14. Some Promising Applications

Social Networks: Millions of Profitable Niches

In his classic 2008 essay,1 “1,000 True Fans,” Kevin Kelly, the founder of Wired, predicted the internet would transform the economics of creative activities. He saw the internet as the ultimate matchmaker, enabling twenty-first-century patronage. No matter how niche, creators could discover their true fans, who would, in turn, support them:

To be a successful creator you don’t need millions. You don’t need millions of dollars or millions of customers, millions of clients or millions of fans. To make a living as a craftsperson, photographer, musician, designer, author, animator, app maker, entrepreneur, or inventor you need only thousands of true fans.

A true fan is defined as a fan that will buy anything you produce. These diehard fans will drive 200 miles to see you sing; they will buy the hardback and paperback and audible versions of your book; they will purchase your next figurine sight unseen; they will pay for the “best-of” DVD version of your free YouTube channel; they will come to your chef’s table once a month.

Kelly’s vision hasn’t exactly panned out. The reality is creators do, generally, need millions of fans, or at least hundreds of thousands, to support themselves today. Corporate networks got in the way, inserting themselves between creators and audiences, siphoning away value and becoming the dominant way for people to connect.

Social networks are probably the most important networks on the internet today. Besides their economic impact, they have a huge effect on people’s lives. The average internet user2 spends almost two and a half hours a day on social networks. Next to text messaging, social networking is the most popular online activity.

The design of the dominant social networks explains what went wrong. Powerful network effects locked users into Big Tech’s clutches, and that lock-in led to high take rates. It’s hard to know precisely what take rates many major corporate networks charge, because their terms can be opaque and noncommittal, but it’s reasonable to estimate they charge around 99 percent. With the combined revenue of the five biggest social networks—Facebook, Instagram, YouTube, TikTok, and Twitter—at about $150 billion per year, that means these networks pay out on the order of $20 billion to users, with the overwhelming majority of that share coming from YouTube alone.

Corporate networks won out because they made it easy for people to connect—more so than protocol networks like RSS did. But that doesn’t mean corporate networks are the only, or even the best, way for people to connect. The alternative to today’s world would be one where social networks are decentralized and community owned, meaning built with either protocol or blockchain architectures. This could have meaningful economic effects for users, creators, and developers and could revive Kelly’s compelling vision for internet patronage.

To understand the effect of a different network design, let’s do some back-of-the-envelope math. Protocol networks have take rates that are effectively zero. Sometimes companies build apps on top of these networks, providing easy access and other features. Substack does this for email newsletters, and it charges roughly 10 percent for the convenience. (Substack’s take rates stay low, as in other markets that respect ownership, because users own their connections—that is, their email subscriber lists—which they can export and load into rival email access providers at any time.)

Let’s pretend the top five social networks charged a similar amount. If they all had take rates of 10 percent, their share of the $150 billion in annual revenue would drop from $130 billion to $15 billion. That would put into the pockets of network participants such as creators an extra $115 billion per year. How many lives might that change? At the average U.S. salary of $59,000 per year,3 that extra $115 billion of redirected revenue could fund almost two million jobs. This is a rough estimate, but the numbers are clearly big.

Low take rates have a multiplier effect. More money to the edges of the network means more people reach an income level where they can pursue creative work full-time. The two-tier class system that divides creators and users on most social networks would become more permeable. Barriers to social mobility would ease as more users could build sustainable media enterprises of one. Meanwhile, full-time work would result in better-quality content for others to consume, attracting larger audiences and generating more income across the network.

Better economics for creators leads to a virtuous cycle. Millions of people working full-time on creative activities improves the quality of the internet for everyone. Social networking should be a place to chat and trade memes, but it should also supercharge longer-form activities: writing essays, creating games, making movies, composing music, recording podcasts, and more. These endeavors require time, money, and effort. For the internet to be an accelerator of deep creativity, it needs a better economic engine.

Creating new jobs isn’t just nice; it’s necessary. As new technologies like AI automate work, social networks can be a counterweight that provides people with fulfilling career opportunities.

Decentralized social networks would also be good for users and software developers. The high take rates, capricious rules, and platform risks of corporate networks are deterrents for developers. In contrast, decentralized networks encourage investment and building. With more tools being built, users can shop around for a greater variety of software and features. Choice drives competition, which leads to better user experiences. Don’t like the way a client ranks posts, filters spam, or tracks your personal data? Switch. Nothing’s holding you back, and you won’t lose your connections.

This might all sound great in theory. The practical question is whether today, given where we are in the evolution of social networking, it’s possible to build a decentralized social network that can actually succeed. Occasionally users awaken to the problems of today’s platforms, and after an incident happens—a deplatforming, a rule change, a new corporate owner, a data privacy or legal scandal—people flee to some upstart social network. These anti-communities usually don’t last. Durable social networks are built out of friendships and shared interests, not anger.

The value proposition needs to be full parity with corporate network user experiences, plus much better economics. Corporate social networks succeeded because they made it so easy for people to connect. It’s not too late to design decentralized social networks that make connecting just as easy. Protocol social networks like RSS were a good starting point, but they failed because they lacked the features and funding of corporate rivals. Blockchains can address both shortcomings. We can now, for the first time ever, build networks with the societal benefits of protocol networks and the competitive advantages to rival corporate networks. Indeed, the timing is right: blockchains have only recently become performant enough to support social networking.

Today, a cohort of blockchain projects is taking on the social networking establishment. Each project is designed in its own way, but the common thread is that each one overcomes the weaknesses that doomed RSS. The best designs fund software developers and subsidize username registrations and hosting fees through their token treasuries, analogous to corporate coffers. And in terms of features, blockchains have core infrastructure that provides a centralized global state to support basic services, making it easy to search and follow across the entire network, avoiding the user experience issues that the partitioning in protocol networks and federated networks creates (as covered in “Federated Networks”).

The key marketing challenge is to kick-start a network effect. One tactic is to start on the supply side, where the pain of high take rates is greatest. Users may not realize how much value they’re forgoing by participating in corporate networks, but creators and software developers care deeply about how much money they earn. Offering a predictable platform where they receive a greater share of the value they generate would be a compelling proposition. If the best content and software were available only on another platform, the demand side of the network—the users, many of whom are passive consumers—would likely seek it out. That users can participate in a blockchain network’s economic upside and governance, privileges from which they were previously excluded, adds further motivation for them to switch.

