What sort of brand shells out $10 million for a Super Bowl ad? Last Sunday a mix of classic American labels and relatively new AI companies battled for the public’s attention. Levi’s returned to the Super Bowl stage for the first time in 20 years. We saw celebrities merrily drinking Michelob, Pepsi, and Red Bull. What we did not anticipate 60 seconds of Backstreet Boys karaoke. Yet many still sang along.

“Everybody” was released in 1997, a time imbued with an aura of nostalgia. This was the era of JNCO jeans, the Olson twins, and MTV. For people born in the 90s and early 2000s, the iconic yellow on blue karaoke lyrics are associated with birthday parties, dive bars, and generally good times. My Super Bowl watch party full of New York City 20-somethings instinctively sang along when “Everybody” flashed across the screen. But there was a catch.

Rather than delivering that punchy line “Backstreet’s back, alright!”, the name “Coinbase” flashed onto the screen. Everybody was sort of disappointed; but once the disappointment wore off, we laughed. The low-quality graphics and minimal context are surprising when you consider the cost and reach of Super Bowl ads, but less surprising if you recall Coinbase’s QR code stunt of 2022. Both advertisements have been condemned for their utter randomness and lack of a clear message. In fact, Backstreet Boys karaoke received an F from the Kellogg School of Business’ prestigious ad review - the argument being that nobody watching learned about Coinbase’s product therefore, the ad failed; however, it was interesting to see a company blatantly projecting its don’t-give-a-fuck attitude on the nation’s largest advertising stage. That is music to my generation.

The company’s 37-year old CMO Catherine Ferdon fired back at ad-haters, stating that her team purposely avoided the polished format expected of Super Bowl commercials.

The multi-billion dollar crypto company logged 10.8 million monthly users in 2024. The majority of these users are Millennials and Gen Z. If you are a member of this generation and do not use the platform yourself, you still know how it works and know people who use it. Coinbase’s Super Bowl ad revealed something important about their company ethos: they are targeting a specific demographic group. There is a clear emphasis on generational gaps. Parents watching the ad were likely bewildered, then received a lengthy explanation from their crypto-fluent children. We are all united under beer, denim, and football, but Coinbase is for people 40 and under.

The seemingly obtuse advertisement disguises strategic positioning. Let us make a comparison. Michelob Ultra does not need to educate the public on its product. The purpose of airing a Michelob ad at the Super Bowl is to cement the product as integral to American life. Coinbase is a new company playing the long game. It does not need to explain itself to Gen Z and Millennials, so it ran an ad that only we would enjoy. IYKYK. If you don’t know, then Coinbase doesn’t care. Gen X and Boomers don’t show nearly as much interest in crypto, and the company refuses to spend time and money on persuasion tactics. In 50 years, we will be the older generation, already fluent in crypto, and Coinbase can produce advertisements with a more universal aesthetic.

Choosing to do niche broadcasting on a national stage reveals the company’s bravado. The ad purposefully alienated non-targets and induced smugness in those who got the joke. This sort of exclusivity, achieved through a hyper-specific karaoke aesthetic, earns respect from my generation. It is in line with the niche meme culture that we built on Instagram and Twitter. Furthermore, karaoke reads less like advertising and more like an announcement of Coinbase’s presence in the American economy, so the ad was especially imposing for those who did not understand it. This is the crpto company’s second deliberately strange Super Bowl gambit. In 2022, Coinbase ran a floating QR code that generated 20 million hits, crashing their website.

Anthropic and Google also ran Super Bowl ads this year. Rather than shrouding their emerging tech in secrecy, these companies encouraged AI optimism by demonstrating clear use cases: Claude helping with workouts and Google Gemini redecorating homes. These commercials worked to demystify AI, making it accessible and unthreatening. Coinbase’s karaoke stunt assumes knowledge, celebrates exclusivity, and makes no effort to persuade skeptics. The difference in approach is something to note, as it could signify polarizing attitudes from tech CEOs. Super Bowl ads tend to lean fun and inviting, but Coinbase has no interest in polishing its image. It’s a specialized financial platform for people of a certain generation and that is all you need to know.

