NFTs that represent digital sneakers14 that owners can display and wear in the video game Fortnite. Owners also get access to new product drops and events like chats with pro athletes.
For users, the NFTs act as a digital twin of the physical object, breaking down the barrier between the on- and offline worlds. They get the usual benefits of owning the physical products plus online benefits like the ability to trade on marketplaces, showcase on social sites, or equip characters in games. The brands get an ongoing digital relationship with their customers, something most don’t have today.
NFTs can also act as identifiers, analogous to DNS names. Recall that by giving users ownership of their names, DNS lowered switching costs in protocol networks. NFT identifiers can play a similar role in newer social networks, letting users switch applications with their names and connections intact.
Users hold and control tokens through software “wallets.” Every wallet has a public address, derived from a public cryptographic key, that acts as an identifier. If someone knows your public address, that person can send you tokens. If you have the corresponding private key, you control the tokens in the corresponding wallet.
The term “wallet” originated back when tokens were used solely to represent currencies, but the label is somewhat misleading today. Wallets continue to serve as persistent inventories of tokens that can be carried across the internet, but they are also used for many other types of token, application, and software interactions. A better analogy is that wallets are to blockchains as web browsers are to the web; they’re interfaces for users.
Like wallets, “treasuries” bundle tokens together and serve as an interface for users, but they do so at a greater scale. Whereas wallets are mostly used by individuals, treasuries make it easy for larger groups to coordinate. On Ethereum, you can write a treasury application that puts a community, often called a DAO, or decentralized autonomous organization, in control. The community can vote on how to administer the treasury’s assets, such as by funding software development, security audits, operations, marketing, R&D, public goods, charitable donations, or educational initiatives. Both wallets and treasuries can also be set to autopilot, automatically investing or disbursing money, or engaging in other programmatic activities.
If tokens are like cells, then treasuries are like full-blown organisms. Treasuries are multiplayer reserves, controlled by software that ensures tokens move only according to prescribed rules. These capabilities give blockchains the ability to stand up to offline organizations like companies or nonprofits—they give blockchains muscle.
The Importance of Digital Ownership
Maybe all this sounds far-fetched or inconsequential. People like to joke that DAOs are just a “chat group with a bank account,”15 that NFTs are just glorified JPEGs, and that tokens are no better than Monopoly money. Even the word “token” is reminiscent of games and arcades. But it would be a mistake to underestimate these technologies’ significance.
Blockchains represent a radical departure from the status quo. Through tokens, they flip the script on digital ownership—making users, rather than internet services, owners.
People are mostly accustomed to the reverse. They’re used to having all the things they acquire online remain attached to digital services. The same is true of many downloads. You don’t really own that e-book you ordered from Amazon Kindle, or that movie you bought from Apple’s iTunes store,16 for example. Companies can revoke these purchases at will. You can’t resell them. You can’t transfer them from one service to another. Every time you sign up for a new service, you have to start from scratch.
The only internet objects most people feel as if they own are their websites, and only when they own their own domain names. I own my website because I own the domain. As long as I stay within the law, no one can take it away from me. Similarly, companies own their corporate domains. That the one digital asset people feel as if they own is built on the web is no coincidence. Protocol networks, like blockchain networks, respect digital ownership. Corporate networks do not.
Most people are so habituated to the corporate network norm they don’t even register its peculiarity. In the physical world, people would be upset if they had to start over whenever they visited a new place. We take for granted that we have a persistent identity and can take objects from place to place. The concept of ownership is so deeply embedded in our lives that it’s difficult to imagine how the world would look if that were taken away. Imagine if the clothes you bought could be worn only in the venue you bought them in. What if you couldn’t resell or reinvest in your house or car? Or what if you had to change your name wherever you went? This is the digital world of corporate networks.