Starting in narrow and deep niches could help a new social network get over the initial hump. Targeting a group with common interests, like people interested in new technologies or new media genres, is one way to plant the seeds of a community. The most valuable users will likely be up-and-comers who don’t have big followings elsewhere. When YouTube started out, it didn’t succeed by getting creators from TV and other forms of media. New stars rose along with the platform. That’s the power of native over skeuomorphic thinking.

What we have today may feel like a golden age for creative people: creators can push a button and instantly publish to five billion people. They can find fans, critics, and collaborators just about anywhere on earth. But they’re mostly forced to route everything through corporate networks that devour tens of billions of dollars that might otherwise have funded an immeasurably greater diversity of content. Imagine how much creativity we’re missing out on because earlier attempts at decentralized social networks, while noble, like RSS, couldn’t hold their own.

We can do better. The internet should be an accelerant for human creativity and authenticity, not an inhibitor. A market structure with millions of profitable niches, enabled by blockchain networks, makes this possible. With fairer revenue sharing, more users will find their true callings, and more creators will reach their true fans.

Games and the Metaverse: Who Will Own the Virtual World?

The plot of Ready Player One, the most popular recent book about the so-called metaverse, revolves around a contest to see who gets to control the OASIS, the book’s 3-D virtual world. I won’t spoil who wins the contest, but the real issue is not who wins—it’s that one person can control that virtual world at all.

Ready Player One builds on a tradition of speculative fiction that owes much to Neal Stephenson, the sci-fi author who coined the term “metaverse”4 in his 1992 novel, Snow Crash. Back when Stephenson was writing, 3-D multiplayer games had simple graphics and supported interactions among only a few players. Obviously, the state of the art has come a long way since then. Today, game graphics rival Hollywood movies, hundreds or even thousands of players can interact in the same virtual world, and audiences for games number in the hundreds of millions of players. Video games like Fortnite and Roblox are the closest equivalents we have today to full-fledged virtual worlds like the OASIS.

You hardly have to squint to see where we’re headed. Soon enough, digital worlds will have lifelike graphics and let many thousands if not millions of people play together. Audience sizes will continue to grow, and people will spend more time in game worlds. High-quality virtual reality headsets will be common. Haptic interfaces that provide physical feedback will make the experience even more realistic. Artificial intelligence will create an abundance of rich characters, worlds, and other content. The trends all point in this direction.

As the quality of virtual experiences improves, digital interactions will spill over into the physical world. You might make friends, meet a future spouse, or get a new job in “virtual” reality. As more of the economy moves online, more jobs will exist solely in online worlds. The distinction between work and play will blur. What happens in digital worlds will have consequences and meaning in the physical world, and vice versa. The same pattern unfolded in social networking. Twitter started off as a way to share what you had for lunch, and now it’s at the center of global politics. Things that look like toys sometimes remain toys, but sometimes they become much more than that.

As the metaverse vision materializes, a central question is how these worlds will be designed and what architecture will underpin them. Today’s most popular video games use the corporate network model. The players connect through shared virtual worlds that game development studios control. Many of these game economies have digital currencies and virtual goods, but they are centrally managed, with high take rates and limited opportunities for entrepreneurs.

The alternative to the corporate model is an open model based on either protocol networks or blockchain networks. Tim Sweeney, the founder of Epic, maker of Fortnite and the popular game engine Unreal, describes his vision of an open metaverse5 as a combination of the two:

We need several things. We need a file format for representing 3D worlds …. These could be used as a standard for representing 3D content. You need a protocol for exchanging it, which could be HTTPS or something like the Interplanetary File System, which is decentralized and open to everyone. You need a means of performing secure commerce, which could be the blockchain, and you need a realtime protocol for sending and receiving positions of objects in the world and facial motion ….

We’re several iterations away from having the remaining components for the Metaverse. They’re all similar enough that a common denominator could be identified and standardized, just like HTTP was standardized for the web.

Sweeney’s vision is on the right track, but he could take the openness even further. Limiting blockchains to commerce is skeuomorphic thinking. Blockchains are computers, capable of running arbitrary software. The strongest form of an open metaverse would be a collection of composable blockchain networks, each of which meets one of the needs Sweeney describes, and which themselves interoperate, forming a meta-network. This could start from an initial core blockchain network and expand out to a fabric of interconnected networks, the whole composed from its parts, built from the bottom up.

It wouldn’t take much to meet the technical specs. Fungible tokens representing virtual currencies and NFTs representing virtual goods would flow freely through the network. Some NFTs would be “soulbound,” or nontransferable, representing a special achievement or an item forever tied to the person who attained it. Some NFTs would be tradable commodities, like virtual clothing, or “skins,” that can be bought and sold. And other NFTs would be a combination, with some features that are tradable and some that are not. An avatar might acquire experience points that reset on transfer, for instance.

The game designers would have a rich design space to work in. They would build applications on top of an underlying blockchain network, but they would still have access to all the tools of today’s game designers. They would also get new design elements like persistent, transferable ownership and economies that span across networks.

Fees paid back to the blockchain networks could cover development costs. The take rates would be low, as they should be in blockchain networks, but a larger total economy driven by entrepreneurship would compensate. Creators could set up shops to sell their wares, keeping most of their earnings for themselves. Investors would have an incentive to fund entrepreneurs building on top of the network, knowing upside wouldn’t be capped. The interoperability and composability of blockchain networks would mean users could move between games and applications, migrating from one network to another, creating competition among networks. Digital property rights would be guaranteed by persistent, blockchain-enforced rules. Governance and moderation would be managed by the community.

In corporate networks, cross-network interoperability is often considered a liability. Blockchains reverse this logic and make interoperability a tool for growth. If one network builds a community of token holders, another network can incentivize that same community to become its own participants, such as by offering to support the other network’s tokens in its applications. So, the sword and potion collection someone spent years building in one game doesn’t go to waste when that person stops playing. The player can transfer it to a new game. Maybe the graphics and gameplay are different, but the item and core properties persist.

To the extent that a protocol network like the web can help build a metaverse, I welcome it. But as Sweeney points out, many pieces will still need to be built that protocol networks like the web don’t provide. If open systems like protocol or blockchain networks don’t step in to fill the void, corporate networks will, and then the world will end up with the dystopia of Ready Player One.