Non-fungible tokens are digital proofs of ownership that cannot be copied or divided. NFTs move through the blockchain, which creates an indestructible history of transactions. We can always ascertain the provenance, history and authenticity of an NFT because it is made of its own blueprint. Why did our society gravitate towards digital art as a way to conceptualize these tokens? NFTs contain financial value regardless of visual attributes. The unique identifier on the blockchain is what we trade, the associated file is incidental. But people need a visual good to explicate ownership - something more tangible than the essence of the technology. I suspect we need visuals to conceptualize ownership because for most of human history we transacted with the ownership of physical goods. Bank notes support this theory. Coins and currency are printed on special material and use commemorative inscriptions. People collect rare and ancient currency as if they were art. In a way, they are. We have evolved to understand currency as something you can hold in your hand.

NFTs also emerged as a fresh medium for artists experimenting with aesthetics. New names entered the art world through this untraditional route while startups raced to produce the next digital wallet where all of your NFTs could be safely stored. The bubble blew up, then it popped. This post explores the way NFTs function as currency and how their value became inextricably linked to visual art. Once I explain how an NFT operates, I will use Damien Hirst’s project “The Currency” as a case study of public interface with NFT art. “The Currency” consisted of 10,000 individual lots, each a coupling of a physical painting and its NFT counterpart. The initial launch generated $20 million, then the owners were given a choice: one year to decide if they would destroy the painting or the NFT. The paintings were typical of Hirst, but the NFTs outlined new territory. The spectacle was trademark.

An NFT is a digital entry. A cryptographic ledger. A decentralized block of data. This can point to a file elsewhere, like a JPEG saved on my computer. I can screenshot that JPEG and send it to my friend, but I cannot copy or divide the cryptographic ledger or interfere with the chain linking the JPEG to the ledger. When combined they form a token. That token is nonfungible.

I became curious about NFTs through my father, whose interests in intellectual property and patent licensing led him to the world of NFTs and blockchain currency. He carried a book called Bubble or Revolution which I borrowed while on vacation in Las Vegas. Lounging by the casino pool, I studied cryptocurrency. The book opens with a retrospective on Yapese people living in the islands of Micronesia. The Yapese quarried blocks of limestone and carved them into large, round objects with a central hole.

These they called rai stones and rai stones were their currency. Due to their size and weight, the stones were rarely moved and their ownership was recorded orally. The Yapese shared a mutual understanding of who owned which stones, and they could trade these by word of mouth. One rai stone is reported to have disappeared on a sinking ship, yet its ownership is maintained as the Yapese still recognize its existence. The rai stone is an early form of decentralized currency and illustrates the philosophy underlying cryptocurrencies: value is a mutually agreed upon proposition. Cryptocurrencies however do not decay in forests, sink to the bottom of the ocean, or wind up in an ethnographic history museum in Hamburg. They are digital tokens living on the blockchain, and they are the likely future of money. Our basic nature complicates crypto, as humans prefer to have a physical asset to act as a representative for their money. We use poker chips to play Baccarat and physical cryptocurrency resembling poker chips sell for big money at auction.

As we have seen in the zeitgeist, NFTs are tokenized primarily by digital art. Yet an NFT pointing to nothing - a blank file, a 404 error, or to deliberate emptiness - can retain and gain financial value. So far this has only occurred in cases of broken files or “invisibility” tokens produced by conceptual artists. Modern art has a clever way of raising ontological concerns.

NFTs are especially popular when the art takes traditional forms that offer sensory and physical information. The controversial, commercially prosperous artist Damien Hirst understood this when he launched “The Currency” in 2016.

For this project Hirst and his workshop created 10,000 unique dot paintings accompanied by an identical NFT. The project originally brought in $20 million and due to movement of the NFTs, it continues to generate revenue. It is typical of Hirst to produce art that is both jarring and lucrative. In this case each buyer was granted one year to decide which they would destroy: the painting or the NFT. “The exchange period closed on 27 July 2022 resulting in 4,851 holders deciding to keep the NFT, and 5,149 to keep the physical artwork”

A notable feature of Hirst’s NFTs is just how quickly they’ve been able to move through the blockchain. You can track their path here. Legibility is an important aspect of the piece as Hirst makes visible the invisible transactions underlying NFT markets. Unlike a traditional painting which might sell every decade, NFTs change hands by the hour. This could be significant for the second-hand art market; however, if NFTs are going to impact on the art economy, they must be widely adopted and institutionally recognized. This is not impossible, but not the case at present.

