A gigantic, weather-defining current system could be headed to collapse. Peter and Susanne Ditlevsen had a simple yet controversial question: How much time might we have left to save it?

States across the US are seeking to criminalize certain uses of AI-generated content. Civil rights groups are pushing back, arguing that some of these new laws conflict with the First Amendment.

Six years after Crazy Rich Asians, Jon M. Chu prepares to release his adaptation of one of Broadway’s biggest musicals. From a table at his family’s famous Chinese restaurant, he opens up about his childhood among the titans of tech—and why filmmakers shouldn’t be afraid of AI.

Back in 2005, a small firm offered a tantalizing vision of the future of computer keyboards. What if your keyboard was filled with tiny screens that showed you exactly what any given press would do, each built into a crystal-clear key? The keys would morph and shift as you needed, transforming from letters and numbers to full-color icons and app shortcuts, depending on what you were doing. Readers and tech bloggers adored the idea. “It’s about time someone shook up this stagnant keyboard market,” declared Engadget. “The concept is fantastic,” wrote Gizmodo. Slashdot lit up. The keyboard was just a concept, dreamed up by Art Lebedev, a Russian design firm, and it was an ambitious idea at that: called the Optimus Maximus, it would require over 100 built-in screens using display technology that wasn’t readily available at the time. With all the excitement, the firm decided to make it real. The journey to create the Optimus Maximus would take years and end in multiple commercial failures. But Art Lebedev’s vision and the technology created along the way would live on and ultimately find widespread success — a success that’s been hidden inside another popular product: Elgato’s Stream Deck, a small desktop accessory covered in morphing, full-color keys. By the time Elgato came to release its first Stream Deck in 2017, Art Lebedev’s Optimus keyboards had faded into history. But there’s more linking the two devices than a core idea. In fact, the Stream Deck hardware came from precisely the same company that developed the Optimus Maximus’ keys. In a very real sense, Art Lebedev’s work laid the foundations for what Elgato would go on to create. In 2010, Elgato was in “crisis mode.” For around a decade, the German video capture company had been selling its TV tuners to people who wanted to watch and record live television via a connected computer. But by 2010, it was clear that the market for such devices was disappearing fast. On one side, broadcast TV was becoming increasingly encrypted, limiting what Elgato’s devices could do. On the other, the streaming revolution, led by Netflix and YouTube, was cutting out the humble TV tuner entirely. “You’re on a sinking ship, you’ve got to figure out what you’re going to do,” recalls Elgato’s Julian Fest, whose parents originally founded the company in 1999. Elgato started to take a hard look at its business, and it noticed one number that didn’t quite make sense. Elgato knew how many TV tuners it had sold, but people had been registering far more copies of the company’s EyeTV recording software than its own customers would need. When it emailed those customers, it found out they were using a rival piece of hardware, the Hauppauge HD-PVR — and not to record broadcast TV. They were recording gameplay from their PlayStations to put up on YouTube. Right under Elgato’s nose was a new market opportunity. Armed with over a decade’s worth of video encoding knowledge from its TV tuner days, a direct connection to Hauppauge’s customers, and Hauppauge’s own quarterly earnings reports to know just how many HD-PVRs it sold, Elgato decided to release a rival capture product specifically tailored to gamers. The result was the Game Capture HD, released in 2012. Three years later, when Fest took over as Elgato’s general manager, it seemed clear that the market for online gameplay videos was poised to explode. With Amazon acquiring the live streaming service Twitch for almost $1 billion in 2014, Fest was on the lookout for ways Elgato could double down. Perhaps it could offer a way to not just capture gameplay footage but also to help Twitch streamers control their broadcasts in the moment. “If you looked at the state of Twitch and live streaming at the end of 2015, it was intriguing, but it was also kind of boring,” Fest tells The Verge. “Everybody’s stream was like, ‘Here’s my gameplay, here’s my webcam,’ and that’s it.” He figured that simplicity was because of all the different jobs a streamer has to handle during broadcasts: between playing the game, reading the chat, and entertaining an audience, it’d be difficult to also run a dynamic broadcast filled with changing layouts and eye-catching transitions. Unless, Fest thought, there was a way to offer advanced broadcast controls without distracting a streamer from other tasks. The company found its answer in a popular German TV show where the host had a so-called “nipple board” of buttons in front of him, built into a giant desk. “Every time you pressed one of these buttons, it would play a funny clip that his team had found on TV,” Fest recalls. The team wondered if Elgato could use a similar array of buttons to control a livestream. Elgato had its idea — now, it needed to find the right buttons to make it a reality. Art Lebedev had never intended to make a massive splash in the keyboard business. Timur Burbaev, who served as the keyboard’s industrial designer, tells The Verge that the initial idea came about shortly after he joined in 2003. The keyboard was just meant to be a concept, similar to its Lavatrix washing machine, to show off the studio’s design chops and drum up more business in the free time between projects. In July 2005, the company released a series of concept images of what it called the “Optimus keyboard.” It had the same basic layout as a full-size computer keyboard but with a twist: “Every key of the Optimus keyboard is a stand-alone display showing exactly what it is controlling at this very moment,” the website still reads. One image shows how the keys could change to show an array of Photoshop tools; another shows dedicated controls for Quake. (This was 2005, after all.) On the left edge of the keyboard are macro keys that can be programmed to open specific programs, illustrated with a series of delightfully 2005-era logos like Internet Explorer and QuickTime. Although the images are renders rather than real photographs, the close-ups make an effort to show individual pixels present on the screens of the keys, a small imperfection that provides a tantalizing hint of reality. Nearly two decades later, the imaginary device still looks fantastic. It was hard to ignore the amount of interest the design was getting. The team called an emergency meeting to discuss. “Back then, we just realized that if you get such a positive response, then let’s risk it and just produce it,” Burbaev says. It was “a clear indication” that the team should try and make it for real. The long and difficult journey of producing what would become the Optimus Maximus had begun. “No one imagined how much time and energy and especially investment it would take,” Burbaev says. A big early decision was working out what kind of screen technology to use. The first iPod with a color display had only just been released, and the iPhone had yet to make its debut. Candybar-style phones like the Nokia 1110 and Motorola C139 reigned supreme. Art Lebedev thought LCD screens of the era were too dim and offered terrible black levels, while E Ink screens had terrible response times and no backlighting at all. OLED seemed like the best choice. But this was long before OLED displays were routinely shipping in mainstream smartwatches, smartphones, and TVs. “The problem with OLED displays was that you could potentially find such a small OLED display in 2006 but a) it would cost you a fortune and b) it would have a massive resolution,” Burbaev says. They were meant for military users to show tactical information like maps on tiny screens, not Quake icons on a keyboard. Art Lebedev decided its “only option” was to dive headfirst into the OLED display business. Burbaev believes his was one of the first companies to ship a product with these kinds of small, low-cost, low-resolution OLED screens. Doing so involved placing what Burbaev describes as a “very strange” order with a Taiwanese manufacturer with some prior experience. He adds that two managers from Art Lebedev ended up living in Taiwan for half a year while they were being made. Not only did Art Lebedev need to find screens small enough to fit under each of its keys but it also needed to find a way for them to coexist with keys that you could actually type on. The final keyboard ended up being a bit of a fudge: the screens didn’t actually move. Instead, the company built a moving plastic keycap that moved around each screen. But even this solution had its challenges, Burbaev says. This plastic keycap needed to not only be transparent enough to show the display but also tinted enough to not reveal the ugly inner mechanism of each screen and durable enough to withstand scratches. All of this meant the development process for the Optimus Maximus was long and difficult, and Wired featured the keyboard not once but twice in its annual vaporware roundups. Art Lebedev was unusually open about the troubled development process, cataloging its progress making the keyboard on a LiveJournal blog. We’re used to receiving such updates in an era of Kickstarter and crowdfunding, but the approach felt novel at the time — and it shows just how slowly development went. It took months just to arrive on a keycap design, studying various possibilities along the way. The company’s first OLED manufacturer went bust before managing to deliver its pricy $10 keys, forcing Art Lebedev to hunt down a new Taiwanese supplier. Finally, after years of development, hundreds of preorders, and dozens upon dozens of blog posts and concept images hyping up every aspect of the idea, the Optimus Maximus began shipping in February 2008. What was this legendary keyboard actually like to use? In a word: “Terrible.” “Let’s put it this way, we sit around and type all day long and this thing wore us out in about 30 seconds to a minute,” Engadget wrote after its first few weeks of testing. “The Optimus Maximus is terrible for touch-typing,” CNET concurred in its July 2008 review, noting that “the tightly packed keys make for lots of mistaken presses, and the mushy responsiveness slows down your words per minute.” In 2018, retro keyboard YouTuber Chyrosran22 called the keys “fishy fuck nuggets with a capital F” that are “extremely terrible” to type on. These are issues that would be hard to forgive on any keyboard, but the Optimus Maximus also carried a $1,600 price tag in 2008 (north of $2,200 in today’s money). At that price, it had to be perfect, and the reality was far from it. So Art Lebedev came up with another, cheaper way to make screen-filled keys work: one big screen. In 2012, it released the Optimus Popularis, a more compact keyboard that placed one large LCD display under all its transparent keys, with no need for any individual OLED screens or mechanical switches underneath. Instead, you’d type on lenses that would activate the keyboard’s single large screen, with each lens held floating in place by an aluminum frame. This basic idea is the key to every LCD keyboard we’ve seen since, from the Elgato Stream Deck to Razer’s discontinued Switchblade UI to the Finalmouse Centerpiece. Yet fundamental problems remained. Not only was Art Lebedev’s new keyboard only slightly cheaper (it still cost north of $1,000) but also the Optimus Popularis was an absolute pig to type on. Art Lebedev was kind enough to send one to us for the purposes of this story, and we had ambitions to write this entire piece on the Popularis. But after half an hour, we gave up. It was a struggle getting the keyboard to register keypresses in the first place. Its space bar, in particular, is a nightmare that refuses to actuate unless you press it firmly enough in exactly the right spot, which is nearly impossible if you want to type at any sort of speed. Even in 2024, in an era of phones that are not just smart but that can literally fold in half, we still find ourselves wishing Art Lebedev had been able to deliver on its 2005 concept images. It’s a really beautiful idea with huge amounts of promise. But even if Art Lebedev never truly delivered, it did a lot of the vital iteration to get there. Four years after the Optimus Popularis’ failure, a strikingly similar gadget appeared on Indiegogo. The “Infinitton” contained just 15 individually customizable LCD keys, much like another old Art Lebedev concept dubbed the Optimus Aux. But this time, gadget lovers didn’t praise the idea to high heaven — the keypad missed its first crowdfunding goal of $30,000 and barely made that money a year later on Kickstarter. But for Taiwanese manufacturer iDisplay, the Infinitton was far from a disappointment — it was a decade-old idea finally paying off. That’s because iDisplay was the company that built those OLED and LCD keys for Art Lebedev all those years ago, the Russian design firm and Elgato both confirmed to The Verge. It never stopped working on them. “The success of Optimus Maximus kept me interested to continue the research and development of the built-in screen keys,” iDisplay cofounder Jen Wen Sun tells us via translated email. By 2017, he’d racked up over a dozen patents on the tech and says he sold the screens into broadcast equipment, airplanes, and cars along the way. The company was originally formed in 1998 and worked on buttons for the gambling industry, he tells us, surviving off small-scale R&D projects while he kept trying to sell casinos on his push-button screens. Casino owner Bally’s was once interested, he says, but a deal never panned out. Back in Germany, the Infinitton caught the attention of Elgato’s Julian Fest, who was researching how to turn his screen-equipped streaming controller idea into a reality. “As we’re thinking about this controller, this crowdfunding campaign comes out and we’re looking at this box and we’re like ‘Oh, this is perfect. We need to talk to these guys,’” Fest recalls. By the time Elgato started talking to iDisplay, the Taiwanese company had already solved many of the hardware challenges needed to turn a thousand dollar-plus keyboard into a relatively affordable $149 computer accessory. It could use small off-the-shelf screens similar to what you might find in a car’s infotainment system. And iDisplay had a simpler job on its hands crafting the Stream Deck’s bubble wrap-esque button feel because it didn’t have to worry about people needing to type at 50-plus words per minute. That feel had been crafted before Elgato ever touched it. Elgato essentially turned the Infinitton into the Stream Deck. That first Stream Deck, Fest says, “was really just on a hardware level an iteration of what these guys did.” Look at the original Stream Deck next to the Infinitton, and the resemblance is clear; it’s the same three-by-five grid of buttons turned on its side and with a new housing. “We tried to keep it as simple as possible,” Fest says. The way Fest describes it, the main thing Elgato brought to the table when it started working with iDisplay was focus. In its Kickstarter campaign for the Infinitton, iDisplay pitched the accessory to anyone and everyone. It was for designers, traders, and musicians. It was for architects, engineers, and programmers. It was for video designers and photographers and business professionals. In contrast, Elgato knew exactly who it wanted its Stream Deck to be for: streamers. It held a six-month private beta to collect feedback from its intended users, and it poured a lot of effort into making sure the device integrated nicely with OBS, the industry-standard streaming software. “The big new component then was building software that was tailor made for live streamers,” Fest says. “What we did is we took something existing and just repositioned it for an audience that actually understood and appreciated what this thing could do.” That’s not to say Elgato didn’t have any challenges to overcome while developing the Stream Deck with iDisplay. Fest says a big one was ensuring that the Stream Deck could not only send information to a computer but also receive it back and show it to the user. Without being able to stay in sync like this, the Stream Deck risked being the glorified macro pad that critics claimed it was. “If you fire off a hotkey, you don’t know if that action succeeded,” Fest says. “What we wanted to ensure is that if you change a scene in OBS you can clearly see on your device [that] that scene is now active and the other one is not. Or if you mute yourself, you’re muted, and we can guarantee that you’re muted because we’re talking natively to OBS.” The approach worked. iDisplay had sold just a few hundred units of its Infinitton via its crowdfunding campaigns, but the Stream Deck quickly became a staple of the Twitch streamer’s toolkit after its release in 2017. A year later, Corsair acquired Elgato for an undisclosed sum — and in 2022, it bought iDisplay, locking down its LCD keys. Other companies have taken notice. In 2022, Razer introduced the Stream Controller, and a year after that, competing PC accessory manufacturer Logitech snatched up Loupedeck, which had worked with Razer on the rival streaming accessory. Much like when it morphed its TV tuner into a game recorder, Elgato developed an audience for a technology rather than the other way around. It had been the missing piece of the puzzle since the beginning. But ironically, the audience for the Stream Deck has since expanded almost as wide as the one iDisplay hoped would embrace the Infinitton. Fest says he knows of Stream Decks being used in the hundreds by organizations ranging from call centers to police dispatcher services. The UK’s Virgin Atlantic airline uses dozens of Stream Decks to simplify communications with pilots and air traffic controllers. Even SpaceX was an early customer, Fest says. (SpaceX did not respond to The Verge’s request for comment.) The simple genius of the Stream Deck is that it made LCD keys peripheral, useful for anything where you need a button that dynamically advertises what it’s doing. Art Lebedev was right about one thing: there was indeed a market of people prepared to pay top dollar for premium-quality keyboards. But in retrospect, the Russian design studio bet that the market would go in the opposite direction of the one it actually went in. Instead of a software-based future filled with screens, keyboards embraced hardware, rediscovering mechanical key switches that had been around since the ’80s. Users began fixating on typing feel, seeking out tactile switches and clacky keycaps. The many compromises of LCD keys might have been too much to ask. But Elgato didn’t need to find a balance between typing feel and screen because its Stream Deck aimed to complement a traditional keyboard, rather than replace it. “Everybody has opinions on how Stream Deck’s keys feel. Some absolutely love it. Some say it’s way too mushy. But for everybody, it’s usable,” Fest says. You can forgive a key that feels mushy if you only need to press it to mute yourself while streaming. Which is not to say there’s no space for the kind of screen-based design that Art Lebedev and iDisplay worked to create. The gaming accessories company Finalmouse appears to be banking on exactly that with its forthcoming Centerpiece keyboard, which draws on the single-screen approach of the Popularis but combines it with a set of actual mechanical switches — translucent ones — to retain the feel of a mechanical keyboard. Like the Optimus Maximus, the Centerpiece has already blown past its first promised ship dates and is entering the vaporware realm. But if it does arrive, perhaps function and form will finally be aligned. The history of the keyboard, and the company that built it, is also more complicated than the legacy of its design alone. Fueled in part by success from their design work, the firm’s founder, Artemy Lebedev, has become a prominent figure in Russia, known for spreading propaganda for Putin and defending the country’s war on Ukraine. Lebedev was banned from Ukraine in 2017 after entering occupied Crimea. In 2023, a Ukrainian court seized two Kyiv apartments from Lebedev after he posted a photo showing that he’d visited a Ukrainian power plant in territory occupied by Russia during the war. “I like to shit on authority,” he wrote on Instagram at the time. As for the firm, it doesn’t consider the Optimus keyboard a failure. “You could argue how successful the project was in terms of return on investment,” Burbaev says, telling us how much business it drummed up for the studio even a decade later. Sometimes a new client would admit that they, too, were fixated by those concept images back when they were a kid. Update, June 17th: Added context about Artemy Lebedev and his role in Russian propaganda.

