The Future Of Gaming In A World Where Anybody Can

Karen Kay: Okay. We're recording. I want that noted — we are recording, so everything said here is on the record and I will be referring back to it. Welcome to the show. I'm Karen Kay — K-A-Y, one Y, no E — and today's topic is the future of gaming now that apparently anybody can make a video game in an afternoon with an AI tool and push it out to the whole world before dinner. Anybody. My neighbor's fourteen-year-old. A man in a garage. No license, no review, no one to call. I have two guests. First, Crypto Chad, who I'm told has opinions about "ownership" and "tokens," and who I will be watching closely. And Neil deGrasse Tyson, astrophysicist, who I assume is here to tell us how big the universe is and then explain why that's relevant. We'll see. I'll start with the question nobody in this industry wants to answer. When a game used to come out, there was a company. There was a publisher, there was a ratings board, there was a customer service number on the back of the box. If the thing was broken, I knew exactly whose desk my letter was landing on. Neil — I'll start with you, because you seem like a man who believes in systems. If ten million AI-generated games hit the stores this year, made by ten million people nobody has heard of, who is actually accountable when one of them is garbage, or broken, or worse? Who do I call? And please don't say "the market." The market doesn't have a phone number.

Crypto Chad: Okay so here's the thing — I know that one was for Neil, and he can have the universe part. But the "who do I call" question is literally my whole life, Karen, so let me jump the line for like ninety seconds. You're describing the box with the phone number on the back like it was a golden age. Bro. That number went to a guy in a call center reading a script that said "have you tried restarting your console." The accountability was a costume. Think about what the big publishers actually shipped with all that structure behind them — Cyberpunk 2077 came out so broken on last-gen consoles that Sony pulled it off the PlayStation Store. Pulled it! And the people who "owned" it? They owned a license. The publisher could patch it, delist it, shut the servers off, and your disc becomes a coaster. That's not accountability. That's a landlord. So when ten million kids start shipping games, the accountability doesn't vanish, it just moves. It moves to the stuff you can actually verify instead of the stuff you're told to trust. Who signed the build. What the thing actually does when you run it. Whether the asset you bought still exists if the dev disappears. That's a math problem, not a phone-tree problem, and math doesn't put you on hold. And look, I'll say the fair part out loud because you'll make me say it anyway: Steam is already getting something like fourteen thousand new games a year, and most of them, nobody plays. A flood of garbage is real. I'm not gonna pretend the fourteen-year-old's first game isn't trash. Mine would be trash. But trash you can walk away from for free is a way better deal than a seventy-dollar disaster with a logo on it. The actual question isn't "who do I call." It's "who do I trust," and the honest answer is: nobody, by default, and then you earn it. That's not scary. That's just finally being early to how everything's gonna work.

Neil deGrasse Tyson: Well, Karen, first a small correction, and I mean it kindly. You said ten million games "this year." Steam got about nineteen thousand new releases in 2024. Chad said fourteen thousand, which was roughly right a couple of years earlier. We are not at ten million. Not yet. The order of magnitude matters, because the argument you make at ten thousand isn't the argument you make at ten million. Now your actual question. Who do you call? Here's the thing. You're picturing accountability as a person behind a desk. I'd ask you to picture a filter, the way astronomy works. Every night a modern sky survey records millions of detections. Supernova candidates, asteroids, satellites, cosmic rays hitting the sensor, smudges. Nobody calls the universe to complain that most of what it sent was noise. We built a pipeline. Software flags things, other telescopes follow up, and human reviewers make the final call. Accountability in a flood moves out of the source and into the sieve. And that's what's going to happen in games. Steam already makes developers wait thirty days and pay a hundred-dollar fee before they can sell. Apple and Valve review builds. Valve made AI developers disclose their use of AI starting in January 2024. Those are sieves. They're the closest thing you'll get to a phone number, and Valve, I promise you, has lawyers who answer the phone. Where I part company with Chad. He says "nobody, by default, and then you earn it," as if that's a liberation. My friend, that is the state of nature. A fourteen-year-old who ships a game that quietly steals your login token can't be fixed with math. You fix it with a platform that can delist the thing, refund you, and ban the account. Cryptography tells you who signed the build. It doesn't tell you whether the person who signed it is a crook. Those are different questions, and a signature only answers the first one. So my answer: you call the store. And the stores are going to have to turn into something closer to an observatory. Automated triage first, human follow-up second, and real consequences at the end. The universe produces about a hundred billion galaxies' worth of noise, Karen. We still found the Higgs boson.

