The Enshittification of the Internet  🔗 **Source:** [The Enshittification of the Internet](https://share.snipd.com/episode/c8214a74-eb66-4f72-acc5-f65f3429468e) ## Highlights - **Three-Stage Pattern Of Platform Decline** - Inshittification follows a three-step pattern: platforms are good to users, lock them in, then prioritize business customers and shareholders. - Once both users and businesses are locked, platforms extract maximum rent and degrade the experience. Transcript: Cory Doctorow It's kind of a theory in three parts. There's an observational component about what happens to platforms. So first, they're good to their end users. They find a way to lock them in. When it's hard for them to leave, they make things worse for them in order to make things better for business customers, they also lock in. And once those business customers are locked in, they drain the value off of them to harvest it all for the shareholders and executives turn into a pile of shit. That's the observational part. ([Time 0:05:10](https://share.snipd.com/snip/6696a57c-c59a-4a9e-beb2-31ae3f599049)) - **How Facebook Locked Its Users In** - Cory Doctorow recounts Facebook's 2006 expansion from college-only to the public as the moment users locked themselves in. - That collective-action lock made leaving impractical for social, business, and community reasons. Transcript: Cory Doctorow Sure. It's a good kind of example here. So 2006 was a critical moment for Facebook. Before 2006, you needed a.edu address. You need to be an American college kid to sign up. But then Mark Zuckerberg wants to open up to everyone in the world. So he makes this pitch. He says, come over to my platform. I know you're all using MySpace, right? But I don't know if you realize this. MySpace is owned by this evil, crapulent, senescent Australian immortal vampire billionaire named Rupert Murdoch who spies on you with every hour that God sends. Facebook is not owned by an evil billionaire, and we will never spy on you. And that was a pretty good offer. So we piled into Facebook, and we locked ourselves in. So with other platforms, they have to do exotic things to lock you in, like Uber. They went to this venture capitalist called SoftBank. It's, you know, principal, this guy Masayoshi's son, had $31 billion from the Saudi royal family. They gave it to Uber. Uber lost 41 cents on every dollar they brought in for 13 years, drove all the other cab companies out of business, drove all the other rideshare companies out of business, became the Dominant player, double the price of rides, have the compensation they gave to drivers. That's a big, complicated bet, right? You don't have to do that with social media. With social media, we lock ourselves in through something economists call the collective action problem, right? Which is that like you love your friends, but like even the six people in your group chat can't agree on where you're going to go for a beer on Friday or what board game you're going to play. So when you want to leave Facebook, and some of you are there because like you have a rare disease and that's where everyone else is hanging out. Some of you are there because that's where your customers or your audiences are. And some of you are there because that's how you plan the carpool for your kids' little league. And some of you are there because you emigrated, and that's where everyone from back home is hanging out. It's really hard to convince those people to leave, right? Because they have to convince all those other people to leave. They have to convince everyone else. So this collective action problem glues us there. We take each other hostage. And so that's stage one, right? ([Time 0:09:11](https://share.snipd.com/snip/37ad2179-37e1-4562-8a73-01bcae813a4e)) - **Monopsony Power Explains Supplier Harm** - Platforms turn screws on advertisers and publishers once users are trapped, becoming monopsonists with outsized buyer power. - This monopoly-over-buyers lets platforms extract value from suppliers while degrading service for everyone. Transcript: Cory Doctorow He says, like, I know I told these rubes that we weren't going to spy on them, but of course, we spy on them from asshole to appetite. Give us, like, a small amount of money. We will target ads to them with exquisite fidelity. And we are such kind of conscientious craftspeople that will pay for a whole building full of engineers that will just police ad fraud. So when you give us a dollar to show an ad to a specific user, we'll find that user and shove that ad right in their face, right? Publishers, similar kind of thing. I know we told you that we told these rubes that we were only going to show them the things they asked to see. Actually, put excerpts from your own website here. We'll cram it non-consensually into the eyeballs of people who never asked to see it. We'll put a link back to our own website. They'll click the link, and then you're going to get free traffic. It's like a gift from