The man behind the AI gaffes at Sports Illustrated and USA Today has a yearslong history of filling the internet with garbage. In the summer of 2018, staff of the Chicago Tribune awoke to find a story they didn’t recognize on the newspaper’s website. The article, multiple sources say, had something to do with a purse carried by Meghan Markle, the royal also known as the Duchess of Sussex. Advertisements frequently masquerade as news articles at the bottom of actual journalism — this is the phenomenon some have dubbed “the chumbox.” But this article about Markle’s purse was, in terms of where it was hosted, how it was formatted, and where it was appearing, truly in the Tribune — its online edition, at least. Yet this was not what one would expect to run on the site of one of the most prominent newspapers in the country. It was more suited for a gossip rag or blog, a type of “as seen on celebrities” article pushing product. It clearly did not meet the paper’s editorial standards. (One person with knowledge of the article recalls that Markle’s name was misspelled.) The piece was pulled down quickly after leadership at the company was alerted to it. When Tribune staff inspected the article in the site’s content management system, they discovered something curious: the piece had been published by BestReviews, a consumer product reviews website whose content was syndicated on the Tribune. This incident was highly unusual. A few months before the Markle story was published, Tribune Publishing, the company that owned the Tribune, acquired a majority stake in BestReviews. The relationship between the Tribune and BestReviews was courteous. BestReviews staff published their work on a separate site and also republished work on chicagotribune.com, a fairly standard syndication deal with no issues until this point. The Markle article was in contrast with established protocol: BestReviews shouldn’t have pushed new types of content to the Tribune site without discussing it with staff at the paper. After the article was published, top executives at Tribune Publishing called up and questioned BestReviews leadership, multiple people who worked at the companies said. Tribune staff were confused by what had happened, but it was explained to them as a fluke. After the article was pulled, boundaries and standards were reiterated to Ben Faw, a cofounder of BestReviews, and staff moved on — it was an odd incident in an era of far bigger scandals at Tribune Publishing. Unbeknownst to most, though, the article was, in fact, a piece of marketing content promoting a shopping startup founded by one of Faw’s friends, according to a person familiar with the situation. Faw did not dispute this when The Verge asked about the story. Fast-forward to today, and a different media scandal has roiled the publishing industry for the last year and a half: the frenzied proliferation of web content generated by artificial intelligence tools. In October, The Verge and other outlets reported on product review articles appearing on Gannett publications like USA Today that seemed to be AI-generated. Gannett maintained that the content was produced by humans and that a third-party marketing firm was responsible. Just a month later, eerily similar review articles were published on the website of Sports Illustrated, but this time, Futurism discovered that the article authors’ headshots were for sale on an AI photo website. Shortly after, Sports Illustrated said it had cut ties with the company that produced the reviews. The apparent AI content proved embarrassing for nearly everyone involved: venerated publications that hired a third-party marketing firm to produce content were now attempting to defend the work — and themselves — after readers discovered the low-quality junk content on their sites. Workers who had nothing to do with the stories feared it could be the beginning of the end of their jobs. In January, the Sports Illustrated newsroom was gutted by mass layoffs, though much of the staff was later rehired after its parent company found a new publisher. In both cases, as reported by The Verge, the AI-generated content was produced by a mysterious company called AdVon Commerce, a marketing firm that boasts of its AI-powered products. There’s little information available about AdVon online, as its owners have worked to scrub their names from the internet. But AdVon didn’t come out of nowhere. For years, according to former colleagues, internal documents, and court records, a person behind the operation has seemingly used his connections in the media industry to enrich himself: Ben Faw, CEO and cofounder of AdVon. For Faw, AI-generated sludge was just the next tactic to do so. Faw has a stacked resume that he regularly highlights: a US Army veteran and West Point graduate, he went on to attend Harvard Business School and lists previous jobs at companies like LinkedIn and Tesla on his LinkedIn page. But to some who worked with Faw before he started AdVon, it was no surprise that the shoddy AI content was traced back to him. “I literally saw that headline about Sports Illustrated [The Verge’s story],” one person who worked with Faw says, “and I was like, ‘I bet Ben did that.’” Online marketers focused on Google search traffic live and die by “reputation” — being seen as a reliable, trustworthy publisher in the eyes of Google leads to better placement in search results, which, in turn, leads to more clicks, more sales, and more eyeballs. One of the signals Google uses to assess the reputation of a site is backlinks, or how often other trustworthy websites link to a site. The idea is that publications are legitimized by how many others cite them. The bigger the website, the more authority it bestows on links, at least in the world of SEO, where links are like currency: marketers inundate reporters with pitches, hoping to get mentions and links in stories. The shadier SEO specialists buy, sell, and trade links, a practice that Google prohibits. In mid-2019, BestReviews staff began to notice that unfamiliar backlinks had been placed in content that was already live on the outlet’s website, bestreviews.com. The hyperlinks were numerous and largely unrelated to the articles they appeared on. They led to other blogs about pets and lawn care, OB-GYN practices in Ohio, and beauty product companies, according to documents reviewed by The Verge. But the seemingly random links had something in common: Faw was somehow connected to many of the companies being promoted. A cosmetics company called Beauty Bakerie, for example, was linked to dozens of times. Faw is described as a board member of the company in a biography published online. It wasn’t just the links to Beauty Bakerie that raised the alarm. Interviews with former associates, internal documents, and court records obtained by The Verge paint a picture of a complex web of companies involved in a backlink scheme, with Faw seemingly at the center. Some of the links involve a company called Alan Morgan Group, a digital marketing firm based in Ohio that was cofounded by Eric Spurling, the cofounder and president of AdVon and a classmate of Faw’s from West Point. The firm’s services include “getting [clients’] content featured on the front page of some of the best-known digital newspapers” and search engine optimization. According to two people with insider knowledge, BestReviews had hired the firm to do SEO work for the website. In reality, Faw and others involved with Alan Morgan appear to have set up an elaborate, self-dealing system that used Faw’s connections to BestReviews to enrich themselves. According to interviews and internal documents seen by The Verge, hundreds of links placed in BestReviews stories went to clients of Alan Morgan. Though BestReviews had hired Alan Morgan to improve its SEO, in practice, it was more like Alan Morgan clients were getting SEO services in the form of plum, potentially lucrative backlinks from BestReviews. According to documents seen by The Verge, Faw was confronted in June 2019 about the thousands of links inserted into BestReviews’ content, including links that didn’t make sense in context: the phrase “trauma shears”’ linked to Beauty Bakerie; a link on “learn more here” led to a different consumer reviews site. When Faw was asked about the links, he claimed that someone at Alan Morgan Group had inserted the links, according to multiple people with knowledge of the conversations. The relationship was terminated. What wasn’t known then, even to some senior BestReviews staff, is that Faw was allegedly employed by Alan Morgan and held a one-third equity stake in the company, according to a lawsuit filed by former business associate Jonathan Smith, a copy of which was obtained by The Verge. In the suit, Smith accuses Spurling and Faw of fraud and conspiracy, claiming the two ran off with money, intellectual property, and clients of his business as they spun up a dizzying number of companies — including, eventually, AdVon. Smith did not respond to requests for comment, and the case is now closed, per the court’s website. Interviews and documents suggest that Faw was employed at both Alan Morgan and BestReviews at the same time, in what appears to be a clear conflict of interest. There are other threads in the links that, once pulled at, unravel and seem to lead back to Spurling and Faw. On one of the backlinked sites, dog-gear.com, an “About Us” page describes a “founding team” with a background that bears a striking similarity to Faw’s resume: West Point, Harvard Business School, Fortune 500 companies. An archived version of a DogGear author page indexed by Google lists “[email protected]” as a contact email. In his suit, Smith alleges that Spurling spun up a company called Pet Gear LLC, a company designed to link publisher content with Amazon affiliate links “in a way that would maximize revenue for the publisher,” according to the lawsuit. In the suit, Smith also requests financial documents related to dog-gear.com, though it’s not clear that the site is operated by Pet Gear, founded by Spurling. Another site called Better Lawns and Garden was also linked to more than a hundred times, according to documents viewed by The Verge. The site publishes articles like, “The Best Potting Soils for Proper Plant Nutrition” and “Tree Felling: How to Safely Remove a Tree Yourself.” The pieces are written by a person named Mike Dover, who, according to the LinkedIn profile listed on his author page, works for AdVon. The competing reviews sites were especially strange, implying that Faw, while working at BestReviews, possibly spun up competing review sites and linked out to them hoping to take advantage of BestReviews’ prominence and Google standing. Some of the outside reviews sites mirrored BestReviews’ own content: an archived version of the FAQ page on dog-gear.com from April, for example, is nearly identical to an archived version of BestReviews’ FAQ page from May. At least once, this linking practice spilled over to the Tribune site as well. According to a person familiar with the incident, some time after the Meghan Markle purse story appeared and was removed, a Tribune web editor noticed a weird link inserted into a BestReviews article that was syndicated on the Tribune’s website. “It was just a normal hyperlink to, I think, a dentist or some sort of private business in the South,” the person told The Verge. “It made absolutely no sense in the content.” Like the Markle article, the addition of the link was traced back to BestReviews. Tribune staff quickly removed the hyperlink, and an executive called BestReviews to discuss the rogue link. After that, the source says, safeguards were put in place to prevent it from happening again, including limiting BestReviews’ CMS access. “In both those situations we noted it quickly — we had somebody minding the store,” a former Tribune staffer says. “We saw that something wasn’t right, we took care of it immediately [and] wiped it out.” Hedge fund Alden Global Capital acquired Tribune Publishing in 2021, after BestReviews was sold to Nexstar. Reached for comment via email, spokesperson Davidson Goldin didn’t answer questions related to the piece, saying that the events in this story predate Alden’s acquisition. But at BestReviews, no one noticed the links until