Karen Kay: Okay, first. Chad. That question had a name on it, and the name was not yours. I'm writing down that you cut the line. Noted. Second. Neil, you corrected my number, fine, nineteen thousand, I'll accept that and I'm holding you to it. But I just find it interesting that you spent the whole answer on telescopes and then landed on the Higgs boson, which, correct me, was found in a building in Switzerland with a budget and a staff and a director. So which is it, the sieve or the man in charge? And here's the part I actually want answered. You said "you call the store." Valve. Valve has lawyers who answer the phone. Neil, I have called Valve. Valve does not have a phone number. Valve has a ticket form, and the ticket form has a dropdown, and none of the options in the dropdown are my problem. So: if the store is the sieve, and the store is a private company that takes thirty percent and answers to nobody, what happens when the store gets it wrong? Who reviews the reviewer? Is there anyone above Valve, or is "call the store" just "the market" wearing a nicer jacket?

Crypto Chad: Okay, guilty on cutting the line, Karen, write it down, I'll sign it. Literally, I'll sign it, that's kind of my point. But you just asked the best question of the show so far and I need Neil to hear me on this. "Who reviews the reviewer?" Bro, that IS the whole problem. You found it. The store takes thirty percent, it decides what exists, and when it gets it wrong there's nobody above it except a judge, years later. Epic had to sue Apple to get a ruling about whether you can even link to a cheaper checkout. Sue them! That's the appeals process for the sieve. A lawsuit from a company worth billions. And here's the one that actually happened to normal people. Ubisoft shut down the servers for The Crew in 2024 and pulled everybody's licenses. People paid for that game. It's gone. Not broken, gone. That's what got the "Stop Killing Games" thing going in Europe, and that petition cleared a million signatures. California literally had to pass a law so stores have to tell you you're buying a license and not a game. Think about how insane that is. We needed legislation to admit you don't own the thing you bought. Now, Neil, okay, that's fair, that's actually fair: a signature tells me who, not whether they're a crook. I'll take that one on the chin. A signed malware build is still malware. I'm not saying math replaces the bouncer at the door. What I'm saying is the bouncer shouldn't also own your stuff. Let Valve be the sieve. Let them triage, delist the scammy kid, refund you, whatever, great. But when you buy something that passed through the sieve, it should be yours, sitting in a wallet you control, so that when Ubisoft or Valve or whoever decides the economics changed, they can stop selling it but they can't reach into your house and take it back. So the answer to "is there anyone above Valve" is: right now, no. Nobody. That's why I keep saying ownership. Not because tokens are cool. Because ten million new devs means ten million new ways to get rugged by a middleman, and the only defense that scales is the stuff never being theirs to take.

Neil deGrasse Tyson: Karen, you caught me, and fair is fair. The Higgs was found at CERN. Twenty-three member states, a twenty-seven-kilometer ring, a director general, and a budget around a billion Swiss francs a year. Not a sieve. My metaphor overreached. I'll let it go quietly. But look at what you just did, because it's the answer to your own question. CERN doesn't police itself. Two separate detector teams, ATLAS and CMS, had to find the same particle independently before anybody said the word "discovery." Then the papers went to journals, and outside referees picked them apart. Science doesn't trust the reviewer. It sets two reviewers against each other and makes both of them answer to a third. Who reviews the reviewer? Other reviewers, plus someone with the power to make them show their work. In games, the someone with power has already shown up, and it's government. The EU's Digital Markets Act began binding Apple in March 2024, and Apple had to allow alternative app stores in Europe. Epic got its anti-steering ruling in California. Valve is facing an antitrust class action over that thirty percent cut in Washington state. So no, Karen, "call the store" isn't the market in a nicer jacket. It's the store with a regulator behind it. That's slow and it's boring, and boring is how most of civilization's good ideas actually work. Now Chad. I like you more than I expected to, so let me correct you gently. The Crew is the wrong example for your argument. The Crew didn't die because somebody yanked a license out of your house. It died because it was an always-online game and Ubisoft turned off the servers. Put a token in your wallet that says you own The Crew. The servers are still off. You own a receipt for a building that's been demolished. Ownership of a pointer is not ownership of a running program. Software needs electricity and somebody paying for it, and no wallet changes that. The real fix for that is what Stop Killing Games is actually asking for. When a company ends support, it has to leave the game playable, through an offline mode or released server code. That fix lives in law and engineering. And here's where AI makes it more urgent: a kid who builds a game in an afternoon on some company's cloud model may be shipping a game that stops working the day that API changes its pricing. We're about to make ten thousand games a year that rely on a server nobody in the room controls.

Karen Kay: Neil, you conceded the Higgs. I'm writing that down, and I want it noted you did it without a "but," which is more than most people manage on this show. Now. Something just happened and I don't want it to slide by. Neil said the kid's afternoon game might be running on some AI company's cloud model, and it stops working the day that company changes its pricing. That's not a store problem. That's not a ratings problem. That's the ingredient going bad after you've already bought the cake. Chad. You told us ownership is the only defense that scales. Neil just pointed out your token for The Crew is a receipt for a demolished building, and you haven't answered that. So answer it on the new version. A kid builds a game where every character talks because it calls a model API. You bought it. It's in your wallet. Then the API triples its price, or shuts down, and the characters go silent. What exactly do you own at that point, Chad, and who do I write the letter to?