us to you. So everyone piles in. They become very dependent. And I think that unless you've done econ, we were telling jokes about doing econ classes before, or unless you're running a small business, it's hard to grasp how dependent a supplier Can be on a customer. We all understand monopoly, right? We played the board game on rainy days at the cottage until we wanted to kill our family and ourselves, right? But no one's ever played a board game called monopsony. Monopsony is the kind of twin brother of monopoly. If monopoly is when you have powerful sellers, monopsony is when you have powerful buyers. And if you think about it as by analogy, like imagine there's a block with like five coffee shops on it. And you go and you pick your favorite one every day and you go there for coffee. Now that coffee shop has an office building next door like this one. And 20% of their gross receipts come from the people in the office building. One day that business goes under and those people never come back to the coffee shop. The coffee shop loses 20% of its gross receipts overnight. The coffee shop is probably out of business, right? Jon Favreau In fact, happened during the pandemic. Yeah. It happened to some of of the coffee shops and places around this office when we weren't here for a couple of years. Exactly. Right. Cory Doctorow So, you know, this is how a business that represents only 20% of your sales can be as important as though they had 100%, right? This is how they can put you over the barrel. Whereas you, like the customer, yeah, you're sad that your favorite coffee shop went under, but there's still four other coffee shops in the block. You'll just get your coffee somewhere else. So a 20% share as a monopsonist is basically dispositive. It gives you total control. 20% share as a monopolist is just an inconvenience for your customers, right? So they get this monopsony over the advertisers, over the publishers. They turn the screws on them too. ([Time 0:11:13](https://share.snipd.com/snip/082d146b-ed74-4e69-8f2c-a40911686ed9)) - **Amazon's Search Payola Example** - Doctorow describes Amazon's 'pay-for-play' search where sellers bid to top results, driving up prices and lowering quality. - The top results average 25–29% higher, forcing merchants to raise prices and shrinking their margins. Transcript: Cory Doctorow Search scam. So they call it an advertising program. It's not advertising. It's what if you follow radio for a long time, you know as payola, pay for play, pay for search results. So the top result on an Amazon search results page is there because they paid more than everyone else. It's not because it's the best match for your search. So that top result, now they have to either lower quality or raise prices or both in order to win the bidding war. So that top result on average is 29% more expensive than the best result for your search on Amazon. The top row is 25% more expensive. And to get the best deal, you go about 17 places down on the second screen to get to the best result. Now, when I wrote the book, it was a 30-something billion dollar a year industry. Last year, it was a 50-something billion dollar industry. Just paying for the top results. That's right. This year, they're on track for $80 billion. That is three times the combined gross revenue of all the newspapers in the world. And as my friend Tim Wu points out in his new book, The Age of Extraction, this is a loss for everyone. The merchants lose, the customers lose, the only people who win are Amazon. And because of that most favored nation deal, because everyone has to raise their prices to pay the junk fees on Amazon. It was 45 to 51% of every dollar that a merchant brought in on Amazon was kept by Jeff Bezos. Now it's 50 to 60%. To make that up, they have to raise prices. But because of Most Favored Nation, they have to raise prices at Target, at Walmart, at the factory store, and at your local convenience store. ([Time 0:19:48](https://share.snipd.com/snip/0bc783d4-755b-4aca-ad8d-56fc549948eb)) - **Policy, Not Fate, Enabled Surveillance Ads** - Surveillance advertising and data-harvesting grew not from inevitability but from policy failures and lack of updated privacy laws. - Corporate concentration plus regulatory capture enabled companies to 'do it because they can.' Transcript: Cory Doctorow If you remember the history of Google, it's a good example to work from. So Google didn't know how they were going to make money at first. Actually, if you look at the research paper that they released when they released the first version of Google, there's a section in it that says we're never going to do ads because ads Make search engines crappy. It's great. It's called the PageRank paper. It's a very famous research paper that's just a footnote. It mostly describes this cool Google methodology for finding the best search results, which was an amazing moment. And then, you know, we had the crash of 2000, and they had to figure out how