Perhaps the closest offline analogue to corporate networks are theme parks, where a single company tightly controls the entire experience. Theme parks are fun to visit, but most of us wouldn’t want our everyday lives to work that way. Once you pass through the turnstiles, you’re subject to the business owner’s unchallenged policies. In the real world, outside the park gates, people have agency. They have the freedom to do with their possessions as they please, such as to open shops and businesses that resell goods, and to bring their possessions wherever they wish. People get value and satisfaction from owning and investing in things.
Ownership has positive secondary effects too. Most people’s wealth comes from appreciation of the assets they own, such as their houses. Homeowners are known to invest17 in and care for their places and, by extension, their neighborhoods far more than renters. Improving one’s lot improves everyone’s lot.
Ownership is also a prerequisite for many startup ideas, so innovation depends on it. A novel service like Airbnb can exist only in a world where people are free to do as they like with their homes, including renting them out. The process of making physical goods usually requires taking other goods as inputs and remixing them without asking for permission. Buy whatever inputs you want and do whatever you want with them because you own them. Many businesses take existing things and reuse them in ways their original creators never imagined—and sometimes might not like. Within limits, like patent law, ownership is a basic freedom that means you don’t have to ask for permission to do something new.
The importance of ownership may seem obvious when laid out as I’ve done above, yet most of us don’t really think about it in the context of the internet. We should. The digital world would be a better place if ownership were as widespread there as it is in the physical world.
The Next Big Thing Starts Out Looking Like a Toy
Today tokens are used by a small group of enthusiasts, a tiny proportion of total internet users, perhaps a few million people. It is easy to underestimate them, these early adopters of an odd-seeming, outside-in technology. But this would be a mistake. Big trends start small.
One of the amazing things about the technology industry is how often tech giants miss major new trends18 and allow startups to rise up as challengers. TikTok mastered short-form video before anyone else, catching tech giants like Meta and Twitter off guard. It wasn’t as if these incumbents were being complacent; most of them aggressively crushed, copied, acquired, and built products to avoid being displaced. Instagram and Twitter had video capabilities well before TikTok became popular, but they prioritized their legacy products instead. Twitter shuttered its short-form video app Vine in 2017. A year later, TikTok went viral in the United States.
The reason incumbents whiff19 is that the next big thing often starts out looking like a toy. This is one of the main insights of the late business academic20 Clayton Christensen, whose theory of disruptive technology starts with the observation that technologies tend to get better at a faster rate than users’ needs increase. From this simple insight follow nonobvious conclusions about how markets and products change over time, including how startups are so often able to take incumbents by surprise.
Let’s review Christensen’s theory. As companies mature, they tend to cater to the high end of a market and improve products by increments. Eventually, they add capabilities that exceed what most customers want or need. By this time, the incumbents have developed myopia, focusing on profitable niches to the exclusion of the low end of a market. And so they overlook the potential of new technologies, trends, and ideas. This creates an opening for scrappy outsiders to offer cheaper, simpler, and more accessible products to a wider array of customers who are less demanding. As the new technology improves, the newcomer’s market share grows until it eventually overtakes the incumbent.
When disruptive technologies debut, they’re often dismissed as toys because they undershoot user needs. The first telephone, invented in the 1870s, could carry voices only short distances. The leading telco of the time, Western Union,21 famously passed on acquiring the phone because it didn’t see how the device could possibly be useful to the company’s primary customers, which were businesses and railroads. What Western Union failed to anticipate was how rapidly telephones and their underlying infrastructure would improve. The same thing happened a century later22 when minicomputer manufacturers, like Digital Equipment Corporation and Data General, ignored PCs in the 1970s and, in ensuing decades, when desktop computing leaders, like Dell and Microsoft, missed out on smartphones.23 Time and again, a sling and rock beat a lumbering swordsman.
Yet not every product that looks like a toy will become the next big thing. Some toys remain just that, toys. To distinguish the duds from the disrupters, products need to be evaluated as processes.