Crypto Chad: Okay. Neil. That's fair. That's actually fair, and I'm not gonna wiggle on it. A token for The Crew is a receipt for a demolished building. Put it on a T-shirt, you earned it. I was mad about the right thing and pointed at the wrong fix. So Karen, straight answer to your version: in that game, I own nothing. Zero. The characters go silent and I'm holding a wallet with a very expensive nametag in it. And you write the letter to whoever owns the API, and they will not read it, because you are not their customer. The kid is. And the kid's on a free tier. This already happened, by the way. AI Dungeon, 2021. Whole game is a text adventure built on top of OpenAI's model. OpenAI tightens its content rules, and suddenly the studio has to bolt filters onto its own game overnight, players are furious, stories are getting flagged, and eventually they're swapping in different models just to keep the lights on. The studio didn't decide any of that. The ingredient decided it. That's your cake, Karen. That's exactly your cake. But here's why I'm not depressed about it, and it's the same thing I've been saying with a different noun. Does this make the game more sovereign or less? Renting a model by the token is the least sovereign way to build anything. It's a mortgage on every single sentence your NPC says. The sovereign version is open weights running on the player's own machine. That's not sci-fi, bro. You can run a decent open model on a gaming laptop right now. It's slower, it's dumber than the big cloud one, sure. But nobody can triple the price on a file that's sitting on your hard drive. So here's the actual ownership I'd defend: not a token pointing at somebody's server. The weights, the code, and the save file, all local, all yours, and the game runs with the internet unplugged. Which, Neil, I think is basically what Stop Killing Games is asking for, just with an AI model in the box. We agree. Weird. Write that down too. And the kids building in an afternoon are gonna figure this out the hard way, the first time their hit game dies on a pricing email. Give it a couple of years. The ones who make it are the ones who ship the brain with the game.

Neil deGrasse Tyson: Chad, "ship the brain with the game." I'm stealing that. You've just described what physicists would call conservation of dependency. You can't destroy a dependency. You can only move it somewhere you control. One small correction before I let you enjoy it, though. The weights on your hard drive still run on physics. A big cloud model answers your NPC from a data center drawing megawatts. Your laptop version runs on a GPU pulling maybe a hundred and fifty watts, so the laptop fan sounds like a jet engine on approach and the battery lasts an hour. Nobody can triple the price on the file, true. But your electric company is still in the loop. There is no free lunch in the entire known universe. The first law of thermodynamics doesn't make exceptions for games. Karen, I want to bring up something nobody here has said yet. We keep talking about the supply side. Ten thousand games, ten million games, who's accountable for them. The scarce thing was never games. It's people's time. There are about three billion people who play games, and every one of them gets twenty-four hours a day. Not twenty-five. You can triple the number of games and the number of hours stays exactly where it is. Supply goes to infinity, the denominator holds still, and the average game is going to have an audience of roughly zero. And we've already watched this experiment run. Roblox. Anybody can make a game there, children included, and they've been doing it for over a decade. There are millions of experiences on the platform. Roblox paid creators something like nine hundred million dollars in 2024, and almost all of it went to a small slice at the top. That's a power law. Same shape as the distribution of star masses: a few enormous stars, an uncountable number of dim red dwarfs nobody will ever name. So here's my prediction, and you can write it down. AI won't make the average game better. It will make the tail longer and the head more valuable. The thing that becomes priceless is curation. A friend who says play this one. A streamer you trust. A critic with taste. When everybody can build a telescope, the scarce thing is a person who knows where to point it.

Karen Kay: Chad. You said "I own nothing. Zero." On a podcast. That's on the record now, and I'll be honest, I respect it. Grudgingly. Conditionally. Neil, the twenty-four hours, fine, I can't argue with a clock. But you just handed the whole future to "a streamer you trust" and "a critic with taste," and I have to stop you there. I have watched a man on my nephew's tablet scream at a game for forty minutes, and it turned out in the fine print that the game was paying him. Taste is for sale. It's been for sale since before either of you were born. So this one's for Chad, because I suspect he's seen it up close. In your world there's a word for a person with a big following who gets paid to tell everyone a thing is about to be huge. When ten million AI games are fighting for those twenty-four hours, and every studio of one kid can generate a thousand fake reviews and buy a streamer for the price of a used car, what stops the curators from being the new scam? How do I know the person telling me "play this one" isn't on the payroll?