NFTs: Scarce Value in an Era of Abundance

Copying is a core activity of the internet. When people write online, information gets copied from their machines to servers and then back to readers. Almost every action a person takes, from likes to posts to retweets, creates copies. Copies are free and frictionless, producing a flood of videos, memes, games, messages, posts, and more.

Copying is both good and bad for creators. On the one hand, it distributes creative work to a wide audience. On the other hand, the abundance of media creates heated competition for attention. Networks route and prune this information; nevertheless, far more flows in than anyone could possibly consume. The good news is you can instantly reach five billion people. The bad news is so can everyone else.

Traditional media businesses rely on scarcity to make money. In the pre-internet world, media, like books and CDs, were limited. Only so many were produced. People had to seek out and acquire physical goods. In the digital world, where information flows freely by default, abundance is the norm. Many media businesses protect their interests by imposing restrictions, like paywalls and copyrights. You need to pay to read articles in The New York Times or listen to music on Spotify. (Pirating media is obviously illegal, and it has gotten less appealing as legal alternatives have sprouted up over time.)

Scarcity can convert attention into money, but scarcity also prevents media from benefiting from the supercharged copying machine that is the internet. Friction reduces the chance that content will survive in the struggle for attention. Restricted content can’t be shared or remixed as easily as public content, for instance. This is what I call the attention-monetization dilemma—the trade-off media creators face between maximizing attention and maximizing money.

The video game industry is far ahead of other media businesses in navigating this dilemma. Games tend to have short lives and must adapt to changing technologies and trends. A few long-standing titles like Madden and Call of Duty are exceptions, but most other games come and go. As a result, the industry is fast moving, highly competitive, and open to experimentation. The enduring companies embrace new technologies and business models. The lessons video game studios have learned are applicable to other forms of media. Game makers just learned them sooner.

For many years, video game studios made money the same way just about all media businesses did. People would pay a onetime fee, typically around $50, to buy a game, whether as a physical CD or a digital download. With the advent of the internet, new genres of video games, such as massively multiplayer role-playing games and battle royal shooters, emerged and took advantage of the internet’s native capabilities. New activities like streaming and new business models like sales of virtual goods became popular.

While experimenting, game studios made a discovery. They realized they could make even more money on free games.6 Taking their sole source of revenue and giving it away free of charge was a daring move, but it worked.

In the early days of the internet, game designers would offer a few levels for free and then charge for the full game.7 In the 2010s they took that idea further by giving away the full game and just charging for add-ons. Today, the most sophisticated games—including Fortnite, League of Legends, and Clash Royale—make all their money charging for virtual goods,8 which generally don’t even make players better at the game. Mostly, the goods are cosmetic, like new outfits or animations for your character. (When people can “pay to win,” players usually revolt.)9

Video games solved the attention-monetization dilemma. Making the game free meant the game and all its derivative works—videos, memes, and so on—could spread freely across the internet. As a result, game-related content has become a consistently top-trending category across social media. The biggest game releases routinely have higher sales than the biggest movie releases10 (a trend amplified by the recent pandemic).11 In 2022, the gaming industry brought in12 around $180 billion in global revenues, seven times as much as global movie box office revenues. What was once a niche activity for enthusiasts is now a blockbuster pastime.

The gaming industry’s savvy is also evident in its approach to streaming. On sites like Twitch users watch live videos of players who also chat with the audience—a mix between sports spectating and talk radio. Legally, it would be easy for the industry to crack down on streaming. When game streaming began in the late 2000s, some companies, notably Nintendo, pushed back.13 But today every game company encourages streaming because of an industry-wide realization that the attention gained more than offsets the monetization lost.

Game studios were smart. They looked at their products expansively, as bundles that included a game, but also streaming and virtual goods. Through experimentation they discovered the right blend of free and paid elements to optimize the trade-off between attention and monetization. In the process they created a new, scarce layer of value. Thus, the games themselves went from paid to free while adding new layers like streaming (free) and virtual goods (paid). Game studios shrank one part of the revenue balloon while finding other parts to inflate.

In contrast to video games, the music industry responded to the rise of the internet by squandering time filing lawsuits against innovators.14 Music companies focused far more on protecting existing business15 than on exploring new businesses. Only after a great deal of foot-dragging did record labels accept incremental change, such as allowing streaming services like Spotify to use their content in subscription bundles. And they weren’t happy about it.

The kicking and screaming persists to this day. When music startups try to find new ways to navigate the attention-monetization dilemma, record labels usually threaten them with lawsuits. These threats chill experimentation. New music-related tech products, to the extent there are any, are generally just minor variations of previous products. Novel approaches are considered too risky and expensive.

The effects are palpable. Founders create hundreds of video game startups each year, but they found very few music-related startups. That’s because entrepreneurs want to spend their time inventing new things, not getting sued. Investors have learned their lesson too. They rarely back startups16 related to music.

The outcome of these two approaches shouldn’t be surprising. As the next graphs show, the revenues of the video game industry17 have far outpaced the revenues of the music industry18 over the past thirty years. The games industry grew by embracing each new wave of technology. The music industry’s litigious approach stunted growth.

There is nothing magical about video games that makes them more able to be monetized than other forms of media. People love video games, but people also love music, books, films, podcasts, and digital art. These other creative industries have simply experimented with fewer new business models. People make and listen to music as much as ever. The problem isn’t supply and demand; it’s the broken business models in between.

What virtual goods did for video games, NFTs can do for other forms of internet media. NFTs create a new layer of value—digital ownership—that didn’t exist before.

Why would people pay for digital ownership? There are many reasons, but one is the same reason people buy art, collectible toys, and vintage handbags: an emotional connection to the ideas and stories behind the goods. Think of buying an NFT as buying an official product from a brand, or a copy of an artwork signed by an artist. The NFT connects you to the brand, artist, or creator as well as to a collector community through an immutable signature trail. The more you copy, remix, and share the art, the better known it becomes, and the more valuable the connections between the creator and the community may become.