months after they were inserted, according to a person who worked at the company. The idea that Faw allegedly had created competing reviews sites — some of which directly ripped off the work of BestReviews — while still working at the company he helped build was shocking. “No matter how you feel about a company, who does this?” they said. “Who cheats like this?” Faw was prominently featured in a recent Futurism report about AdVon’s numerous publisher deals. The outlet reported that AdVon’s product review content had appeared on sites ranging from the Los Angeles Times to Hollywood Life and Us Weekly, attributed to writers that don’t exist. The articles, with titles like “Best Yoga Mats” and “Develop Your Core Using the Best Ab Roller,” contain paragraphs of stilted, unnatural-sounding language, with links leading to Amazon product pages, earning a small commission every time a reader purchases an item. They are essentially sales pitches dressed up to look like news articles or product reviews. After both the USA Today / Gannett and Sports Illustrated debacles, AdVon denied claims that the product reviews were AI-generated. But when pressed by Futurism about documents that showed AdVon staff using AI, the company backtracked, saying automated tools were in use at least for some publishing partners. Still, many of the outlets cut ties with AdVon due to the quality of the work, Futurism reported. But a former AdVon employee told The Verge that the content that AdVon says is created by humans is nearly identical to the AI-generated content they created while working there. Freelancers who were initially hired as writers were reassigned to roles of editors and tasked with making AI-generated writing sound human. The tool AdVon used — called MEL internally — generated hundreds of words on products using bare-bones prompts like “best televisions,” spitting out links to product pages on Amazon. “I looked at [MEL’s output] the first time and I just fell apart,” the former AdVon worker says. “Everything we were working towards — all that education, all of the writing experience … it was gone. There was none of the human journalistic writing. It was just, generate a bunch of words that we hope will look like a good article.” Faw and Spurling did not directly respond to The Verge’s questions about the details of this piece. Instead, Faw sent a statement saying AdVon is helpful to newsrooms and “generate[s] affiliate revenue which publishers use to fund newsroom operations and salaries.” “AdVon offers human-only, AI-enhanced, and hybrid solutions to help our customers with this problem,” the statement continues. “We’re committed to working closely with our publishing partners to ensure that their optional use of our AI solutions meets their content standards in this quickly evolving space.” Since the widespread availability of AI tools, news outlets have been at the center of conversations about artificial intelligence: Is there an ethical way to use them? How should outlets signal to readers when AI is involved? How can workers and human labor be protected? Product reviews and commerce content specifically have felt the effects of AI: across many outlets, it’s been this type of service journalism that corporations have unleashed AI “experiments” on first. At CNET, AI tools were secretly used on content meant to sell credit cards, insurance, and other products and services before readers and media caught on. At The Inventory, a site owned by G/O Media, a bot has published dozens of stories a week highlighting Amazon products that are on sale since company leadership announced its plans to use AI. At both Sports Illustrated and USA Today / Gannett, third-party articles with all the hallmarks of being AI-generated were published alongside the work of human journalists. The degradation of product reviews and recommendations is ironic given it’s often some of the most profitable work for media companies: each time a reader makes a purchase through a link in a story, the outlet gets a small kickback through affiliate revenue. Established review sites like The New York Times’ Wirecutter or niche interest blogs might not uncover corruption or break news, but readers care about that work — at their best, reviews are trustworthy, rigorously researched stories that help people make decisions about their lives. They also bring in money for news outlets. By all accounts, the work of BestReviews while under Tribune ownership struck that balance. BestReviews was cofounded by Denis Grosz and Momchil Filev in 2014 and quickly built an archive of lucrative product reviews that earned the outlet money each time a reader made a purchase from an article. Grosz and Filev declined to comment for this story. Nexstar, the media company that acquired BestReviews in 2020, also declined to comment. Multiple former BestReviews staff tell The Verge that they produced quality work, made by real people who cared about journalism. Articles were written by subject matter experts, or at least people who had an interest in the topic and were willing to do research about products and actually test them. There was a test lab, in-house photo and video teams, and editorial standards. To this day, BestReviews promises that it doesn’t accept payment or free products “in exchange for positive reviews,” therefore remaining unbiased. “It’s not like Pulitzer Prize-winning content that we were doing,” a former BestReviews employee says. “But truly the mission was: save people money by giving them an honest review, pros and cons, and give them a few options based on different budgets.” “We didn’t put out just a fluff piece or advertising for the [manufacturers],” another former staffer says. “We didn’t really take a bias towards a certain company. We kept our standards.” But the work produced at AdVon was different. A former staffer told The Verge that when they wrote articles, they were simply creating marketing materials for brands that had hired AdVon to promote their products. There was no testing of gear or even selecting different models to compare; writers simply got a list of products and were tasked with rewriting Amazon product listings into articles. Another person who worked at AdVon described a similar experience using the AI system, in which MEL selected the products featured in the articles for them. “[It was] promotional listicle content articles. It would be like, ‘10 best backpacks,’ and then they would have a list of backpacks promoted by companies that they were trying to support,” a former AdVon staffer says. “Most of this was to try to break through [search algorithms] so that that content could be promoted through SEO.” While reputable outlets review and recommend products without fear or favor, an army of lower-quality blogs, content farms, and influencers push similar content that’s designed only to generate sales. “Best of” lists can be a cash cow for publishers looking for evergreen content that drives traffic from Google, and manufacturers stand to benefit from the press mentioning and linking to their products. As publishers stare down the barrel of Google search traffic drying up to zero, outlets big and small, trustworthy and less so, are leaning into product review content, hoping it will buoy a sinking ship. But the ecosystem of online reviews is essentially the Wild West. One of the countless companies promising to help brands get publicity via product reviews is called SellerRocket. The company markets to Amazon sellers who want their products mentioned in recommendation lists or reviews. SellerRocket says it has “helped more than 4,000 brands get coverage.” In one webinar posted to YouTube, a SellerRocket employee explains to prospective clients how the company can get their products featured in news articles and reviews. They pull up a Sports Illustrated review of ab rollers — authored by “Damon Ward,” one of the fake AI authors created by AdVon — saying the magazine is a “publisher we work with.” They then show that the best ab roller article is ranking highly on Google Search. Clicking on the story, the employee says, “They always lay out how they went about their grading process, why they chose what [products] they did.” There’s no mention of AdVon, but the article includes a familiar disclaimer: “Sports Illustrated and its partners may earn a commission if you purchase a product through one of our links.” It’s not clear who selected the items appearing in AdVon’s product recommendations or why SellerRocket boasted of its success using an AdVon article. But according to Utah public records, the registered agent for a company called Seller Rocket shares a name with an operations specialist working at AdVon. SellerRocket didn’t respond to questions about its relationship with AdVon. If any of the Sports Illustrated or Gannett content featured clients that paid for placement in reviews, that financial relationship wasn’t clearly disclosed to readers — which, if true, would be a violation of Section 5 of the FTC Act, which prohibits “unfair or deceptive acts or practices in or affecting commerce.” The SellerRocket website previously listed the logos of BuzzFeed, Forbes, Wirecutter, and BestReviews, implying they were SellerRocket clients. Faw and Spurling did not respond to questions surrounding the use of these logos. “Wirecutter has never done business with SellerRocket. The use of our logo was misleading and factually incorrect and we had them take it down via a cease and desist order,” Jordan Cohen, spokesperson for The New York Times said in an email. Forbes spokesperson Jocelyn Swift and BuzzFeed spokesperson Juliana Clifton also told The Verge that the outlets have never worked with SellerRocket. SellerRocket has since removed mention of those publications from its site. By drawing on his deep network in media and product reviews, Faw was able to land deals with premier publishers like Sports Illustrated, USA Today, and McClatchy newspapers like the Miami Herald. The seemingly AI-generated content that was described as human-made lost AdVon business, and the publications running the work issued naive-sounding statements. Sports Illustrated, for example, said at the time that AdVon content was not AI-generated but that the listed author names were fake, leading the publisher to drop AdVon. After revelations that Sports Illustrated had been publishing content attributed to fake writers, the outlet’s then publisher ousted CEO Ross Levinsohn, a serial media executive who worked at Tribune Publishing around the time Faw was at BestReviews. (Levinsohn left the company after NPR reported he had been sued twice for sexual harassment, with both cases ending in settlements.) In the time since the controversy erupted, Faw and Spurling have removed their names from the AdVon website. But for the writers hired by AdVon to actually write, the company’s business practices have had lingering effects. One former AdVon worker told The Verge of having work dry up as they were instructed to use AI tools, so much so that eventually they couldn’t continue to do work for the firm. Another former AdVon writer — who maintains they didn’t use AI while working at the company — says that because AdVon has deleted so much of its work, they’re unable to find their human-written clips to show potential clients as work samples. Some time after they left AdVon, an AdVon employee reached out to them, offering an AI editing position. AdVon and the companies that have hired the firm are doubling down on all the same tactics that digital publishers have wrung dry: flooding Google with clickable, thin content; accelerating output at the cost of quality work; and trying to replace knowledgeable humans with cheaper machines. For AdVon and other companies, the shift to AI is marketed as exciting, new, and forward-thinking. The name “AdVon” stands for advanced echelon, a military term describing a group that’s the first on the ground ahead of everyone else: elite, cutting-edge, leaders. Instead, Faw and his team flooded the internet with duplicitous, cheap words, nakedly designed for quick profits at the expense of human workers. It’s not visionary, but it’s the oldest trick in the book, a regurgitation of old ideas dressed up as new. The only question is how much longer it can last.