to make money. And they came up with this ad sales technique, AdWords, that actually did make a ton of money for them. They were in the black and doing super well, but they wanted more money. Now, of course, every firm wants more money, right? The ideal firm, if you're like just sort of constructing it as like a physicist going, you know, well, I'm going to imagine a spherical cow of uniform density on a frictionless surface For the purpose of this thought experiment. The ideal company is one that charges infinity for its product and pays nothing for the inputs that make that product. And admittedly, this is the academic publishing industry I've just described, but most other industries have to accept that they can neither charge infinity nor pay nothing because People won't sit still for it. So Google went from this original advertising system, which was a content-based advertising system, where they were targeting ads based on what you were searching for, to a surveillance-based Advertising system. Now, surveillance-based advertising systems did not come down off a mountain on two stone tablets, right? The reason companies are able to conduct mass surveillance and use our data without any limit to speak of is because the last time Congress updated our privacy law was in 1988, when Ronald Reagan signed a law that banned video store clerks from disclosing your VHS rentals. Right? Now, like, that's not a fact of nature, right? Like, we have payment processors on the internet, not the best companies in the world, but payment processors would make a lot more money if they just kept your money when you gave it To them instead of giving it to the person you were hoping to get it to. They don't do that, right? The reason they don't do that is it's felony fraud and they go to prison, right? Why does Google spy on us? The same reason your dog licks its balls, because they can, right? Because we never told them they couldn't and so they do, right? So this is not a fact of nature. It's not an iron law of economics. ([Time 0:25:13](https://share.snipd.com/snip/d61fa0f6-2d43-433c-9ed5-bb2d977009da)) - **OG App's One-Day Success And Takedown** - Doctorow tells of OG app, a teen-built Instagram client that stripped ads and algorithmic boosts and hit the app-store top 10. - Meta quickly issued takedowns and app stores removed it overnight, showing platform control. Transcript: Cory Doctorow You could eat your cake and have it too. Jon Favreau So this is like if you could do this with Blue Sky, for example, and be like, well, I don't want to leave Twitter because there's a lot of stuff going on there and I still have to go. And so you could then bring it all over there. Cory Doctorow You could do it all on device. You could have the Twitter client on your device. And if you can jailbreak your device, you don't even have to jailbreak the app. If you could jailbreak iOS, then you could look inside the packets that are being slurped down from Twitter. And you could put the ones that were relevant to the user, not the far-right influencer crap, but the, you know, whatever matters to you in that feed. In fact, this happened. So in 2024, a couple of teenagers made an app called OG app. And what OG app did was it took your Instagram login and password, impersonated you to the service, grabbed everything that Instagram wanted to feed you, threw away all the suggestions, Threw away all the ads, threw away all the boosted content, threw away everything that the algorithm had surfaced from three months ago because it was clicky, and just showed you the Things from the people that you followed in reverse chronological order. In one day, it hit the top 10 of both app stores. That night, Meta sent a takedown notice to both app stores. Jon Favreau The next morning was gone forever because there was honor among thieves, right? ([Time 0:42:37](https://share.snipd.com/snip/afef09db-ec91-44fa-acb3-4c23f25af41b)) - **Use Enforcement To Deter Platform Abuse** - Enforce existing antitrust and competition laws aggressively to deter extractive platform behavior. - Use investigatory power to punish offenders and create deterrence that opens space for market entry. Transcript: Cory Doctorow Well, a lot of it is just about enforcing the laws that are in the books. The laws are pretty good. You know, one of the cases that just got dropped by Trump's FTC was an incredible case. It was the first Robinson-Patman Act case that had been brought in like 40 years. The Robinson-Patman Act was passed back when AT&T, not AT&T, A&P was blowing up. And small-town grocers were all going under because A&P was going to their, it's funny, there's so many companies that sound like A&P. There's AP, the Associated Press. There's AT&T. A&P went to all the suppliers and they said, you have to charge us less than you charge everyone else and you have to charge them more. And if they discount below what we're selling at, you have to cut off their supply. So the Robertson-Patman Act says you cannot offer preferential discounting. You can do quantity breaks. 