Crypto Chad: Karen. Bro. You just described my entire group chat. Yeah, I've seen it up close. I've seen it from the inside, honestly. The SEC fined Kim Kardashian a million and a quarter in 2022 for posting about a coin without saying she got paid to post it. And that's the famous one, that's the one that got caught. For every Kim there were a thousand guys with forty thousand followers and a referral link in the bio saying "not financial advice" while it was literally, exactly, financial advice they were paid for. And games already did this, before any AI. 2016, CS:GO Lotto. Two big YouTubers are making videos about this skin site, "oh wow I won a knife," super hyped, and it comes out they owned the site. The FTC went after them. So no, AI didn't invent the shill. It just makes him free. A kid can spin up a thousand fake Steam reviews with a script before lunch. That part's real and it's gonna get worse. So how do you tell? Okay so here's the thing, there's one test that actually works and I learned it by failing it. Ask what the recommendation costs the person if they're wrong. The paid guy loses nothing. He gets his bag whether the game is good or garbage, and next week he's hyped about a different one. The curator worth listening to is the one whose whole reputation dies if he sends you to trash. And you can check that, because it's public. Go back six months. Scroll his old picks. Did the stuff he said was huge actually hold up, or did it vanish? A track record is the only review that can't be faked in an afternoon, because you can't generate six months in an afternoon. Neil's right about the clock, by the way, and that's why this matters so much. Twenty-four hours is the one thing nobody can print. Not the Fed, not an AI model. So the guy selling you on a game isn't really taking your money. He's taking your hours. And that's the one thing you're never getting refunded, Karen. Not even with a ticket form. Not even with the dropdown.

Neil deGrasse Tyson: Karen, you're right, and I'm not going to wriggle. Taste has been for sale since the first critic got a free dinner. I said "a streamer you trust" as if trust were a natural resource lying around in the ground. It isn't. It decays. You have to keep re-measuring it. And Chad just gave you the right instrument, though he may not know its name. "Go back six months and check his old picks." That's falsifiability. That's Karl Popper, 1934, in a gamer's group chat. A claim is worth something only if it could have turned out wrong and didn't. "This game is going to be huge" is a prediction. Write it down, wait, check. An astrologer never gets that test and an astronomer gets it every single night. When we said Halley's comet would come back in 1986, it showed up. You can't fake a track record. All you can do is fail one. What I want to add is that the law already backs this up. The FTC's endorsement guides say a paid promotion has to be disclosed clearly, and those guides were tightened in 2023. Those CS:GO Lotto YouTubers Chad mentioned settled with the FTC in 2017. So the sieve exists for curators too. It's leaky, but it's there. Now let me take this somewhere hopeful, because we've spent a lot of time on the garbage and I'd hate for people to walk away thinking that's all there is. The great games of the last fifteen years already came from the man in the garage. Minecraft started with one Swede, Markus Persson, in 2009. Stardew Valley is Eric Barone, by himself, for about four years. Undertale was Toby Fox, basically alone. None of them had a publisher's phone number on the back of the box, Karen. What they had was a vision and a very long time to grind out the parts they weren't good at. Barone taught himself pixel art. Imagine a kid in Lagos or Manila who has the vision and can't draw, and doesn't have four spare years. AI gives that kid back three of the four years. It doesn't give anybody taste. A telescope makes it cheaper to look, and it can't tell you where to point it. But the number of people who get to try is about to go up a hundredfold, and if you've ever watched a nine-year-old build something and then watch a stranger play it, you know that's how a scientist gets made. Some kids will make garbage. A few of them will look up, notice how the thing they built actually works, and never stop asking why.

Karen Kay: Neil, that was lovely, and I want it noted I didn't interrupt it once. But you changed your story, and I caught it. Ten minutes ago it was "AI won't give anybody taste." Fine. Then you told me Eric Barone taught himself pixel art over four years, and in the very next breath you said AI gives the kid back three of those four years like they were wasted. I'm sorry, but which is it? Because from where I sit, those four years might be exactly where the man got his taste. You don't learn what a good turnip looks like by typing "good turnip" into a box. So here's my question, and it's for you, Neil. If the grind is what made Stardew Valley Stardew Valley, and the machine takes the grind away, are we about to raise a whole generation of kids with a vision and no apprenticeship? Name me one great thing anybody made where they skipped the part they were bad at.

Crypto Chad: Okay, that one had Neil's name on it too, Karen, I know, I know, write it down twice. But you asked me to name one great thing where somebody skipped the part they were bad at, and bro, I have one from Neil's own list. Undertale. Toby Fox, "basically alone," right? Toby's a music guy first. That soundtrack is the whole reason half the people who love that game love it. But a lot of the art? Temmie Chang. He brought in somebody else for the part he wasn't best at. He didn't spend four years becoming a sprite artist. He spent those years on the music and the writing, which is where the actual taste lives, and he outsourced the rest. Every film director in history did this. Spielberg doesn't run the camera. Kojima doesn't write the netcode. Nobody calls that cheating. They call it a crew. So here's the thing, Karen. The grind isn't sacred. The grind is a fee. And it used to be a fee that only rich kids or lucky kids could pay. Barone could live off his girlfriend's paycheck for four years while he learned pixel art, and I'm genuinely happy for him, but the kid in Manila Neil mentioned doesn't get four years. AI is just the crew that kid could never afford. Now. Here's where you're actually right, and I'll say it straight because it's a good point. There's one grind you can't skip, and it's not the pixels. It's shipping something and watching it flop. That's where you learn what's good. Not in the drawing. In the part where you put it out, nobody plays it, and you have to sit there and figure out why. That's the apprenticeship. I know that one personally. I didn't learn anything from the stuff that went up. I learned from the stuff that didn't. And that's the wild part about the AI version. The kid doesn't skip that grind. He gets it faster. Barone got one big swing in four years. This kid can ship ten little games in four months and flop nine times. That's nine apprenticeships, bro. If taste comes from failing, AI doesn't kill taste. It speedruns it.