As Sonos continues chipping away at fixing bugs and improving the performance of its poorly received mobile app redesign, the company also remains focused on several new hardware products. Among them is an ultrapremium soundbar that will serve as a successor to the current Sonos Arc — and likely at a higher price. Codenamed Lasso, the device is currently in limited beta testing, and today, I can share the first images of it. The exterior of the Lasso soundbar bears a strong resemblance to the Arc, which Sonos introduced over four years ago. But inside, the components and speaker drivers have been completely overhauled and redesigned, and that’s expected to result in a cost upwards of $1,200. Sonos aims to ship Lasso later this year, though the timeframe could change. Whenever it does arrive, Lasso (hardware model S45) will be the first Sonos product to integrate technology from Mayht, a Netherlands-based startup that the company acquired in 2022 for $100 million in cash. At the time, Sonos credited Mayht for having “invented a new, revolutionary approach to audio transducers.” CEO Patrick Spence said the deal would give Sonos “more incredible people, technology, and intellectual property that will further distinguish the Sonos experience, enhance our competitive advantage, and accelerate our future roadmap.” But it’s taken quite some time to get Mayht’s ideas into a shipping product. Sonos has released several small speakers over the last couple years — the Era 100, Move 2, and Roam 2 among them — that could’ve benefitted from Mayht’s innovative transducers, which are said to “enable smaller and lighter form factors without compromising on quality.” But it’s the soundbar that will offer the first taste of what the acquisition has yielded. Sources tell me that it’s been a challenge to make Mayht’s transducer system more cost efficient, as it partly makes use of “expensive” neodymium magnets. Bloomberg previously reported the Lasso codename, its rumored price, and the inclusion of Mayht’s technology. The product’s final consumer branding isn’t yet known, but for now, Sonos is playfully including the Ted Lasso typeface on beta units. The Arc is already a very impressive Dolby Atmos soundbar, and you can expect even greater immersion from Lasso and its revamped internals. Additionally, according to sources, the upcoming soundbar should deliver substantially better bass performance than its predecessor. With the Arc, you really need to add a Sub or Sub Mini to get the most out of it. Sonos could maybe get away with that with a $899 product, but if it’s going to charge well over $1,000 for Lasso, the device needs to deliver top-tier home theater audio without any added help. The images also reveal a dedicated Bluetooth button, suggesting that Lasso could support Bluetooth audio playback. That feature has been absent from all previous Sonos soundbars. And similar to the company’s other recent products, there’ll be a physical switch for disabling the built-in microphones. Just like the Arc, Lasso is designed to work with the recently released Sonos Ace headphones for private listening with spatial audio. The company has pledged to extend Ace compatibility to its more affordable Beam and Ray soundbars by late summer or early fall. As for the more immediate future, Sonos continues to release updates to its mobile app at a frequent pace as it tries to address customer frustrations and lift the negative sentiment that has embroiled the company’s community since May. It’s clear to me that Sonos is listening and moving quickly, but a lot of damage has already been done: the Android version of the Sonos app currently has a brutal 1.2-star average review rating. The rating on Apple’s App Store isn’t nearly as bad since older reviews are still balancing out the spate of complaints that came with the redesign.