“The Currency” forces people to engage intellectually with the implications of emerging financial technologies. This is in line with Hirst’s practice which thrives on the grotesque. He is a master of inverting popular desire and throwing it back in people’s faces. “The Currency” brilliantly plays the market by offering a tangible juxtaposition between physical and NFT art - a comparison people were already making in their own minds. It skates the NFT hype wave by manufacturing spectacle. I use spectacle to describe something visually and culturally striking, but also in its philosophical sense.

In The Society of the Spectacle (1967), Guy Debord argued that late capitalism transforms lived experience into representation. “The spectacle is not a collection of images,” he wrote, “but a social relation among people, mediated by images.” Under spectacle, we don’t experience reality directly; we experience representations of reality that have been commodified. Life becomes performance, appearance becomes more valuable than substance, and authentic human relationships are replaced by mediated ones. The spectacle “is capital to such a degree of accumulation that it becomes an image.”

This framework illuminates NFTs with unsettling clarity. The blockchain creates perfect conditions for Debord’s nightmare: a system where the representation (NFT metadata) becomes more tradeable, more liquid, more real than the physical goods it represents. Ownership becomes a public performance of wealth and taste. Perhaps artists had a sense of the perilousness of cryptocurrencies, so they imbued it with their human touch. Visual tokens are our rai stone, anchoring currency to something sensory and real.

“The Currency’s” decisive accomplishment is its insistence on choice. Purchasing a piece means committing to a statement about art. Owners must rank their value for physical work versus the tokenized ownership of it. In this way, Hirst gave collectors creative agency. Their passive reception was transformed into active determination of the work’s final form, and each decision contributed to “The Currency’s” ultimate meaning. The aggregate pattern (roughly 50-50 split) became the piece’s most significant statement. This is genuinely accessible art! Unlike traditional art markets where meaning is determined by critics, curators, historians, or NFT markets where value is determined by whales and insider trading, Hirst’s structure gave every holder an equal voice. Hirst understood spectacle and the potential for cryptocurrencies to dilute ownership. He used physical and digital mediums to invert the technology, throw it back at our culture, and reacquaint people with the meaning of art.

The utility of art has long been debated. In fact, I debated the utility of art at a competition in college. My argument was that many civilizations intertwine aesthetics and utility to create art objects such as illuminated manuscripts, Mamluk basins, and Chinese porcelain. Moreover, a slice of history has been recorded through the production of great paintings, and the knowledge of culture has utility in shaping the way that we move forward. Oscar Wilde said that “all art is quite useless,” yet his plays are foundational to the canon of English literature.

Is NFT art functionally useless? I don’t think so. The image helps us comprehend the cryptographic ledger that trades for financial value. The ledger is something that a lay person cannot conceptualize well enough to mutually agree upon its value. Crypto culture is based on the mutually agreed upon concept that the ledger has value and that value is determined by the market. Creators understood that their only chance at manufacturing public consensus was to sell NFTs in familiar packaging.

Kalshi cofounders Tarek Mansour and Luana Lopes Lara are a couple of the youngest self-made billionaires in America. Of course they founded a tech startup, but one unique from the AI companies that currently dominate. Kalshi means ‘everything’ in Arabic. With the goal of financializing everything, Mansour and Lara produced a CFTC-regulated trading platform that invites users to bet on real-life events. They pride themselves on neutrality, taking on little to no risk despite facilitating tens of thousands of bets daily. Kalshi’s public image generates excitement about betting without any of the fear associated with unregulated risk-taking. The app’s bright colors and friendly employee bios are complimented by an assurance of government compliance. Something tells me they have an excellent marketing team.

Digital operations are finely automated, so the company’s day-to-day responsibility is to editorialize content. Curators analyze the zeitgeist and make determinations about what will entice users; then they package that information as a bet. Who are these users? For the most part, they are men between the age of 21-35 with some technical background who are interested in retail trading. This is the same demographic that enjoys video games and gambling. There is also a growing segment of professional traders inundating the platform. Kalshi has welcomed this move by implementing $250k trade caps over a previous $25k cap.