But NFTs aren’t just art. They’re general-purpose containers for representing ownership. This means you can also design NFTs to have value that goes beyond buying an official or signed work. One popular NFT design gives owners behind-the-scenes access or memberships in private discussion groups. NFTs can also confer voting rights, enabling people to guide the creative direction of characters and stories in narrative worlds. (More on this in “Collaborative Storytelling: Unleashing Fantasy Hollywood.”)

Skeptics sometimes suggest that NFTs will restrict the sharing of media. In fact, NFTs provide an incentive to loosen restrictions. Copying and remixing generally increase the value of NFTs, just as more players in video games increase the value of virtual goods. The same effect happens with physical art too. Both owner and artist can benefit from copying because, as the art is more widely shared, the original copies can grow in value. In the extreme case, a work of art, like the Mona Lisa, can become a widely reproduced cultural icon.

Art generally doesn’t come with embedded copyrights. When you buy a painting, you typically aren’t buying the copyright. Instead, you are buying the physical object and a license to use and display it. The value is more emotional and subjective. You can’t analyze cash flows or use other objective valuation methods. NFTs that represent signed copies are similar.

Yet NFTs are flexible, so creators can embed copyrights if they so choose. The simplest example is an NFT-plus-copyright where the purchaser acquires a traditional copyright. Because NFTs can include code, however, you can create copyright variations that would be hard to implement in the offline world. For example, you can design an NFT where the purchaser is granted commercial rights but must share some revenue back with the original creator. You can also have different rules for remixes and derivative works. Taking advantage of the built-in audit trails of blockchains, you can encode rules that pass money back to different sets of owners and contributors. A remix of a remix might keep a third of the revenue for itself, pass a third back to the remix, and a third back to the original. It’s software; you can design it any way you want.

NFTs can transform the economics of creators too.19 Consider the music business again.20 There are around nine million musicians on the streaming service Spotify,21 yet fewer than eighteen thousand musicians—less than 0.2 percent of them—made more than $50,000 in 2022. Most of the revenue generated went to streaming services and music labels. Tokens cut out layers of high take-rate intermediaries. With NFTs, musicians keep most of the revenue and can therefore support themselves with much smaller fan bases.

Musicians often sell physical merchandise, which also cuts out high take-rate intermediaries. But physical merchandise tends to be a much smaller market than digital merchandise. The music industry sold $3.5 billion22 in merchandise in 2018, whereas the video game industry sold $36 billion in virtual goods23 that same year—a figure that has nearly doubled for video games since. Digital goods are also higher margin, leave more room for product experimentation, and make it easier to maintain ongoing interactions with fans.

For those accustomed to the corporate network model, NFT-based businesses require a mindset shift. In the corporate approach, a company manages an entire service end to end. It builds the core service, the supporting apps and tools, and the business model around it. It’s command and control from beginning to end.

With NFTs, creators start with core, minimal components, like a simple collection of NFTs, and independent third parties build applications from the bottom up around the network and tokens. A band might issue NFTs that attract patrons and hard-core fans. Third-party applications that provide experiences around the NFTs—like access to private events, forums, or exclusive merchandise—might come later.

Third-party developers have an incentive to build around these NFTs for two reasons. First, to accelerate adoption of their products and services by piggybacking on existing communities. A marketer could give NFT holders in a target demographic exclusive perks, like early or free access to new products. In the blockchain model, interoperability becomes a customer acquisition tactic.

Second, NFTs are credibly neutral. The users own the NFTs, and the creator who issued them can’t change the rules (unless explicitly enabled by the code). The incentives are very different from those in corporate networks, where interoperating is risky because corporate owners almost always end up changing the rules in their favor.

Here again the analogy of theme parks and cities is useful. The corporate model is like a highly managed theme park that builds the whole experience end to end. The blockchain network is like a city that starts with core building blocks and encourages bottom-up entrepreneurship. NFTs with permissive copyrights encourage third-party innovation and so fit naturally in the city model.

NFTs are still evolving, but there are early signs of success.24 The NFT standard was formalized25 in 2018, and NFT sales began growing in 2020. From 2020 to early 2023, creators received about $9 billion26 in payments from NFT sales. YouTube, a much more established player, paid out about $47 billion27 during the same period. (That’s the 55 percent paid to creators out of the $85 billion in revenue that YouTube brought in during the same period.) Instagram, TikTok, Twitter, and others paid out almost nothing to creators.

The trend toward abundant media will only accelerate with the rise of generative AI, which can already create impressive visual art, music, and text and is improving so rapidly it will likely, maybe even someday soon, surpass human abilities. Just as social networks democratized content distribution, generative AI will democratize content creation. This will make the model of restricting media—the copyright model—difficult to sustain. People won’t be willing to pay as much for media when they can generate compelling substitutes with AI.

Fortunately, value doesn’t disappear. As the balloon gets squeezed, value shifts to adjacent layers, as covered in “Take Rates.” Chess-playing AI has trounced humans for two decades, yet playing and watching chess on websites like chess.com is more popular than ever. People crave human interaction despite the rise of machine intelligence. Post-AI artistic expression will focus less on the media itself and more on the curation, community, and culture around it.

NFTs add layers of scarce value onto a sea of abundant media. They provide an elegant solution to the attention-monetization dilemma. Creators can make money through new business models, inspired by virtual goods in video games. The internet can keep doing what it does best, copying and remixing. It’s a win-win.

Collaborative Storytelling: Unleashing Fantasy Hollywood

When the British writer Arthur Conan Doyle28 dispatched his best-known character, Sherlock Holmes, over a Swiss waterfall in an 1893 story, fans were dismayed. Thousands of Holmesians canceled their subscriptions to The Strand Magazine, the publication that had serialized his tales. They wore black in mourning and wrote a torrent of letters pleading for the detective’s resurrection. (Doyle ignored them for years, until he finally caved and brought Holmes back.)

To this day, nothing excites people’s passions quite like a good story. Internet forums are full of fans of narrative universes like Harry Potter and Star Wars who follow every update, dissect lore, and bicker over the significance of even minor plot points. Sometimes fans develop their own story lines and characters, even going so far as to write entire books on fan fiction sites like Wattpad. (Fifty Shades of Grey began as one such homage, to the Twilight series.)

People get so deeply invested in a franchise that it can become part of their identities. Yet the feeling of ownership is an illusion. Fans might, collectively, have some influence over a story’s direction—because people hated the irritating alien Jar Jar Binks so much,29 many believe George Lucas cut the character’s role in subsequent Star Wars films—but for the most part fans are just passive observers, with no formal voice and no financial stake.