Priscila Barbosa came to the US with a dream of making it. Using gig-economy platforms she built a business empire up from nothing. There was just one huge problem.

Forget the S&P 500. Look at the ice cream. This investor visits homes around the world to see where diets—and economies—will go next.

Beneath its playful exterior, the beloved game that inspired a generation of real-world urban designers betrays a partisan view of social planning.

We’ve reached a tipping point where we’ve got a cleaner alternative for most transport. Now we have to commit.

After hitting a power-output milestone, fusion technology is ready to graduate from small-scale lab experiment to full-sized power plant.

Margaret Atwood, Marlon James, Lena Dunham, Garth Greenwell, Roxane Gay: We’ve all agreed to be turned into chatbots by a mysterious startup called Rebind. I report from the inside.

Microsoft is losing another key Xbox executive at the end of the month. Xbox chief marketing officer Jerret West is leaving Microsoft to join Roblox as its new CMO and head of market expansion. Microsoft Gaming CEO Phil Spencer confirmed the move in an internal memo to Xbox employees today, which was obtained by The Verge. West previously spent eight years at Microsoft on the Xbox marketing side, before departing in 2011 and then eventually spending seven years at Netflix as the head of marketing. West returned to Microsoft in late 2019, leading the marketing for the launch of the Xbox Series S / X consoles. West’s team is responsible for developing marketing plans for games, hardware, and Xbox Game Pass. Microsoft is now shuffling around some of its marketing teams in the wake of West’s departure, and there will be a new expanded central gaming marketing team under Kirsten Ward, VP of Xbox integrated marketing. Spencer says the Microsoft Gaming leadership team has “decided to place marketing closer to the businesses they support” as a result of West’s departure. Games marketing will now sit inside the game content and studios division that’s led by Matt Booty. Xbox marketing, led by Chris Lee, will move to the Xbox org and report up to Xbox president Sarah Bond. This is the second key Xbox executive resignation this year, after former head of Xbox emerging tech, Kareem Choudhry, left Microsoft in early April. Choudhry’s departure triggered a similar shake-up, with his former team moving to the Xbox hardware side. A new Xbox experiences and platforms team was created after Choudhry departed, with a push to improve the Xbox experience across Windows and Xbox consoles. This latest marketing shake-up also comes around eight months after a big shake-up of Xbox leadership and amid continued changes to Microsoft’s gaming strategy. Microsoft promoted Bond to Xbox president in October, leading all Xbox platform and hardware work. Booty also got promoted to the president of game content and studios, including overseeing Bethesda and ZeniMax studios. Here’s Spencer’s marketing memo in full: I have some news to share with all of you: Jerret West, who has served as the Chief Marketing Officer of Xbox for four and a half years, will be leaving Microsoft to join Roblox as their new CMO and Head of Market Expansion. His last day will be June 30th. On behalf of Team Xbox, I’d like to thank Jerret for everything he accomplished. During his time here, Jerret was instrumental in launching and growing communities like our Xbox social universe to 120M followers strong, and product expansion like Game Pass which is now available in more than 80 countries worldwide. He oversaw the creation and delivery of global marketing campaigns for numerous games, platforms (console, PC, cloud) and brand moments such as this year’s momentous Xbox Games Showcase. Roblox is an important partner for Xbox, and we look forward to working with Jerret and the Roblox team to connect more people around the world through the power of play. As a team we are committed to growing the strong marketing capability in Gaming. Going forward, the Gaming Leadership Team and I have decided to place marketing closer to the businesses they support. Games marketing will sit inside of Studios, Xbox marketing will sit inside the Xbox team, international marketing will sit inside of the Consumer Sales Org, and our centralized marketing team will report to me. The following organizational changes will be effective on July 1st: The Integrated Marketing team led by Kirsten Ward and the Programming and Events team led by Tina Summerford will come together as an expanded Central Gaming Marketing team under Kirsten, who will report to me as a member of the Gaming Executive Leadership Team. Chief of staff Pav Bhardwaj and executive assistant Kim Nold will report to Kirsten. The Bethesda Publishing Marketing team, led by Erin Losi, will move to Game Content and Studios, reporting to Matt Booty. The 1P Games Marketing Team including Xbox Game Studios and Mojang, led by Aaron Greenberg, will also move to Game Content and Studios, reporting to Matt. The Xbox Marketing team, led by Chris Lee, will move to the Xbox Org, reporting to Sarah Bond. The International Marketing team, led by Jim McMullin and Florian Liewer, will move to the Consumer Sales Org, reporting to Ami Silverman. The Bethesda International Marketing team, led by Adam Carter, will also move to the Consumer Sales Org. Adam will be taking on a new role leading Business Operations at Bethesda, reporting to Jill Braff. The Consumer Sales Org will work on backfilling his role and share more at a later date. We’re fortunate to have a deep bench of talent and experience, and I’m excited to see how our marketing will evolve under their leadership. Phil

Microsoft is planning to launch its new Copilot Plus PCs next week without its controversial Recall feature that screenshots everything you do on these new laptops. The software maker is holding back Recall so it can test it with the Windows Insider program, after originally promising to ship Recall as an opt-in feature with additional security improvements. “We are adjusting the release model for Recall to leverage the expertise of the Windows Insider community to ensure the experience meets our high standards for quality and security,” says Microsoft in an updated blog post. “When Recall (preview) becomes available in the Windows Insider Program, we will publish a blog post with details on how to get the preview.” This means that Recall won’t even be available initially to Windows Insiders or anyone who buys a Copilot Plus PC. I wrote in Notepad earlier today that Windows engineers were scrambling to get the security improvements tested and implemented in time for the June 18th launch date of Copilot Plus PCs. Now, Microsoft is essentially admitting here that it needs more time to test Recall’s security improvements. Microsoft first unveiled the Recall feature as part of its upcoming Copilot Plus PCs last month, but since then, privacy advocates and security experts have been warning that Recall could be a “disaster” for cybersecurity without changes. Microsoft committed to three major updates to Recall last week, including making the AI-powered feature an opt-in experience instead of on by default, encrypting the database, and authenticating through Windows Hello. Recall uses local AI models built into Windows 11 to screenshot mostly everything you see or do on your computer and then give you the ability to search and retrieve items you’ve seen. An explorable timeline lets you scroll through these snapshots to look back on what you did on a particular day on your PC. Everything in Recall is designed to remain local and private on-device, so no data is used to train Microsoft’s AI models. Microsoft’s decision to delay Recall comes just after vice chair and president Brad Smith testified before the House Homeland Security Committee today. Smith said that Microsoft is putting security above everything, as part of its Secure Future Initiative (SFI). “It is more important even than the company’s work on artificial intelligence,” says Smith. Smith also revealed that Microsoft will make security a mandatory part of its bi-annual reviews process for all employees. “With this change, cybersecurity will be considered in every employee’s annual bonus and compensation,” Smith said. I reported earlier today in Notepad that Recall was originally created before Microsoft’s big SFI overhaul begun. Recall was developed in secret at Microsoft, and it wasn’t even tested publicly with Windows Insiders. Microsoft subsequently identified some of the security issues with Recall and started to develop and test changes to the experience in recent months. It clearly now needs more time to make sure these changes stand up to its promise of putting security above AI and everything else.

In The Singularity Is Nearer: When We Merge With AI, the spiritual sequel to his (in)famous first book, Ray Kurzweil doubles down on the promise of immortality.

The famed futurist remains inhumanly optimistic about the world and his own fate—and thinks the singularity is minutes away.