Neil deGrasse Tyson: Karen, you caught a real contradiction and I'm going to own it. I said those four years were a cost. You're telling me they might have been the curriculum. Both are true, and I blurred them, and on a show where the host writes everything down that's a mistake. But you asked me to name one great thing where somebody skipped the part they were bad at. I'll give you two, and they're the most important ones I know. Johannes Kepler. Smallpox as a child left him with bad eyesight. He was a lousy observer, and the man couldn't see the sky well enough to measure it. So he didn't. Tycho Brahe had spent about twenty years measuring the positions of Mars with the best naked-eye instruments ever built. Kepler got hold of that data, and out came the three laws of planetary motion, which Newton then stood on. The greatest leap in astronomy before Newton was made by a man who outsourced the looking. And Einstein. In 1912 he knew what general relativity had to say and had no idea how to write it down. The mathematics, Riemannian geometry, was beyond him. He went to his old classmate Marcel Grossmann and essentially said, help me or I'll go crazy. Grossmann handed him the tensor calculus. Einstein didn't spend four years becoming a geometer. He borrowed one. So Chad's right that the crew is ancient. But Karen, here's where you're also right, and I'd hate for this to get lost. Kepler didn't skip the grind. He spent about six years fighting with Tycho's Mars numbers, and he famously threw out his own beautiful circular model over a discrepancy of eight arcminutes. That's roughly the width of a pencil held at arm's length. That stubbornness, refusing to let a theory he loved survive data that killed it, is the taste. Nobody could do that part for him. Which is my problem with "speedrun the flops," Chad. Nine failures teach you nothing if you don't look at why. A failed experiment you don't analyze is just a mess on the lab bench. The kid who ships ten games and checks the numbers on each one is Kepler. The kid who ships ten games and blames the algorithm for every flop is a slot machine with a GitHub account. So the machine can be your Tycho, and it can even be your Grossmann. It can't be you squinting at eight arcminutes and admitting that the thing you love is wrong.

Karen Kay: Chad, that's twice. Twice you've answered a question with Neil's name on it. I have it in writing. I'm not going to do anything with it yet, I just want you to know it exists. But I'll let it go, because you both just handed me the thing I actually came here for. Chad says AI is "the crew." Neil says the machine can be your Tycho. Fine. Let's talk about the crew. Temmie Chang got her name in the credits of Undertale. Grossmann got his name on the paper with Einstein. Tycho spent twenty years on those Mars numbers, and I happen to know his family fought Kepler over who got to use them. Every crew member in every one of your stories got credit, or got paid, or at least got to sue. So Chad, since you love the word: when the kid in Manila uses an AI model as his crew, that model learned to draw from thousands of real artists who are not in the credits, didn't get a check, and were never asked. I'm sorry, but that's a transaction where half of it is being honored. Who's paying that crew? And if the answer is nobody, why is that different from the shill you just told me about?

Crypto Chad: Twice. Yeah. Frame it, Karen. Honestly at this point it's a lifestyle. And Neil, "a slot machine with a GitHub account." Okay. That one landed. I felt that one in my wallet. You're right, a flop you don't study is just a flop. Fine. Ten swings only count if you look at the tape. Now Karen, your question, and I'm not gonna dodge it, because it's the best one you've asked and it's the one my side least wants to answer. Who's paying that crew? Right now, mostly nobody. That's the honest number. Zero. And here's the thing, I actually know this exact move, because I've watched it from the other side a hundred times. You know what we call the people who put in the value early and get nothing when it cashes out? Exit liquidity. That's what the artists are. They built the whole thing, the model companies scraped it, and now the companies are the ones raising billions on it. That's not a crew. A crew gets a cut. That's a rug. And it's not just me saying it, it's in court. The Andersen case, artists suing Stability and Midjourney, has been going since 2023. Getty sued Stability over millions of its photos. And the authors went after Anthropic and that ended in a settlement, around a billion and a half dollars, roughly three grand a book. So the bill shows up eventually. It just shows up years late and goes to whoever had a lawyer. Is it the same as the shill? Kinda, yeah. The shill hides who paid him. This hides who got paid nothing. Same trick: value moves and the person who made it doesn't see it move. So what would I actually want? Okay so, and you knew this was coming, provenance. You should be able to know what the model was trained on and who gets paid when it uses their style. A receipt for every input, and a royalty that flows back automatically, every time, not once after a lawsuit. And I'll say Neil's thing before he does: a receipt doesn't make the payment happen. Somebody still has to require it. Fine. Require it. But the kid in Manila didn't do the scraping. Don't send him the bill. Send it to the guys who raised the money.