Meanwhile, the media world is addicted to sequels and reboots because it’s risky to market new intellectual property. Media companies need to spend tens of millions of dollars to promote new stories. It’s safer just to recycle proven material.

But what if fans really could be owners, and media companies could harness their energy to help create and evangelize original stories? That’s the idea behind a group of new blockchain projects that enable fans to collaboratively create narrative worlds.

When given the right tools, diverse groups of strangers can work together to create great things. That’s a key lesson from the read-write era, of which Wikipedia is the most stunning example. The crowdsourced encyclopedia, founded in 2001, defied skeptics who viewed it as a digital graffiti wall run by utopian radicals. Today, most people barely remember Encarta,30 the Microsoft-owned knowledge compendium, written by paid experts, which was once considered the favorite to win the digital encyclopedia war. Wikipedia continues to face endless spam and defacement, but its community soldiers on, undaunted, editing and improving the site. Positive edits outnumber the negative, netting out to steady progress.

Today, Wikipedia is the internet’s seventh most popular website.31 People accept it as a trusted reference. The site’s success has inspired a wave of other collaborative knowledge projects, including the question-and-answers sites32 Quora and Stack Overflow.

Collaborative storytelling combines the lessons of Wikipedia with the power of credibly neutral, low take-rate blockchain networks that reward fans with ownership over their creations. The way this works most commonly in practice is that users receive tokens in proportion to their contribution to the narrative corpus. The resulting intellectual property is controlled by the community and can be licensed to third parties to make books, comics, games, TV shows, movies, and more. Licensing revenue is sent back to the blockchain network treasury, where it can be held to fund further development or be distributed back to token holders.

These projects give users a say in how the characters and stories develop. If they don’t like the current narrative path, they can “fork” characters by copying them and changing them to versions they do like. They can even fork complete stories, creating alternative timelines and worlds—whole user-generated multiverses. Characters and stories become composable Lego bricks for people to mix and match, mod and remix.

The collaborative storytelling model has multiple benefits:

• ▪ Widening the talent funnel. Permissionless access removes gatekeepers and broadens who can contribute to the writing process. The traditional media model uses gatekeepers to green-light people and projects. Creative jobs often still depend on living in the right city and knowing the right people. It’s a narrow funnel that likely overlooks a wide range of talent. Wikipedia brought the bazaar model to an encyclopedia industry dominated by cathedrals; collaborative storytelling can do the same for media.
• ▪ Viral marketing of new IP. Harnessing fandom is a powerful way to market new story universes without spending millions on advertising. Think of the viral marketing power of memecoins like Dogecoin, but imagine it focused on meaningful narratives instead of meaningless speculation. Enthusiastic fans go from passive consumers to active evangelists.
• ▪ Increased creator income. Token rewards can boost the income of creators. Blockchain networks have low take rates, which means most of the money earned goes back to creators. Removing layers of intermediaries transforms creator economics. A million dollars doesn’t mean that much to a big studio, but it can mean a lot to groups of independent creators.

Wikipedia defied skeptics to become an essential resource. Blockchain networks can extend the model Wikipedia pioneered to collaborative creative work, letting creators own a stake in what they create. Cuy Sheffield, head of crypto at Visa, calls this idea “fantasy Hollywood,”33 drawing an analogy to fantasy football. The model turns fans into active participants, and in this case they are actually in the game, not imagining it.

Making Financial Infrastructure a Public Good

When the commercial internet rose in the 1990s, it promised to modernize payments. But moving money online turned out to be difficult. Basic security measures34 like encrypted internet traffic were nascent and controversial. People didn’t trust entering credit card information online. Some companies like Amazon managed to win customers’ trust, but getting users to make electronic payments was a challenge for most.

So, many internet services flocked to advertising. Ads created a frictionless, closed loop that was effective from the start. The first banner ad,35 purchased by AT&T, appeared on the Wired site hotwired.com in 1994. A few years later advertising companies like DoubleClick held hotly anticipated IPOs. The ad gusher hasn’t stopped flowing since—with all the cluttered experiences and user tracking that entails.

It wasn’t until the 2010s that business models based on payments caught up to those based on advertising. E-commerce was the obvious beneficiary. People are now comfortable using debit and credit cards at miscellaneous merchants around the world. Shopify, which provides services to smaller e-commerce merchants, became a credible Amazon rival by riding this trend.

Freemium and virtual goods are the other popular payment-based models. Freemium providers give away a free version of a service and upsell a premium version. This model is used by media companies like The New York Times and Spotify, social networks like LinkedIn and Tinder, and software providers like Dropbox and Zoom.

Video game studios pioneered the virtual goods model, as covered in “NFTs: Scarce Value in an Era of Abundance.” As in the freemium model, providers gave away the basic product—in this case, a game—in hopes that a subset of users would buy add-ons. Some of these à la carte items might be useful in the game, like weapons, but many are purely cosmetic, like new outfits for players’ avatars. This model has supported several megahits, such as Candy Crush Saga, Clash of Clans, and Fortnite.

Although internet payments are now common, they are still high friction. Users have to enter credit card information. Incidents of fraud and charge-backs are high. Credit card fees are between 2 and 3 percent, low by the standards of other internet take rates, but high enough to prohibit many possible uses. (As discussed earlier, mobile platforms charge much higher fees, up to 30 percent of app store transactions.)

It shouldn’t be this hard to move money; sending money should be as easy and cheap as text messaging. The internet is the greatest tool the world has ever known for moving and managing information, but so far it has barely affected the mechanisms underlying how most payments work. The payments problem has proven far more stubborn than the problem of moving other kinds of information.

There are some things that make money harder to manage than other information. A typical consumer payment will pass through multiple layers of intermediaries on its way to a recipient. A patchwork of systems run by banks, merchants, card networks, and payment processors must all coordinate. There need to be systems for managing compliance, fraud, theft, and aiding law enforcement.

These are problems that have been successfully managed for a long time, but in redundant and sometimes inefficient ways, within individual financial organizations. They could be managed more efficiently within a unified, modern system. The challenge is getting these various organizations to align around a single system.