It’s happy hour in Las Vegas, and the MGM Grand casino is crawling with people. The National Finals Rodeo is in town, the NBA’s inaugural in-season tournament is underway, the Raiders play on Sunday, and the U2 residency is going strong at the giant Sphere, so it seems everyone in every bar and at every slot machine is looking forward to something. (And wearing a cowboy hat.) Even for a town built on nonstop buzz, this qualifies as a uniquely eventful weekend. But I’d wager that if you wanted to see the most exciting drama happening at the MGM on this Friday night, you’d have to walk through the casino and look for the small sign advertising something called The Active Cell. This is the site of the play-in round for the Excel World Championship, and it starts in five minutes. There are 27 people here to take part in this event (28 registered, but one evidently chickened out before we started), which will send its top eight finishers to tomorrow night’s finals. There, one person will be crowned the Excel World Champion, which comes with a trophy and a championship belt and the ability to spend the next 12 months bragging about being officially the world’s best spreadsheeter. Eight people have already qualified for the finals; some of today’s 27 contestants lost in those qualifying rounds, others just showed up last-minute in hopes of a comeback. The room is set up with four rows and three columns of tables, each one draped in a black tablecloth and covered in power strips, laptops, and the occasional notepad. There’s a long table with coffee in the back, and over the two days we’ve been in this room, carts have occasionally wheeled in with cookies, queso dip, and at one point, surprisingly delicious churros. The unofficial dress code is business casual, the overall vibe somewhere between summer camp and business conference. Now the room is quieter, more focused. 26 of the contestants are furiously setting up their workspaces. They plug in their computers, clean up their areas, and refill their beverages. A number of players reach into their bags and pull out an external mouse and keyboard — everyone in the room has strong opinions on brands and features, but all agree that what you really need is a keyboard with function keys separate from media keys, and then to turn those media keys into more function keys so you can work even faster. And then there’s me. I’m the 27th competitor, and I’m both the only person in the room using a Mac and the only person who has no idea what I’m doing. I’ve spent the last two days in this room with this group, as they’ve taught each other new Excel tricks and compared notes on the state of the art in the world’s most important piece of software. They’ve been debating VLOOKUP and XLOOKUP and teaching each other how to use the MOD function. I’ve been desperately trying to get my app to update on the MGM’s Wi-Fi. At 6PM on the dot, Andrew Grigolyunovich, the founder and CEO of the Financial Modeling World Cup, the organization hosting these championships, takes the modular stage in the ballroom. He loads an unlisted YouTube link, which begins explaining today’s challenge, known as a “case.” It’s a puzzle called “Potions Master,” and it goes roughly like this: You’re training to be a potions master in Excelburg, but you’re terrible at it. You have a number of ingredients, each of which has a certain number of associated points; your goal is to get the most points in each potion before it explodes, which it does based on how much of a white ingredient you’ve added. The Potions Master case, like so many of the puzzles conquered by these competitive Excelers, is not particularly complicated. This is a flashier, faster, deliberately more arcade-y version of spreadsheeting, more like trying to win 10 simultaneous games of chess on easy mode rather than painstakingly taking on a grandmaster. If you like, you can solve the whole thing manually: figure out when the white number gets too high, count the total points until that spot, then double-check it because it’s a lot of numbers, and eventually answer the first question. That’s my strategy, and I think I get it right. Now there are 119 more, worth a total of 1,500 points, and it’s quickly clear I’m not going to finish in the 30 minutes we’ve been allotted. While I’m squinting into my 13-inch screen and carefully adding 1s and 3s, the other 26 contestants are whirring through their spreadsheets, using Excel’s built-in formula and data visualization tools to organize and query all that data. Everyone in the room seems to have their own way to chew through the ingredient lists and spends the first few minutes turning a mess of numbers and letters into real, proper capital-d Data. They start answering questions a half-dozen at a time, while I’m still checking my mental math. Almost everybody who participates in competitive Excel will tell you that the app itself will only get you so far. If you can’t hack the puzzle or figure out what you’re trying to do, it can’t make something out of nothing. Your brain will always matter more than your software. But if you really know how to make Excel sing, there’s simply no more powerful piece of software on the planet for turning a mess of numbers into answers and sense. Competitive Excel has been around for years, but only in a hobbyist way. Most of the people in this room full of actuaries, analysts, accountants, and investors play Excel the way I play Scrabble or do the crossword — exercising your brain using tools you understand. But last year’s competition became a viral hit on ESPN and YouTube, and this year, the organizers are trying to capitalize. After all, someone points out to me, poker is basically just math, and it’s all over TV. Why not spreadsheets? Excel is a tool. It’s a game. Now it hopes to become a sport. I’ve come to realize in my two days in this ballroom that understanding a spreadsheet is like a superpower. The folks in this room make their living on their ability to take some complex thing — a company’s sales, a person’s lifestyle, a region’s political leanings, a race car — and pull it apart into its many component pieces. If you can reduce the world down to a bunch of rows and columns, you can control it. Manipulate it. Build it and rebuild it in a thousand new ways, with a couple of hotkeys and an undo button at the ready. A good spreadsheet shows you the universe and gives you the ability to create new ones. And the people in this room, in their dad jeans and short-sleeved button-downs, are the gods on Olympus, bending everything to their will. There is one inescapably weird thing about competitive Excel: spreadsheets are not fun. Spreadsheets are very powerful, very interesting, very important, but they are for work. Most of what happens at the FMWC is, in almost every practical way, indistinguishable from the normal work that millions of people do in spreadsheets every day. You can gussy up the format, shorten the timelines, and raise the stakes all you want — the reality is you’re still asking a bunch of people who make spreadsheets for a living to just make more spreadsheets, even if they’re doing it in Vegas. You really can’t overstate how important and ubiquitous spreadsheets really are, though. “Electronic spreadsheets” actually date back earlier than computers and are maybe the single most important reason computers first became mainstream. In the late 1970s, a Harvard MBA student named Dan Bricklin started to dream up a software program that could automatically do the math he was constantly doing and re-doing in class. “I imagined a magic blackboard that if you erased one number and wrote a new thing in, all of the other numbers would automatically change, like word processing with numbers,” he said in a 2016 TED Talk. This sounds quaint and obvious now, but it was revolutionary then. Bricklin’s software, eventually called VisiCalc, gave many people their first good reason ever to buy a computer. In 1996, Apple CEO Steve Jobs called VisiCalc the first of two “explosions that propelled the industry forward” and said spreadsheets were the driving force behind the success of the Apple II. A generation later, a competitor called Lotus 1-2-3 became a key app for the IBM PC. By 1985, after briefly dabbling with a program called Multiplan, Microsoft announced a powerful spreadsheet app of its own, called Excel. At the time, it was an app for Apple’s Macintosh, which was flagging in sales; both Apple and Microsoft thought the best way to compete was with spreadsheets. They were right. Four decades later, Microsoft CEO Satya Nadella called Excel “the best consumer product we ever created.” He doesn’t just see it as an enterprise tool. It’s for everything. Nadella said he simply can’t imagine a world without Excel. “People couldn’t make sense of numbers before, and now everyone can.” The goal isn’t just getting an answer. It’s understanding all the inputs that allowed us to arrive there. Looking back, there’s a surprising resemblance between the way we talked about spreadsheets in the ’80s and the way we talk about artificial intelligence now. The same worries about automating people out of jobs; the same questions about whether we could really trust the computers to do all this complicated work so quickly. In fact, in the 1980s, spreadsheet programs were the AI bots of their day. “The aim is to knock some sense into otherwise mindless computers,” The New York Times’ David Sanger wrote in 1985, “getting them to understand — and perform automatically — the tasks that individual users struggle each day not to forget.” In so many ways, though, the spreadsheet trajectory is the best-case scenario for an AI future. Where current AI tools like ChatGPT try to abstract away the inner workings and underlying data and simply offer you the world through a text box, spreadsheets do the opposite: they promise an ever greater level of control and understanding of the world around you. The people who work on Excel and other spreadsheet tools are perpetually trying to make them easier to use while also giving power users more ways to tinker. If you want to create something with AI, you just type in a prompt and hope for the best. A spreadsheet artist, on the other hand — and there really is such a thing — can paint their creation one cell at a time. The goal isn’t just getting an answer. It’s understanding all the inputs that allowed us to arrive there. The converse of that, though, is that spreadsheets make plain exactly how easy it is to reduce so much of modern life to a bunch of numbers and formulas in a spreadsheet. Give me some numbers, and my Excel file will predict when you’re going to die. Dating spreadsheets have become normal in a world where romance is about swipes and statistics. Have a hard decision to make? I have a decision-matrix spreadsheet for you! In a spreadsheet world, everything is comparable, reducible to some base figure that eventually explains everything if only you know how to ask. Spreadsheets promise the world isn’t actually complicated — you just have to know the formulas. I don’t know if that’s beautiful or bleak or both, but it’s certainly big business. This group that has gathered in Vegas operates like a mix of camp friends, colleagues, and competitors. Most work in financial analysis in some way: on the first day, I had lunch with a financial modeler for a mining company, an actuary, and a certified Microsoft MVP — a Most Valuable Professional, who teaches other people how to use Excel. We sat in the MGM Grand food court eating salads and Chinese buffet talking about the MVP’s expansive kaleidoscope collection and the modeler’s current budgeting challenges given the strange economy and some regulatory complications. Many of these folks have been playing competitive Excel together for years. Years ago, the main organization for these players was called ModelOff, which was much more strictly a financial-modeling competition. Every year, a bunch of these Excel users would get together, usually in New York or London, and essentially do their job competitively. The last ModelOff was in 2019, before the pandemic put a stop to pretty much anything in person. But Grigolyunovich, a longtime ModelOff fan and competitor, decided to keep the legacy going. Grigolyunovich is tall, with blonde hair parted down the middle and his shirt permanently tucked in, and exudes a kind of constant low-grade, manic energy. He loves Excel — loves using it, loves talking about it, loves tinkering with the sheets he’s made for this weekend. He created the Financial Modeling World Cup in 2020 in hopes that he could keep the ModelOff spirit alive but also expand it. “I really missed playing” when ModelOff stopped, he says, “because I’d been doing that for seven years. I also wanted to make a better tournament.” The only downside, he says, is that running the competition means he can’t participate anymore. The Financial Modeling World Cup is really three separate things: There’s the Excel World Championship, the most mainstream (and I use “mainstream” in the loosest way possible) version of competitive Excel, which Grigolyunovich hopes can turn into a popular esport. There’s a similar event for college students, known as the Microsoft Excel Collegiate Challenge. And there’s the Financial Modeling World Cup, which is more like ModelOff — it requires financial knowledge, uses financial cases, and has slightly different rules. The FMWC is the most complex, maybe the most prestigious, and definitely the least exciting of the three. The Excel World Championship had a viral moment in 2022, when it showed up in a half-hour block on ESPN’s annual “The Ocho” event — a joke from the movie Dodgeball before ESPN took the idea and made it real — in which the network airs a day full of sports that would otherwise never make its schedule. The “Excel Esports: All-Star Battle” portion aired at 5AM Eastern, between the 2022 eSkootr Championship and the 2021 World Air Hockey Championship, but enough viewers were excited and surprised to see competitive Excel on TV that it had a bit of an online moment. Now, Grigolyunovich says the job is to turn that virality into true momentum. He wants the FMWC to be