Disruptive products ride exponential forces that cause them to improve at surprising rates. Products that get better incrementally are not disruptive. Bit-by-bit improvements yield weak forces. Exponential growth comes from stronger forces that have compounding effects, including network effects and platform-app feedback loops. Software composability—a property describing code that is reusable so developers can more easily extend, adapt, and build on what exists—is another source of exponential growth. (Much more on this in “Community-Created Software.”)
The other critical feature of disruptive technologies is that they are misaligned with incumbent business models. (I discuss in detail how tokens fit this mold in part 5, “What’s Next.”) You can be sure that Apple is working on phones with better batteries and cameras. It would be foolish for a startup to try to compete with the company on that basis. Apple knows that improving its phones will make the phones more valuable and help it grow its core business: selling phones. A more interesting startup idea would be something that makes phones less valuable. This is something Apple is far less likely to pursue.
A product doesn’t have to be disruptive to be valuable, of course. There are plenty of products that are useful from day one and continue being useful long-term. These are what Christensen calls sustaining technologies. When startups build sustaining technologies, they are often acquired or copied by incumbents. If a company’s timing and execution are right, it can create a successful business on the back of a sustaining technology.
Few people doubt the significance of many modern technology trends, including artificial intelligence and virtual reality. These inventions play to the advantages of companies like Meta, Microsoft, Apple, and Google, which have the computing power, the data, and the resources to fund their costly development. Big Tech is investing heavily in these areas. Upstart competitors, like OpenAI, need to raise billions of dollars just to compete. (OpenAI has reportedly raised $13 billion from Microsoft.)24 While some people question how these technologies will square with the traditional ways these companies make money, it’s likely that they will extend preexisting business models. In other words, they’re sustaining technologies.
To be clear: I believe AI and VR have profound potential, so much so that I co-founded an AI startup back in 2008 and was an early investor in Oculus VR (which Facebook bought in 2014). My point is just that Big Tech recognizes the potential of these technologies too, which makes them less disruptive in the strict sense intended by Christensen. Although people now use “disruption” casually, the term has a precise academic meaning. Disruptive technologies are, by definition, harder to spot than sustaining ones. They elude experts—and that’s the point. Incumbents missing disruptive innovation is what makes it disruptive.
One might be forgiven for mixing up the categories. Even Christensen, the expert on the matter, erred. He famously misread the iPhone25 as a sustaining technology, miscalculating that the device would merely extend the market for phones when, in fact, it would disrupt a much bigger potential market—the market for computers. Such is the innovator’s dilemma; even innovators nod.
Incumbents are opening themselves to disruption yet again. Few big companies have taken blockchains and tokens seriously to date, unlike what they’ve done with AI and VR. Established players don’t recognize their significance. In the years since Bitcoin and Ethereum debuted, only one tech giant has made a real run at tokens. Meta started a blockchain project called Diem, formerly Libra, in 2019. Two years later the company sold off its assets and shut down its related digital wallet product,26 Novi. It’s no coincidence, in my view, that Meta also happens to be the only Big Tech company still led by its founder. It takes a visionary to even try to buck convention.
Tokens have all the earmarks of a disruptive technology. They are multiplayer, like websites and posts, the disruptive computing primitives of earlier internet eras. They become more useful as more people use them—a classic network effect that primes them to be much more than mere playthings. The blockchains that underpin them are also improving at a rapid rate, driven by platform-app feedback loops that generate compound growth. Tokens are programmable, so developers can extend and adapt them for myriad applications, such as social networks, financial systems, media properties, and virtual economies. They are also composable, meaning people can reuse and recombine them in different contexts, amplifying their power.
Skeptics who once dismissed websites as “dot-bombs” and who similarly derided social media posts as nothing more than idle chatter failed to see their power. They misunderstood—and missed out on—the extraordinary forces that network effects unleash. New trends and inventions take hold when the networks that sprout around them kick off compound growth. Websites rose in tandem with the read-era protocol network of the web. Posts rose in tandem with read-write-era corporate networks like Facebook and Twitter.
Tokens are, in the read-write-own era, the latest computing primitive to grow and flourish amid a new kind of internet-native network.