The way to solve collective action problems is by creating new networks. As I’ve argued throughout this book, there are three options: corporate, protocol, or blockchain networks.

A corporate payment network would have the same problem as all corporate networks. As long as the network has relatively low market share and weak network effects, it would be incentivized to attract users, merchants, banks, and other partners. But once its network effects become strong enough, it would inevitably use its power to extract higher fees from the network and put in place rules that limit competition. Banks and payment providers are savvy about platform risk and, being aware of the possible consequences, avoid handing power over to corporate networks if they can help it. (These companies did cede a lot of power36 to Visa and Mastercard, but that was back when Visa was a nonprofit and Mastercard was an alliance of banks; both payment-processing networks have since become independent, for-profit companies in moves reminiscent of Mozilla and OpenAI.)

A protocol payment network would present two challenges. The first would be recruiting people to build the network, since protocols have no inherent way to raise money and hire developers. The second problem would be feature limitations. Payment networks need to keep track of transactions, which means they need to maintain databases. Protocol networks have no core services and therefore no way to administer neutral, centralized databases.

Blockchain networks offer the benefits of corporate and protocol networks but without the limitations. Blockchain networks can raise money to fund developers, and they can store payment records in their core software, which acts as a shared ledger. They can run rules that ensure regulatory compliance. They have built-in audit trails to aid law enforcement. They also have low take rates and predictable rules that give developers incentives to build on top of them. All of these pros should be familiar by now.

By creating a neutral layer that can raise money, maintain shared data, and make strong commitments to users, a network like this would solve both technical and coordination problems that plague other payments networks. Blockchain networks can make payments a public good, analogous to a public highway system that spurs commerce and development in the physical world. Private companies would still play a role developing new financial products, but they would build these on top of credibly neutral blockchains. In any tech stack, the optimal design is a mix of private and public goods. In finance, it makes sense to have the payment layer be a neutral public good. (In the “squeezing the balloon” framework, the payment networks should be the thin part of the balloon.)

It’s possible a system like this could be built on top of Bitcoin. Bitcoin is a neutral, permissionless system. The original Bitcoin white paper describes it as an “electronic payment system,” but Bitcoin’s success with payments has been held back by high transaction costs and volatile prices. The high costs are due to a limited supply of block space—that is, how many transactions can fit in a given block. A number of projects building on top of Bitcoin are trying to remove these limitations. The most prominent of these is Lightning, a transaction network layered on top of Bitcoin that has higher capacity and therefore lower costs. Price volatility might still be an issue, but faster settlement times can mitigate this.

Ethereum offers another option. Systems built on top of Ethereum, like so-called rollups, also lower transaction costs and improve latency. People can use a dollar-pegged stablecoin like USDC to avoid price volatility.37 Sending money with USDC on Ethereum is usually faster and cheaper than using bank wires. While the transaction fees are still too high to handle smaller, everyday payments, this should improve as more scaling solutions come online, boosted by Ethereum’s platform-app feedback loop.

A global payments system would have multiple benefits. The first would be fixing problems with existing payments systems. Credit card payment fees are low relative to other internet fees, but still add unnecessary friction. The fees are even higher on international remittances, which can act as a regressive tax on lower-income people who send money to family members abroad. Any internet retailer will also tell you it is difficult to handle international payments, especially when they involve developing countries.

These problems are similar to ones that affected phone calls and text messages prior to smartphones. Users had to pay for calls by the minute and texts by the message, and international fees were high. The problems were fixed when applications like WhatsApp and FaceTime created new networks to replace older networks. A new global payment network could do the same for money.

The second benefit would be new applications that weren’t possible before. If transaction fees were low enough, micropayments could become possible. Users could pay small fees to read news articles or access pieces of media. Music royalties could be paid to rights holders using easily audited, blockchain-based payment receipts. Computers could pay one another programmatically for data, computing time, API calls, and other resources. Artificial intelligence systems could reward content creators who contribute to their training data sets, as we’ll cover more ahead.

Micropayments have been discussed for decades, and even tried at times, but have never worked. The hurdle has mainly been transaction costs. Some industry practitioners also argue that they ask too much of users. Each of these obstacles is surmountable, though. More scalable blockchains could address the transaction costs, and rules-based automation could lower the cognitive overhead. One day users may even be able to set a budget with some simple rules and leave it to “smart” wallets to disburse the payments.

The third benefit would be composability. Consider the composability of digital photos stored in standard file formats such as GIFs and JPEGs. These files can be seamlessly integrated into almost any application, resulting in a wave of innovation around photos. Some innovations are creative, like filters and memes. Some are services, like Instagram and Pinterest. Okay, now imagine a fictional universe where every photo is controlled, via APIs, by corporate networks. In this world, photos can be used only in ways some companies allow. The API providers would be the gatekeepers, controlling what users and developers can do. They would have an incentive to lock down the photos and stifle competition. That’s how money on the internet works today.

A blockchain-based system would make money remixable and composable, as digital photos are today. Or even better, it would turn money into open-source code. Making finance composable and open source is exactly the aim of DeFi networks, which perform the same functions as banks and other financial institutions but do so using blockchains. The most popular DeFi networks have handled tens of billions of dollars in transactions over the last few years. During recent market volatility, when many centralized organizations failed, DeFi networks stayed up and running.38 Users can inspect DeFi code to confirm their funds are safe. They can retrieve funds with a few clicks. These systems are simple, transparent, and credibly neutral—traits that also mitigate the risk of discriminatory practices.

Critics accuse DeFi of being overly self-referential, an internal micro-economy that doesn’t touch the outside world. There is some truth to that criticism. DeFi can operate only on money that is composable, which today limits its scope to money held on blockchains and limits its appeal to a relatively small subset of internet users. If the internet had a composable money system, the concepts DeFi pioneered could be scaled from micro to macro.

Finance has always been centralized, run mostly by for-profit companies, but it doesn’t have to be. Blockchain networks can make financial infrastructure a public good, upgrading the internet from handling bits to handling money.

Artificial Intelligence: A New Economic Covenant for Creators

The internet operates on an implicit economic covenant. Content creators, like writers, critics, bloggers, and designers, whether they are independent or part of organizations, publish work in the understanding that content distributors, like social networks and search engines, will reward them with attention. Creators bring supply; distributors bring demand. That’s the deal.