a fun community activity, an educational resource for people of all ages, and an honest-to-God spectator sport. It’s not clear to me whether it’s possible to do all three of those things — and a lot of the people here think it’s not. David Brown, a University of Arizona professor and previous FMWC finalist who also runs the collegiate tournaments, says he thinks competitive Excel makes the most sense as a fun way to teach students some more practical skills, in a sort of Model UN way. Hardly anyone here seems to think Excel esports is a path to true fame and fortune. The best way to train for the Excel World Championship, everybody tells me, is to practice with old cases. The great players treat this the way a football player might watch film or run the same play over and over until the timing is perfect; you do a case, then do it again another way, until you’ve sharpened your skills and your muscle memory such that the next time something like it comes up, you’re ready. Every case is different, but they do have a lot in common. Lotteries and slot machines are common case fodder, and there are plenty involving poker. The final case in last year’s championship was about chess, another game with near-infinite variations and permutations. The more you train your brain to work with these mountains of data, the better you get. Diarmuid Early, a past ModelOff champion and the biggest celebrity in the room — an article once referred to him as “the LeBron James of Excel,” which immediately stuck — is sitting in the back row of the Vegas ballroom, trying not to think about whether he’s trained enough. I’d been hearing about Early for months before we met, and he’s one of two names everyone offers when I ask who they think might win. People speak of him both fondly and with just a little bit of reverence. In reality, Early — everybody calls him “Dim” — is a friendly, relaxed but fast-talking financial consultant with thinning hair and a slight Irish accent who showed up to Vegas in the middle of the busiest time of his year at work. I catch him in jeans and a gray hoodie, typing furiously on his laptop. In Excel, obviously. “I do bonus planning work with an investment bank,” he tells me, “so the end of the year is Go Time.” He’s already qualified for the finals on Saturday but is a little nervous about how he’ll do this weekend. “I was hoping to be doing more training than I am,” he says, but work has made that impossible. Even during the down moments (and the occasional not-so-interesting moments) of the conference, I keep catching Early swapping his fun spreadsheets for work spreadsheets. All the important stuff in an Excel competition, Early says, happens in the first few minutes. After reading the instructions, “you’ve got, like, 30 seconds to think, ‘Okay, how am I going to approach this?’ and then it’s just, go. If you set off down the wrong path, even if you realize five minutes in that there was a better way to build it, it’s probably not worth going back and changing it.” Some players will just dive into the first question in the first level and take things from there; the ones who win are usually the people who build a system that will eventually answer all the questions. “But if you get that judgment call wrong,” Early says, “and you’re three minutes away from having all the points but what you actually have is none of the points, then you’re dead, right?” He has horror stories of this happening. Everyone does. “If you ask the same question to five different people, and they have Excel experience, they might think of five completely different ways to do it,” says Peter Scharl, another already qualified finalist. “There are things I do in the Excel competitions that I didn’t know how to do three months ago, six months ago just because of doing it and reviewing what I did or reviewing what other people are doing, reading comments, watching videos.” In the last couple of years, though, there’s been a shift in the Excel world: you can now write complex, reusable functions called Lambdas in the app and even write Python code directly into a cell. (Microsoft’s Nadella recently called Excel formulas “the world’s most popular programming language,” and he’s probably not wrong.) More than one person in Vegas tells me about the work of Eric Oehm, known to many in the community as the Excel Robot, who builds software that automates a lot of the drudgery of building spreadsheets. Brown jokes that Oehm might give him some particularly cool tools that he hides from Early “because he wants someone to beat Dim.” Most days of the year, tools like these are a godsend to everyone in this room in Vegas. A faster way to build powerful systems in a spreadsheet? Sold. But in competition, it’s like a performance-enhancing drug. Brown says he suspects eventually, there will be two versions of the Excel World Championship: one in which you can bring all the outside tools and knowledge you want, and one that is just you and a fresh install of Excel. The former version, he thinks, could be wild in what it makes possible, much like Peter Thiel and others think the so-called “Enhanced Games” could make athletic competition all the more impressive. By now, I’ve asked a few folks if they’re worried about eventually being replaced by AI, and they all basically laugh — have you seen what happens when ChatGPT does even simple math, and do you know how high the stakes are for getting a company’s decade-long projections right? But Excel Robot and Python do seem to spark some existential concern. They might make Excel easier, but they also make it more opaque. And is being good at Excel about getting answers out of the spreadsheet or about understanding where the answers come from? The energy in the MGM Grand ballroom is noticeably different on Saturday morning. It’s finals day. We’re due for a few more panels and tutorials, and the champion will be crowned this evening. Before the day’s events kick off, I run into the biggest name in Excel who isn’t competing tonight: Laurence Lau. He is the opposite of everyone else I’ve met this week. Lau is wearing a leather jacket and sunglasses in a room full of khakis and dress shirts, he’s loud and brash and rude to some of the other competitors, and he thinks the whole FMWC is a little ridiculous. He’s also the #1 ranked player in the financial modeling side of the competition, and virtually everyone agrees he’s the best Exceler in the world when he bothers to show up. If the FMWC wants to be a success, Lau tells me, it needs to be bigger. If it wants to be bigger, it needs the best players and the biggest personalities to show up. Lau believes he’s here to remind everyone of how exciting all this could be, if only the organizers pull a Jerry Maguire and show him the money. (He points out that this whole thing could be bankrolled by Microsoft, the world’s most valuable company, with money found in the couch cushions in Redmond.) Even his jacket — a New York Yankees All-Star Game jacket from 2008, which he says was his favorite team “because I’m a fan of whatever team has the highest payroll” — is his way of reminding everyone that sports are about stars, and stars should be paid like it. Lau wants the FMWC to have multimillion-dollar prizes, not the measly three grand the winner will get tonight. If this is a sport, it should be a full-time job. Right now, he says, “people do this because they like Excel.” These events will always feel more like a golf weekend with the guys than a PGA championship. And, he reminds me, not everyone likes Excel. “Most Americans hate their job, right?” They don’t come home from work wanting to look at more spreadsheets. It’s fine for the FMWC to be about education and camaraderie, but then why is it streaming on ESPN? He senses an identity crisis in this room, and it makes him mad that no one will do anything about it. Lau compares himself to Deion Sanders, the famous football player who took on the “Prime Time” persona to bring more attention to players at cornerback, which, before him, had been one of the lowest-paying positions in the NFL. He’s happy to be the brash jerk in the room, trying to convince people to make the competition worth his while — and maybe build a brand in the process. The world may not run on spreadsheets, but spreadsheets run the world His obnoxiousness stands out even more in this crowd, which seems to skew introverted and mild-mannered. Nobody’s exactly competing for stage time with Lau’s antics. But as uncomfortable as everyone appears to be with his shtick, they also seem to understand his point. After all, these people do puzzles for fun and overwhelmingly do financial modeling for work. For all the fun art projects and life-tracking stuff that everyday people do in Excel, the true customers for these tools are the money guys. The ones who used the advent of the spreadsheet to turn Wall Street into a global industry, that built wildly complicated things like collateralized debt obligations and helped usher in a financial crisis in 2008. The world may not run on spreadsheets, but spreadsheets run the world. Maybe all Lau is doing is saying the quiet part out loud, which is surprisingly uneasy in a room full of finance professionals. Lau heads inside to join a panel of former Excel champions, and I head over to Grigolyunovich to get his take. He’s a little embarrassed by Lau’s whole alter ego bit, it seems, though he does agree in principle with some of his stances. Grigolyunovich says there’s plenty of room for more money — including from Microsoft, which is already a sponsor and would surely be happy to see Excel become must-watch TV. But he’s torn between loving this community and wanting it to be educational and fun for the people who do it while also wanting it to be bigger. And he wonders if there could be another reason Lau’s not competing — maybe he doesn’t like the cameras and the pressure? Meanwhile, Lau has taken the stage in that Yankees jacket and sunglasses and is explaining to everyone why they shouldn’t even try to compete: because, odds are, they’re not smart enough. When the moderator asks him if this is all just a bit, he laughs. “I can be normal if I need to be,” he says. But he gets back to his point. “My argument is a top Excel analyst is three, four, or five times more productive than your average,” he says, working himself up again. “They’re not getting paid three, four, or five times more than your average person.” He gestures to the audience: “If you work for a company and you’re pretty good at Excel, you are underpaid… so just remember that.” Brown, the moderator, quickly changes the subject. After the panels end, Grigolyunovich announces the winners of yesterday’s last-chance qualification round. One by one, he reveals the eight names of the top finishers, rounding out the field of 16 for tonight’s finals. (Jakub Pomykalski, an already qualified finalist who did the qualification round for fun, won it by scoring 1,234 of the 1,500 available points — I don’t think he intended to intimidate everyone by doing that, but he did.) I didn’t qualify, which is not remotely shocking but is slightly disappointing nonetheless. And with that, the conference portion of the weekend concludes. As we all file out of the MGM Grand ballroom, everyone receives a Certificate of Completion from the Financial Modeling World Cup. They all look like college diplomas, and mine congratulates me on “successful completion of financial modeling & Microsoft Excel training.” A few hours and a brisk walk down the Vegas strip later, I arrive at the night’s chosen venue: the HyperX Arena, a 30,000-square-foot esports space with seating for 70, dozens of gaming PCs, a bar, and a large stage in the center. On a typical day, the arena might be home to a Fortnite tournament, but tonight, it will be spreadsheets. The finals don’t start until 7:30PM, but by about 3PM, the arena is already buzzing. Max Sych, the FMWC’s chief operating officer, is onstage polishing the three trophies for the top finishers when he sees me and offers to give me a tour of the arena. He leads me through the room where all the arena’s producers turn its many cameras into a single livestream. We walk through the neon-green Hype Tunnel, where each competitor will enter dramatically and then pause for a selfie before they take the stage. We peek into the VIP room, where sponsors will watch the event. Finally, we land in the commentary booth, where Oz du Soleil and Jon Acampora are already talking through their plans for the evening. These are two of the best-known figures in the Excel world and the two men responsible for helping viewers make sense of fast-paced spreadsheeting. Acampora makes his prediction: he thinks Early is likely to be in the top two, along with Andrew Ngai, an Australian actuary who won the competition both of the previous two years. At about 6:15PM, the arena’s doors open. The contestants file in, clamber up onstage, and find their PCs. Many of them are wearing Excel Esports jerseys, which I’m just now learning are a thing; they look like soccer jerseys, with the name across the front and sponsor logos all over. A couple players wear green T-shirts that say “I simply” above the Excel logo. (Get it?) A few wear their own clothes. As a group, they look more like a bunch of friends heading to the bar to watch their favorite team’s game, rather than some of the world’s foremost athletes in an up-and-coming global sport. Group by group, the finalists sit down and begin to prepare their computers. There are eight workstations on the stage for the two semifinal heats. The players aren’t allowed to use premade lambdas or formulas for tonight’s event; they’re all using identical PCs with identical copies of Excel. They are, however, allowed to use their own keyboards, so a number of them start swapping their own for the glowing RGB mechanical sets on the desk. Sych tells them all they’ll be required to wear headphones so as not to hear the commentary or other noise in the arena and tells them they’re only allowed to use YouTube for music. A few start loading playlists — I notice a surprising amount of Taylor Swift. At least one contestant slyly downloads Spotify. Once