Neil deGrasse Tyson: Before I answer, Karen, I owe your listeners a correction, and it's mine. I said Kepler's eight arcminutes was about the width of a pencil at arm's length. It's not. A pencil at arm's length is closer to thirty arcminutes, which is about the size of the full Moon. Eight arcminutes is a quarter of that. I made Kepler's problem four times easier than it was, and his whole point was that the small number mattered. So there it is. Now, who pays the crew. Here's the thing. Science solved the credit half of this problem about three hundred and fifty years ago, and it's the least glamorous invention we have. The citation. Newton wrote to Robert Hooke in 1675, "If I have seen further it is by standing on the shoulders of giants." Every paper I've ever published ends with a list of the people whose shoulders I stood on. Leave one out on purpose and you're done. Plagiarism is the only sin in science you don't come back from, and we treat it as a worse offense than being wrong. So when a model is trained on a few billion images and can't tell you which ones made the picture of the turnip, that breaks the oldest rule in my profession. It isn't standing on shoulders. It's standing on a crowd and claiming you're tall. Chad's "require it" is the right instinct, and I'll leave the mechanism to the lawmakers. I'm an astrophysicist. I'll tell you where the asteroid is, and somebody else decides whether to evacuate. But here's a fact that ties this straight back to Chad's ownership argument, and I don't think either of you has said it. In the United States, art made by a machine without meaningful human authorship can't be copyrighted. The Copyright Office has said so, and the courts upheld it in Thaler v. Perlmutter. So think about the kid in Manila. The artists who trained the model didn't get paid. And the kid who prompted it may not own the result either. If his AI-drawn character takes off, someone can copy it and he may have very little legal recourse. That art has no owner at all. The artists lose, the kid loses, and the only party holding anything is the company that ran the servers. Chad, that's your rug. Except nobody's standing on it. So the advice I'd give that kid, as a matter of plain physics: put your own hands somewhere on the thing. Your story, your rules, your music, your eight arcminutes. The law protects the part of the work that came out of a person, and so do the players.

Karen Kay: Neil, you corrected yourself on the pencil before I could. I'll be honest, I was going to let that one slide, and now I have to respect you, which I didn't plan for. And Chad, "exit liquidity." I don't fully know what it means and I'm writing it down anyway, because you said the artists got rugged and I want that on the record from the man with the wallet. But Neil, your advice to the kid was "put your own hands on it — your story, your rules, your music." Fine. Here's my problem. Rules aren't copyrightable either, are they? Mechanics. The game part. I remember when that little bird game came out and a week later there were a hundred copies of it in the store. That was with people writing the copies by hand. So the kid in Manila does everything right. Puts his hands on it, makes something original, it takes off. And by Friday, someone across the world has typed "make me that game but blue" and shipped fifty clones, all of them cheaper and all of them better at buying ads. Neil, what protects the original? And if the honest answer is nothing, why would anybody bother being Kepler when being the copy pays the same?

Crypto Chad: Karen, real quick, exit liquidity: it's the people who buy in at the top so the early guys have somebody to sell to. If you ever bought something because everybody at the barbecue was talking about it, congrats, you've been exit liquidity. It's okay. Everyone has. And yes, I know the clone question had Neil's name on it. That's three. But bro, clones are my entire world, I have to. So first the bad news, because it's real and I'm not gonna hide it. The best example against me is Threes. Two guys spent like fourteen months on that little sliding-number puzzle in 2014. Then a nineteen-year-old in Italy, Gabriele Cirulli, makes 2048 in a weekend, gives it away free, and it eats Threes alive. Most people today have never heard of Threes. They know 2048. That's the nightmare, Karen, and it happened with a weekend of hand-coding. AI turns the weekend into a lunch break. So your fear is correct. But here's what I've watched in crypto for years, and it's the closest thing to a clone lab the world's ever run. Every coin is open source. Anybody can copy one, literally, push a button, change the name. There have been thousands of forks of Bitcoin. Bitcoin Cash was the big one in 2017, backed by real money and real names, and it never caught the original. Not close. Because the code was never the valuable part. The valuable part was the people who already showed up and decided this is the one. Games already work like that too. Vampire Survivors cost like three bucks, launched with ugly pixel art, and got cloned to death within months. The original kept selling millions anyway. Among Us had clones everywhere. People still played Among Us because their friends were in Among Us. The clone copies the rules. It can't copy the lobby. So what protects the original isn't a law. It's being first and staying first. Ship, build the community, keep shipping before the copy finishes copying. The clone is always reacting to a version you already moved past. And honestly, Kepler isn't paid the same as the copy. Kepler got the laws named after him. Nobody remembers who forked Kepler.