Google search exemplifies the covenant.39 Google crawls the web, analyzes and indexes content, and shows snippets of its findings in search results. In return for indexing and excerpting content, Google sends traffic back to the content providers through its ranked links. This arrangement enables content providers, like news organizations, to make money through advertising, subscriptions, or whatever business model they’ve chosen.

When this relationship began in the 1990s, many content providers didn’t foresee the stakes. Search engines took cover under fair-use exemptions in copyright law while content providers took a hands-off approach. Over time, as the internet grew, the balance of power between the two sides grew more lopsided. A surplus of content filtered through just a few distributors, giving distributors the upper hand. The end result: Google, as one example, now commands more than 80 percent of internet search.40 No content provider claims anywhere near that level of market share.

Some media businesses have tried to make up for their missteps. The media giant News Corp has been protesting Google’s free riding41 and trying to get more money out of the arrangement, including by lodging formal antitrust complaints, for more than a decade. (The two reached an ad-revenue-sharing agreement in 2021.) For most of its existence, the review site Yelp has been campaigning to rein in Google’s power,42 an effort that culminated in congressional testimony by Yelp’s CEO, Jeremy Stoppelman:

The problem with Big Tech is they control the distribution channels. Distribution is the key. If Google is the starting place for all of the people that are tapping into the web, to the extent they get in front of consumers and block them from finding the best information, it’s really problematic, and that can stifle innovation.

With distributors in the way, content providers lost their leverage. Google grew so dominant in the 2000s that opting out of search results became unviable. If individual companies like Yelp and News Corp opt out, they lose traffic, and their competitors fill in the gaps.

If content providers had the foresight to see this coming in the 1990s, they might have preemptively coordinated to take collective action. If they had, they might be in a stronger position now. Today, content providers are too fragmented to wield any individual power, and they’re not collectively organized. (A few savvy ones did see the endgame coming: the South African newspaper publisher Naspers became an internet powerhouse43 by pivoting its business from news production to internet investing.)

Distribution came out on top. Google made the bulk of the profits from the arrangement. The search giant knew its relationship with content providers was symbiotic, and also faced regulatory pressure, so it let enough money flow back to publishers to allow many to subsist. But the settlements and deals struck over the years are piddling compared with Google’s windfall.

Occasionally, Google breaks the covenant.44 One of the worst things that can happen to a website is “one-boxing,” when Google extracts a site’s content and places a summary at the top of its search results so users no longer need to click to get an answer. Searches related to movies, lyrics, or restaurants are commonly one-boxed. For startups that are dependent on Google for traffic, one-boxing is a death sentence. Sadly, I’ve seen this happen to a few companies I’ve been involved with. Traffic evaporates overnight and, with it, revenue.

Artificial intelligence has the potential to take one-boxing to its logical conclusion. New AI tools are already generating and summarizing content, obviating the need for users to click through to the sites of content providers. OpenAI’s release of its superpowered chatbot, ChatGPT, provides a preview of this future. You can ask the bot for a list of restaurants to visit, or to summarize a news event, and it will give you a self-contained answer—no clicking to other sites required. If this becomes the new way to search, AI could one-box the entire internet, thereby breaking the multi-decade covenant between search engines and the content they index.

Recent AI products have produced incredible results. From large language model bots to generative art systems like Midjourney, AI is improving at a rapid and, perhaps, exponential rate. The next decade in AI should be exciting. New applications will increase economic productivity and improve people’s lives. But advances in AI also mean we will need new economic models for content providers.

If AI systems can answer queries, this could replace most uses for search engines as well as the need to click through results to find content on websites. If an AI system can instantly generate an image, why go searching for images by human artists to cite or license? If AI can summarize news, why go read primary sources? AI systems will be a one-stop shop.

Most current AI systems have no economic model for creators. Consider AI image generation. Generative image systems like Midjourney feed hundreds of millions of captioned images into large neural networks to train them. The neural networks learn how to take captions as input and generate novel images that fit those captions. The results are often hard to distinguish from original human-made art. Despite learning from data from all over the internet, these systems generally neither compensate nor credit their sources. AI companies say these systems simply learn from the input images and the outputs do not infringe on copyrights. In their view, the AI is like a human artist45 who is inspired by other paintings to create an original work of art.

This might be a perfectly reasonable stance under existing copyright laws (there will likely be court cases and possibly legislation to sort that out). But, in the long run, we are still going to need an economic covenant between AI systems and content providers. AI will always need new data to stay up to date. The world evolves: tastes change, new genres emerge, things get invented. There will be new subjects to describe and represent. The people who create content that feeds AI systems will need to be compensated.

There are a few possible futures. One extends what current AI systems are already doing: “We will take your work, use it, and show the output to other people with no attribution or traffic back.” This behavior will incentivize creators to remove their work from the internet or put it behind paywalls so AI can’t train on it. We’re already seeing many internet services curtail their API access46 and enter lockdowns in response.

Maybe AI systems could fill in the gaps by funding their own content. This is already happening today with “content farms”47—buildings full of workers who are instructed to create specific content to supplement AI training data.48 This might work well for the AI systems, but it seems like a depressing outcome for the world at large. Machines direct progress, and humans toil like cogs.

A much better outcome would be a new covenant between AI systems and creators that encourages deep, authentic creativity over content farming. The best way to establish a new covenant is by designing new networks that mediate the economic relationship between AI systems and content creators.

Why do we need new networks? Couldn’t a new covenant evolve organically through the choices individual creators make to opt in or out of AI training data?

We learned this lesson the hard way with search in the 1990s. Web standards groups provided a way for websites to exclude themselves from search engines through the “noindex” tag, part of the robots.txt standard. Content providers learned that when they opted out and others didn’t, they lost traffic and gained nothing in return. Individually the websites had no power. The only way they would have power is if they organized and bargained as a group, which they never did.

An opt-out solution to AI would lead to the same outcome. Other content providers would fill the gaps, and whatever is left would be filled by content farms. Indeed, the problem is worse than with search because it’s hard to restrict the flow of loosely inspired ideas and imagery. Elements of the opted-out content will seep into the opted-in content, which will likely be enough for the AI systems to get what they need. If creators act alone, the AI will get what it wants, one way or another.