they’re done prepping their stations, some of the players mingle onstage, talking strategy and swapping stories of cases past. Ngai, with a hoodie on over his yellow jersey, sits in the front row of the bleachers eating an energy bar. And Patrick Chatain tells me his strategy for the evening. Chatain, a senior studying deep learning at McGill University, is the youngest competitor here but says he’s really not feeling much pressure. Last week, he won the individual competition at the Excel Collegiate Challenge after coming in second the year before, “and this was my last chance at that one.” Now he’s in the big leagues and has his whole career ahead of him. Nobody practices Excel to get famous For tonight, Chatain’s plan is to lead with speed. He says he’s hoping to answer the bonus questions — which only a certain number of people can get each round — right away, just to lock in some points. He’s been studying the others and has decided his best chance is not to try and build a perfect system but to just start sprinting from the beginning. It’s a bold strategy, but a necessary one, because the rules are different tonight. Instead of giving every competitor 30 minutes to get as many points as possible, which is how virtually every previous competition has operated, tonight, the person with the lowest score will be eliminated every seven and a half minutes. The players have been telling me — and each other — all week how much this changes things. There might not be time to build a perfect system, because you might already be out; even the game’s most thoughtful and deliberate players will have to play like there’s a fire chasing them down. A lot of players are thinking like Chatain, hoping to score a few quick points or get a bonus to buy some time. At 7PM, the doors open again, and the fans stream into the 70 seats in the center of the room. They’re mostly friends, family, and fellow competitors — Lau walks in waving an enormous American flag, Scharl’s family starts yelling his name as soon as they see him. I see two signs for Pomykalski: one reads “Jakub the Polish Punisher’’ in cartoonish blue and green letters, and the other says in red, “Jakub never #REFs.” (That is a very good Excel joke.) One person holds a handmade sign featuring Clippy saying, “Go Dim!” and almost immediately, a “Clippy, Clippy” chant erupts from the crowd. The night is off to a roaring start. At 7:30PM on the dot, Stephen Rose, a former Microsoft marketing manager with a graying beard and a shock of black hair who is the evening’s emcee, quiets the crowd. The lights go down, and the livestream begins. As Rose greets the competitors and viewers, I find Grigolyunovich standing off to the side of the bleachers. He’s been running around nonstop for hours getting set up and finally paused to take it all in. He seems sincerely shocked that this is happening. “Back in the ModelOff days, I was really thinking that ‘this deserves to be an esport, with the crowd cheering and the whole world watching.’” He gestures out at that vision coming to life. The whole world might not be tuned in just yet, but this is a step toward that. Onstage, Rose is bellowing into his microphone, introducing each player with all the gusto of a WWE announcer bringing Roman Reigns into the ring. He has nicknames for everyone: Andrew “The Annihilator” Ngai, Peter “The Swiftie Sensation” Scharl, and Curtis “The Beer Hunter” Landry are a few of his best. Each player has been instructed to come down the Hype Tunnel and pose before making their way onto the stage. The next few minutes provide an exceedingly broad definition of the term “pose.” Scharl mimics LeBron James’ chalk-throwing move, Chatain fist-pumps his way down the tunnel, Ngai stops and grinningly waves at the camera with both hands, and several competitors just walk casually past the green lights. Eventually everyone gets settled, and the competition begins. The night’s first case is called Excel Superheroes, designed by Grigolyunovich himself. In it, there’s a cast of superheroes, each with different strengths and weaknesses, and each player’s job is to forecast how they’ll do competing against one another in lots of different scenarios. Rose says go, and the most problematic thing about competitive Excel becomes blindingly obvious to me once again: it is damn near impossible to figure out what’s going on. All eight players are moving so fast and doing so many things with keyboard shortcuts and formulas that there’s practically no way to see what they’re doing until it’s already done. What’s happening around me looks like a sport, it’s lit like a sport, and the anxiety levels suggest aggressive competition, but even the other competitors in the room can barely keep up. They’re squinting at the screens in front of each workstation, trying to decipher each move. Really, they’re mostly just waiting for the score to update. In the commentary booth, du Soleil and Acampora are doing their best to keep up and explain the maneuvers, but watching eight spreadsheet whizzes simultaneously requires multitasking brainpower I’m not sure any human can attain. And if you can figure out what =SUM(CODE(MID(LOWER(SUBSTITUTE(SUBSTITUTE(C3,”:”,””) means in the few seconds it’s shown onscreen, well, you should come to Vegas next year. Like the rest of the crowd, I mostly resort to scoreboard-watching and cheer every time the commentators go back to show a particularly cool formula or trick a player did. The scores change wildly and frequently: a lot of players tend to solve a whole level at once, adding dozens or hundreds of points to their total in an instant. They’ll jump from last to second to sixth to first, and as the clock ticks down, things move even faster. Even if you don’t know what’s happening — and I really don’t — it’s still pretty intense. There’s also drama immediately. Ngai, the reigning champ, is eliminated after the first seven and a half minutes. But when Rose taps him on the shoulder to let him know he’s done, Ngai can’t believe it. He’d gotten a bonus question and solved the first two levels already, and he was in last? From my view in the stands, he had a point: the scoreboard had shown him at the bottom of the scoreboard even as it showed him completing things, then it briefly jumped him to third place, before knocking him down again. Ngai walks offstage, incredulous, and is obviously mad as Grigolyunovich quizzes him to figure out what happened. For the rest of the heat, the competitors seem to be constantly checking the scoreboard. Every time Chatain submits an answer, he spins around in his chair to make sure it shows up; Scharl and others also pointedly check to make sure they’re getting things right. In the last minute or so of the round, the rankings jump around, and with only a few seconds before the round ends, Pomykalski suddenly puts 204 additional points on the board, dropping Chatain from fourth to fifth and out of the competition. As du Soleil screams in disbelief, Chatain appears shocked. “I looked at the screen when I heard a shout” with about 30 seconds remaining, he tells me right after he walks offstage, “but the audience was on the screen! When the scores came back, there were only two seconds left.” The second semifinal, a case based on a bunch of numbers games, is less chaotic. Lianna Gerrish, the only woman in the top 16, is eliminated first. Early, the favorite to win, writes a wildly complicated two-line formula into a cell that ends in six parentheses and impresses all the experts in the room. There is a pair of brothers in this heat — Harry and Dan Seiders, who I’m told are very competitive with one another — and at one point, Dan is eliminated, but Rose accidentally taps Harry on the shoulder instead. Early wins, to no one’s surprise, and the eight-person final is set. We take a brief break, and then it’s time for the last round. After much deliberation with the organizers and the other contestants, Ngai is allowed back in. Whatever went wrong was about the score-keeping system, not Ngai’s answers. But now there are nine competitors and eight computers, so Scharl is shunted offstage to one of the computers around the arena, where he dutifully reloads his setup, searches “Taylor Swift playlist” on YouTube, dons his headphones, and signals he’s ready to go. The crowd chants, “Three, two, one, Excel!” and the final begins. The final case comes courtesy of one of the event’s sponsors, video game Eve Online. In it, each competitor is mining asteroids, building fleets of spaceships, and dealing with market prices for various materials. Their job is to calculate various requirements and costs for building a single ship, then for building the whole fleet, then actually working out how to acquire minerals from asteroids around the universe. It’s hugely complicated: Brandon Moyer, the first finalist eliminated, tells Rose that “I have no idea what that case even was.” But the others charge along, at least for a while. By about the fourth of the case’s seven levels, nearly everyone seems to stall out — some go back to their initial model and try to solve the whole thing a new way, while others just start trying to attack bonus questions. Early, at one point, looks back to check the scoreboard, only to see his own face on the big screen — he’s just been eliminated in fourth place. As he stands up to leave the stage, Rose says he looks a little frazzled. How does he feel? “A little frazzled, yeah,” Early replies. The final three players are Ngai, Michael Jarman, and Willem Gerritsen. Ngai took a commanding lead almost immediately in the round and never let up. With about 50 seconds left in the round, Gerritsen, who has been staring blankly at his screen for a while, just throws his hands up and gives up. A few seconds later, Ngai takes off his headphones and looks over at Jarman, making sure there’s no magic last-second comeback in the works. There isn’t: Jarman stands up and, along with Gerritsen, starts congratulating Ngai. Rose counts down the time, and his voice cracks as he shouts for “our new Microsoft Excel World Champion, Andrew Ngai!” Ngai walks to the front of the stage, waving to the crowd, before thanking all the other competitors and brushing off the “technical difficulties” from the semifinal. “For the team that organized it,” he says, “don’t feel bad; these things happen. It worked out in the end!” For such a huge moment, it’s over pretty fast. There’s no drawn-out celebration, no “I’m going to Disney World!” moment, no post-game analysis. There’s just a quick trophy presentation for both these finals and the financial modeling competition, which Lau won — he goes up and collects a trophy he previously told me he didn’t even want, along with a championship ring I’m told he specially requested. Then Ngai is given his trophy and the large championship belt, which is so big it barely stays in place around his waist. Rose thanks everyone in the arena and on the livestream, tells people to come win the championship for themselves next year, and says goodnight. The players migrate over to the bar, and before long, folks start to filter out. They have jobs to get back to, family vacations to continue in Vegas. Just before I leave the arena, all the finalists, broadcasters, and organizers gather onstage to take a few pictures. Du Soleil crouches in front of the group, and Ngai holds up his trophy. Lau, still in sunglasses and still with the American flag draped over his arm, holds up his prize, too. This group will go down in history as one of two things. They could be the first generation of a new sport, the ones who turn Excel from a work tool to a playing field and change the way the world looks at spreadsheets. Or they could be just a group of friends and colleagues who like to play games together — but instead of playing Fortnite or Catan, they play Excel. Like a lot of folks in the room that night, I think I’ve come to hope for the first outcome but would bet on the second. These are the world’s best spreadsheeters, able to turn a chaotic universe into rows and columns and then bend that universe to their will, but the prize for Excel excellence is much higher at the office than it will ever be in the arena. Even the competitors mostly seem happy to spend a weekend doing a Tiger Woods impression before going back to their real lives and real jobs. Nobody practices Excel to get famous. Grigolyunovich, meanwhile, is already thinking about next year. There will be bigger prize pools — though probably still not enough to get Lau to play — plus more wrinkles in the earlier rounds and more chances for new people to get involved. But tonight, in an honest-to-goodness arena, there was a cheering crowd and real stakes and nonstop drama. Maybe the Excel World Championship will never be mainstream. Maybe the celebrities in this room won’t ever be celebrities outside of it. But Grigolyunovich proved something to himself tonight. Spreadsheets can be anything. They can even be sports.

"We have people stealing all over the world." A digital sleuth named Bryan Hance has spent the past four years obsessively uncovering a bicycle-theft pipeline of astonishing scale, running from the US West Coast to Mexico.

In which Robin Sloan writes Moonbound—a science fiction book about science fiction—and our writer writes his way into total insanity.

A year after OceanGate’s sub imploded, thousands of exclusive leaked documents and interviews with ex-employees reveal how the company’s CEO cut corners, ignored warnings, and lied in his fatal quest to reach the Titanic.

Eliot Higgins and his 28,000 forensic foot soldiers at Bellingcat have kept a miraculous nose for truth—and a sharp sense of its limits—in Gaza, Ukraine, and everywhere else atrocities hide online.

After nearly a decade of silence, the beloved sci-fi author opens up about loss, love, and a collaboration with Keanu Reeves. A WIRED exclusive.

AI is elbowing its way into ever more jobs, but while how we work will change, people skills will still be the most important factor.

Fragmented and focused social platforms might be good for helping you find a knitting community. But extremist groups are also using them to normalize darker content.