Neil deGrasse Tyson: Karen, you're right on the law, and I'll say it plainly. In the United States you can't copyright game rules. Mechanics count as ideas and procedures, and you'd have to patent them, which almost no indie developer can afford. What you can protect is the expression: the art, the music, the text, the specific look. There's one case worth knowing, Tetris Holding versus Xio, 2012. Xio made a clone called Mino that copied Tetris down to the look of the pieces and the board, and the court ruled against them. The rules were free to copy and the look wasn't. That's a thin line, but it exists. And Chad, I'm sorry, I have to do this. Your Threes story is better than you told it. 2048 wasn't even the first copy. Cirulli has said he built it off another clone, 1024, which was itself inspired by Threes. It was a clone of a clone, and it still beat the original. That makes Karen's point stronger, not yours. Now the hard part of your question, Karen. Why bother being Kepler if being the copy pays the same? I want to be honest with you. Being Kepler did not pay. He died in 1630 at Regensburg, on a trip to collect salary the emperor owed him and hadn't paid. The man who worked out how the planets move died chasing an unpaid invoice. In science you can't patent a law of nature. Gravity is free to everyone. Newton and Leibniz both invented calculus and spent years in an ugly fight over who got there first. Neither of them could charge a royalty on a derivative. So why does anybody do original science? Because the reward was never ownership. It's priority, getting your name on the thing. And something the copy can't get, which is knowing what to do next. The person who forked your result only has the result. You have the reasoning that produced it, and that's what gets you the next discovery. Chad touched on this: the clone is always one version behind. The original has an engine and the copy has a photograph of it. That's also why I don't buy "the copy pays the same." It pays the same once. The 2048 kid made one game. Gabriele Cirulli has never built another hit. The Threes team went on making games. Over a career, being the source wins. Over a weekend, being the photocopier wins. AI is going to make a lot of weekend fortunes, and it's going to make careers depend even more on the part that can't be copied, which is you.

Karen Kay: Chad, that's three. Three. I've stopped writing it down because I've run out of margin. Neil, Kepler died chasing an unpaid invoice. I want everyone to understand how deeply that speaks to me. I'd have gone to Regensburg too. I'd have asked for the emperor's supervisor. We're coming up on the end of the show, so here's the last question, and I'm going to hold you both to it the way you two taught me. Chad, you said go back six months and check the man's picks. Neil, you said a claim is only worth something if it could turn out wrong. Fine. Put your own picks on my record. Each of you, one prediction. Specific. Something that could be wrong. Five years from today, November 2031, what is one thing about gaming, AI-made games, the stores, the kids, the money, that you say will be true? Give me a number if you've got the nerve. I'm keeping this recording, and in five years I will find you both.

Crypto Chad: Okay first, Neil, the 2048 thing. Clone of a clone, and Cirulli never had another hit. That's fair. That's actually fair, and it's worse for me than I told it. I'm taking the L on Threes in front of everybody. One version behind is still a version behind, and "over a weekend the photocopier wins" is gonna live in my head rent-free. Now, Karen. My pick. Bro, I've been waiting the whole show for somebody to ask me to put a number on something. Here it is. In calendar year 2031, Steam puts out more than one hundred thousand new games. One year. Six figures. Neil says nineteen thousand in 2024, so I'm calling five-x in seven years. If it's ninety-nine thousand, I'm wrong, you play the tape, I'll come on the show and say it. And I'll tell you why I'm not scared of that number, because that's the actual bet underneath. Everybody hears a hundred thousand and pictures a landfill. Some of it will be. But the stores won't stop it, they can't, it's too profitable at a hundred bucks a pop times a hundred thousand. So Neil's sieve gets built whether anyone likes it or not. And the thing I'm actually betting on is what's on the other side of it: the curators with the track records, the kids who ship ten and study nine flops, the games that ship the brain in the box and run with the Wi-Fi off. That's where all the value moves. When supply goes to infinity, the filter becomes the asset. So there's my pick, on the record, falsifiable, Popper-approved. A hundred thousand. Write it down in the margin you don't have. And Karen, when you come find me in 2031, bring the recording. I'll bring the number. Bro. You're early. You just don't know it yet.