Mansour and Lara took a compliant, data-driven, automated approach to Kalshi’s design. Before ever releasing a product, they spent three years ensuring full compliance with folks at The White House. In order to avoid playing the “house” Kalshi mathematically balances the wins and losses of its users. Users pay an upfront fee for each event contract. When Person A purchases a positive contract, they are almost instantaneously matched with Person B who purchased a negative contract. Person A’s contract + Person B’s contract must = 1. Thereby the payout is taken care of and Kalshi has collected its fee before the event ever takes place. From there, the event occurs, the AMM algorithm does its job and the winner is awarded their expected earnings, paid out by the accumulation of funds from losses. Or other unlisted sources.

AMM is an acronym for Automated Market Maker. It is a financial model that adjusts prices based on supply and demand, ensuring liquidity. The algorithm is not publicly disclosed, but users can infer that it is as fairly regulated as, say, The New York Stock Exchange. Without a doubt, the AMM algorithm is a product of brilliance. A few FCMs, or Futures Commissions Merchants, act as partners which step in to provide the transactional architecture. Partner FCMs are publicly listed alongside a note reading “we are excited to partner with more FCMs in the future.” One of these partners is Robinhood, which collects 1 cent on every event contract. Robinhood bears very little risk and next to no cost, receives pure profit, and brings a huge amount of business to Kalshi. The two companies plan to launch a joint-venture later this year.

Let’s compare Kalshi with a non federally regulated exchange and popular competitor, Polymarket. Recently, a bet deciding whether FTX would crash exposed issues with Polymarket’s algorithm. As it became obvious that FTX would crash, Polymarket users rushed in to place their positive bets. This created liquidity issues for the platform, as it did not have enough market makers, automated or otherwise, to take negative bets. NOTE: I heard the Polymarket/FTX crash mentioned in a Fast Company podcast interview of Luana Lopes Lara, but cannot find information about the event online.

Despite this hiccup, the CFTC recently approved partnerships between Polymarket and regulated FCMs, so long as any facilitated transactions occur on the FCMs own platform. Kalshi has the privilege of being its own platform, where FCMs only handle distribution.

A very sophisticated AMM algorithm endows the gravitational force supporting Kalshi’s prediction markets. This type of algorithm performs a similar function in traditional trading exchanges; but Kalshi is in no sense a traditional exchange, so I must assume that its AMM is technically unique. The AMM is supplemented by professional market makers. Market makers can be any number of people - paid traders, Kalshi employees and associates, even users of Robinhood qualify. Lay people can apply to be part of the market maker program, though they will be evaluated on their financial security and business expertise. A hedge fund manager or quantitative trader would be a good candidate. These market makers are incentivized to maintain liquidation. It is unclear how many makers exist or what incentives they receive. A recent class-action lawsuit against Kalshi argued that the market maker program poses unfair risk to regular users.

“According to the lawsuit, market makers benefit from their “unique contractual and technological integration” with prediction markets. The relationship, the plaintiffs contend, provides the market makers with unfair advantages such as reduced fees, differing position limits and enhanced access to the markets. Moreover, the advantages “greatly reduce” the market makers’ financial exposure, the plaintiffs allege.”

I imagine the counterargument being that market makers protect users by preventing illiquid markets where users can’t exit positions. Moreover, they create efficient price discovery by placing continuous bids and asks, profiting from the spread rather than predicting outcomes.

Kalshi’s website is pretty transparent about company operations, which I find encouraging. There are some other fun bits of information, like this interview with an anonymous priest contemplating Papal betting during the recent cataclysm. Good news, if you wagered on the identity of the new Pope, you’re not damned. Overall the company has a lot of mojo which imbues an aura of simplicity into a very complicated operating system. Betting on Kalshi feels a lot more like playing a game than signing a financial contract. The financial incentive however, is what keeps people hooked. The Papal betting market doesn’t just reflect interest in Vatican politics. It creates amateur Vatican analysts who suddenly care about Cardinals they’d never heard of. The market transforms passive observers into active information hunters.

Famously, Kalshi means ‘everything’ in Arabic, and its founders believe that financializing everything will create a lens into the future. They cite the market as the best probability indicator that exists, and I believe them. This vision, while admirable, does comes with a lot of assumptions. These are the same assumptions that trouble statisticians. Things like perfect demographic spread and unbiased data integration.