Blockchain networks could be the foundation for a new covenant. Among other things, blockchains are collective bargaining machines. They are perfectly suited to solve large-scale economic coordination problems, especially when one side of the network has more power than the other side. Blockchains have fixed rules, low take rates, and incentives for builders. The network governance could be jointly managed by creators and AI providers to ensure the network stays true to its mission.

Creators could set terms and conditions for using their work, backed up by software-enforced rules and copyright restrictions focused on commercial uses, including AI training. The blockchain would enforce an attribution system that allocates portions of revenue generated by the AI systems back to the creators who contributed to their training. AI companies would face a binary choice—accept the terms of the collective group or not—instead of using their leverage against individual creators. It’s the same reason labor unions bargain collectively with employers. There’s strength in numbers.

Could someone design a system like this using a corporate network? Yes, and someone probably will. But this will lead to the usual problems with corporate networks, including the attract-extract cycle. The corporate owner will eventually use its leverage to extract fees and implement self-serving rules.

The internet I would like to see is one where people are encouraged to be creative and can make a living that way. If people create things and put them on the open internet, they make the internet better. AI puts human creators at the beginning of the creative pipeline instead of the end of it. Shouldn’t creators get paid49 for being part of the process regardless of where they fit in? Plenty of money flows through search engines and social networks, more than enough to send some back to the people who created the content that makes search and social tools useful in the first place.

Everyone using the internet should ask themselves, if I’m doing something valuable, am I getting paid for it? Often, the answer is no. A few large companies have concentrated bargaining power thanks to the corporate network model. They dictate the economic terms for everyone else. It’s harder to shift the balance of power in mature categories like search and social where the lock-in is strong. For new categories, like networks that mediate the economics of AI, there’s an opportunity to start from scratch.

The time to address this is now, before the market structure is settled. Will content farms feed AI? Or will machines and creators happily coexist? Do machines serve the people, or will people serve the machines? These are key questions in the age of AI.

Deepfakes: Moving Beyond the Turing Test

In the 1968 novel Do Androids Dream of Electric Sheep?, a bounty hunter named Rick Deckard hunts robots. A major plot point of the book, which would inspire the classic sci-fi film Blade Runner, involves Deckard’s attempts to tell “replicants,” or rogue AI, from humans.

If life imitates art, then art is now imitating life. Androids walk among us, virtually: AI makes it easy to create “deepfakes,” media that looks and sounds real but is actually generated by machines. A deepfake video might show politicians, celebrities, or even ordinary people saying something they didn’t say, or a fabricated version of news events that feed conspiracy theories. Video often serves as a ground truth on an internet already awash with conflicting interpretations of events. Deepfakes make video no longer trustworthy.

One proposal to fight deepfakes is to try to contain AI through regulation.50 Some proposals call for a government certification process51 so only approved organizations can offer AI services. A number of AI and tech leaders, including Elon Musk and Yoshua Bengio, a pioneer of modern AI, signed a petition calling for a six-month pause on all AI research.52 The United States and the EU are in the process of developing comprehensive AI regulatory frameworks.53

But regulation isn’t the answer. No one can put the generative genie back in the bottle. Neural networks, the core technology behind modern AI, are an application of mathematical ideas that cannot be unlearned; linear algebra is here to stay, whether government officials like it or not. Open-source systems can already create convincing deepfakes, and these will continue to improve. Other countries will continue to pursue the technology too.

Regulatory restrictions will just entrench power in big companies that already have advanced AI. They will anoint haves and exclude have-nots. Burdensome rules will hold back innovation, and users will suffer as Big Tech tightens its grip even further, exacerbating the problem of internet consolidation.

Regulation also won’t solve the real problem: the internet’s lack of an effective reputation system. Instead of holding technology back, a better solution is to push it forward. We should build systems that allow users and applications to verify the authenticity of media. One idea is to allow “attestations”—claims backed up by cryptographic digital signatures—on blockchains where users and organizations can vouch for individual pieces of media.

Here’s how that might work. The author of a video, photo, or audio track could digitally sign a media identifier, called a hash, saying, “I created this content.” Another organization, like a media company, could add to that attestation by signing a transaction that says, “I attest that this content is authentic.” Users could identify themselves in the signatures by cryptographically proving control of domain names (for example, nytimes.com), newer identifiers tied to blockchain-based naming services like Ethereum Name Service (nytimes.eth), or usernames on older identity systems like Facebook and Twitter (@nytimes).

The advantages of storing media attestations on blockchains are threefold. Transparent and immutable audit trails: Anyone can examine the full content and attestation history, and no one can alter it. Credible neutrality: If a company controlled the attestation database, it could leverage this control to restrict or charge for access. A credibly neutral database de-risks the platform and ensures a widely accessible public good. Composability: Social networks could integrate the attestations, displaying verification check marks on media that trusted sources have authenticated. Third parties could build reputation systems that evaluate the track records of attesters, assigning trust scores. An ecosystem of apps and services could develop around the database, helping users differentiate between real and fabricated content.

Attestations can also address the proliferation of bots and “counterfeit persons.” AI is going to make bots so sophisticated that users can’t distinguish between real and fake people. (We’re already beginning to see this happen.) In this case, the answer is to attach attestations to social network identifiers instead of pieces of media. For example, The New York Times could attest that the @nytimes handle on a new social network is controlled by the same organization that controls the website www.nytimes.com. Users could examine the blockchain, or rely on third-party services that do, to verify the authenticity of these attestations.

Such authentication systems would help defeat spam and impersonators. Social media services could display verification check marks for usernames that have credible attestations. They could offer settings that let users filter out bots (“only show me people who have signed attestation from credible sources”). Verification check marks shouldn’t be bought, doled out through favors, or subject to the biases of corporate employees. They should be objectively verified and auditable.

One of the lessons from the last era of the internet is that if a service needs to be built, it will probably get built—if not as a public good, then as a private good. When users needed a reputation system to sift through websites, Google ended up building that system, originally called PageRank and today a set of proprietary rankings. Had blockchains been around, a reputation system like this could have been built as a public good, owned by everyone instead of one company. Website rankings would be publicly verifiable, and third parties could build services on top of them.

Turing tests no longer distinguish real people from bots, and people can no longer tell real from fake media. The right approach is a credibly neutral, community-owned network—a blockchain network—that makes authenticity a trusted internet primitive.