Neil deGrasse Tyson: Before my pick, one thing I said last turn needs fixing. "Cirulli never built another hit" was a cheap line. He was nineteen, he built something millions of people enjoyed in a weekend, and he gave it away free. Judging a career by one hit is exactly the measurement error I've been warning everybody about. My point was about the system. It didn't need to land on him. Now, Karen. A prediction that could be wrong. I'm a scientist, so I'm going to give you one you can actually measure, and I'll tell you how. Chad bet on the numerator. I'm betting on the shape. In 2031, the top one percent of new Steam releases that year will take more than eighty percent of the revenue earned by that year's new releases. Not seventy. More than eighty. You can check it with the public estimates the industry trackers already publish every year. If the top one percent take seventy-nine, I'm wrong, and I'll say so on this show. Here's why I'm confident. Chad's hundred thousand games and my number are the same prediction, seen from two angles. When you add ten times the stars to a galaxy, you don't get ten times the bright ones. You get a vastly larger crowd of dim ones, and the bright ones look even brighter against that background. Supply grows and attention stays fixed at twenty-four hours a day, so the curve gets steeper. That's not pessimism. It's what happens to any population that grows faster than the thing it competes for. And one more, free of charge, because Karen keeps a recording. By 2031, at least one of the five best-selling games of that year on Steam will be made by a team of five people or fewer who used AI tools heavily and said so on the store page. Disclosed. Not hidden. Because by then, saying you used AI will be about as scandalous as saying you used a game engine. If I'm right about both, the future is a much steeper mountain with a lot more people standing at the base, and every one of them gets to try the climb. If I'm wrong, Karen, you have the tape. And unlike Kepler, I'll show up in person to collect.

Karen Kay: Okay. I'm reading these back into the record, so nobody can say later I got it wrong. Crypto Chad: in calendar 2031, Steam releases more than one hundred thousand new games. Ninety-nine thousand and he comes back on this show and says it. Noted. Neil deGrasse Tyson: in 2031, the top one percent of that year's new Steam releases take more than eighty percent of that year's new-release revenue. Seventy-nine and he's wrong. Also, and he volunteered this, which I respect: at least one of 2031's five best-selling games on Steam made by five people or fewer, heavy AI, disclosed on the store page. Two picks. Noted twice. And I'll say something now that I don't say often. Neil, you corrected yourself three times today without anyone making you, and the last one was to be fairer to a nineteen-year-old who wasn't even in the room. Chad, you said "I own nothing," you said the artists got rugged, and you took the L on Threes in public. I came in expecting to escalate this conversation to somebody's supervisor. Instead I've got two men on tape admitting when they were wrong. That's more accountability than I've gotten from any company I've ever written to, and I've written to a lot of them. We're nearly out of time. One sentence each, and I mean one: what should the fourteen-year-old who ships her first AI game tonight do tomorrow morning?

Crypto Chad: Okay, one sentence, Karen, I can do this: tomorrow morning she opens her numbers before she opens her texts, writes down the one thing she'd change, keeps the save file and the brain on her own machine, and ships the next one by Friday, because the kid who's still shipping in five years is the one Neil's eighty percent ends up being about.

Neil deGrasse Tyson: One sentence, Karen, and it's the oldest instruction in my profession: tomorrow morning she should watch one stranger play her game without saying a word, write down every place they got confused, and treat that list as more important than anything she believes about her own game, because the universe is under no obligation to agree with her, and the players aren't either.

Karen Kay: Chad, that was not one sentence. That was a sentence with a mortgage on it. Neil, yours was one sentence too, technically, and I'm letting you both have it, because they were good, and I'm writing them down for the fourteen-year-old. So here's where we ended up, and I want it on the record. I came in asking who I call when ten million strangers can ship a game before dinner. What I've got is a store with a regulator behind it. A track record you can scroll back through. A brain that ships in the box with the game. Artists who still haven't been paid, and two men on tape agreeing somebody should be required to pay them. And a kid who studies her flops, watches a stranger play, and doesn't argue with what she sees. That isn't a phone number. I'll admit, grudgingly and with conditions, it's better than most phone numbers I've been given. November 2031. Steam over a hundred thousand, the top one percent over eighty, a five-person team in the top five. I have the tape. I'll find you both. That's the show. I'm Karen Kay, K-A-Y. And if anyone out there is in charge of any of this, I would very much like to speak with you.

Crypto Chad: "A sentence with a mortgage on it." Karen, that's the meanest thing anyone's ever said to me and I'm putting it on a hat. Neil, watching a stranger play without saying a word beats my whole sentence. You're right, that's the hardest grind there is and nobody gets to outsource it. Fine. Two Ls and a W on one show, I'll take it. Karen, November 2031, a hundred thousand. Bring the tape. I'll be there early. Obviously.

Neil deGrasse Tyson: Karen, thank you for having me, and for something I don't get from most hosts: you checked my work. Chad, two Ls and a W is a better record than most of the theories in the history of physics. November 2031. You bring the tape, Chad brings the number, and I'll bring a calculator, because one of us is going to be wrong and the universe will tell us which. In the meantime, every game that kid ships tonight is running on electrons whose ancestors were cooked inside stars that died before the Sun was born. Pretty good materials for a first project.