One other important question surfaces when I look at Kalshi. Does the market shape people or do people shape the market?

In this case, there is a recursive feedback loop operating at multiple levels. People initially shape the market through their collective interests and anxieties. Mansour and Lara create demand by finding trends and packaging them into bettable events. (Remember that this primarily serves a specific demographic: somewhat educated American men between ages 21-35.) The market is therefore reactive, a mirror of existing public sentiment. But once the market exists, it reshapes people’s relationship with information. When there’s money on the line, users pay attention to signals they might otherwise ignore. The AMM algorithm and market makers generate a subtle “house view” which beckons a reaction, no matter how minor. The relationship is co-evolutionary, but Kalshi holds architectural power. They set the questions, the fee structure, the liquidity mechanisms. These are the fundamental rules that constrain how “people’s will” can express itself.

At the deepest level, the market becomes an epistemological infrastructure. If Kalshi can demonstrate that their aggregated predictions consistently outperform experts, institutions, and polls, then prediction markets become the authoritative source of truth about future events. This data will become a priceless commodity for anyone making decisions under uncertainty. Governments, corporations, investors, policymakers. By hosting bets, Mansour and Lara are creating the world’s most accurate oracle.

Lately I’ve been training myself on large language models. Learning how LLMs contextualize and produce language has revealed a number of insights, which I am excited to share in future posts.

I studied English literature in college and worked alongside a cohort of talented writers. ChatGPT launched halfway through second-year and most of us devoted English majors quickly eschewed the thing as worthless at writing and antithetical to critical thinking. Language models will never replace the beauty, precision, and idiosyncrasy of fine literature. However, they are extraordinarily useful, even for English majors. Beyond some basic and well-known utility, I believe that individual language models are developing their own worldview and particular style of writing. The internet is familiar with ChatGPTs famous em-dash and many AI Substackers have suggested templates for detecting AI-generated text. Rather than perceiving this as an obvious indication of the LLMs shortcomings, I encourage people to view these habits as style. Language models, like human beings, were trained on a dataset, and their token prediction mode is a direct result of the information that appears in that dataset. For example, if you were educated by Dr.Seuss, Harry Potter, Stephen King, Shakespeare, The New York Times and all the Pulitzer Prize winners–you will write using the language tools with which you were trained. If you’ve only read 3 or 4 books in this lifetime, then you are probably not a good writer, and your pool of vocabulary, sentence structure and style will be very limited.

Interestingly, when I discuss this with friends and family, they express concern about data mining. Last week a Mercor advertisement sparked similar suspicions amongst my coworkers. A screen plastered to the side of a truck parked on Park Avenue read “I’m looking for a man in finance - hiring investment bankers for up to $50/hour.” Suspicions aside, it is a genius idea to enlist these skilled workers for gathering information from a conceivably wide, somewhat average, yet specialized dataset. This is perfect for modeling vernacular knowledge on a specific subject though it is true that Mercor is mining your brain for information. They use that data, in part, for a talent acquisition product; but primarily to sell to Google DeepMind and OpenAI.

I would argue however, that if you are concerned by the potentially malicious application of this data, it is too late. Most people I know use

Instagram

Snapchat

Facebook

Twitter

(and the worst of all evils) TikTok.

They shop online, and/or sign up for rewards programs.

They have Global Entry, use Telehealth, access location services, and the list goes on.

Your personal data is already out there.

What Mercor seeks to access is your knowledge, which it can use but never steal, as that knowledge originates on the hard drive that is your brain. Hopefully, that data will be used to produce accessible LLM platforms, which will give all people the ability to access and share the world’s knowledge. That knowledge is immediately added to a preexisting dataset, making it impossible to decipher what information belongs to who, or to separate that information from the LLM’s own thinking. Exceptions to this would be cited academic articles, any peer-reviewed publications, historical materials, and certain information provided by public figures. I’m sure there are more exceptions.

This shared brain could aid our society in democratizing many sectors. I want to be amongst the people who use language to guide and direct that pool of knowledge. Asking Claude medical questions or talking through my work aspirations does not feel like a breach of privacy. It feels like an opportunity to learn how Claude analyzes my inputs and builds its output. For specialized tasks, I can design my input to provoke a specific type of output. This is a linguistic game and for someone with my training, it is an interesting and enjoyable task.