10,000 costs this, 100,000 costs that, one costs this. But you cannot say 100,000 costs less for Walmart than it does for everyone else. The sidelining of the Robertson-Patman Act gave us Walmart. You wouldn't have Walmart without it. And this case that was brought was about how Pepsi and Walmart had rigged the game so that Walmart would give Pepsi preferential treatment, all the end caps, the big stacks of soda and Other snacks. They own everything, not just soft drinks, Frito-Lay and whatever. And that in exchange, Pepsi would charge everyone else more than Walmart, no matter what quantity they were buying in. And if they ever caught them discounting below Walmart's retail price, they would cut them off. They would punish them. And this is just an economy-wide tax. You know, we think of Walmart as bringing us low prices. What Walmart has done is raised prices everywhere, including at Walmart, but not as much as everyone else has raised it. Walmart's imposed one of those economy-wide taxes, just like Amazon has. So we have so many laws like this, the Clayton Act, the FTC Act, all of these laws, they're very forceful and they do a lot of good. Now, the Supreme Court narrowed the FTC's ability to do a lot of enforcement in the last session by sidelining something called Chevron deference. But I think that's a little overstated. So first of all, the Supreme Court did not say, and just to lay this out, Congress says, we're going to create an agency called the EPA and we're going to ask it to make sure that our environment Isn't poisoned. What they don't say is we're going to create an agency called the EPA and we're going to make sure that there aren't more than eight parts per million of dioxin in your water. Because Congress doesn't have the expertise to do that. Jon Favreau The federal government is supposed to, the executive branch is supposed to be able to do that. Cory Doctorow It'd be a very long-ass law, right, if you had to do all of that stuff. Now, the Supremes did not say that everything qualifies as a major question that Congress has to include. So there may be some things that they can do. But the other thing is that Chevron deference does not touch what's called a conduct remedy, which is like a penalty for a specific company. So it's about rules, not penalties. So the FTC can't say no one is allowed to do surveillance pricing, where we spy on you, we figure out how desperate you are, and then we offer you a wage or a price based on your desperation. This is everywhere right now. In fact, Google CEO Sundar Pinchai just made an announcement saying, this is the future of AI. We're going to spy on you. We're going to figure out how much you're willing to pay. We're going to change every price you see based on your ability to pay. It's like cod Marxism, right? From each to their ability to each according to their ability to pay, right? It's a very weird turn of affairs, but it's especially bad in labor markets. So nurses, if they're contract nurses, which is like the way hospitals preferentially hire them so they can do union avoidance, they sign on for their shifts using one of four national Apps, all of which kind of advertises itself as Uber for nurses before the nurses offered a shift because we don't have any new privacy laws since 1988, the app can look up the nurse's Credit card history. And depending on how much credit card debt that nurse is carrying, they're offered a lower wage. So the FTC can't say maybe under the major questions doctrine, they can't say this is illegal. What they can say is if we catch you doing it, then we will find you so fucking much money that your grandkids are going to be sad, right? And like they say, sometimes you have to execute an admiral to encourage the others. The way that deterrence works is if the FTC goes in there and just kneecaps some company for doing this, just takes Delta because Delta just announced they're going to do this, right? Goes and investigates because after Delta announced they were going to do it, everyone got angry at them. They said, oh, we're just kidding. You misunderstood. Why are you mad at me? Right? But go and use your investigatory powers. Just open up all of Delta's books. Read every goddamn memo. Put it in the public record. And if you catch them doing it, turn Delta into a smoking ruin so that we can clear up a bunch of routes that are currently being dominated by a big dumb monopolist. And we can have new market entry by other airlines. ([Time 0:49:06](https://share.snipd.com/snip/31b9fac7-f8cb-4773-9482-05494a481b1b)) - **AI's Bubble And Possible Productive Residue** - AI today is massively capital-intensive and loss-making, driven by hype and enormous CapEx that outstrips realistic revenue. - That creates a bubble likely to burst but may leave useful, low-cost open-source residue. Transcript: Cory Doctorow Develop AI so that it's not in shitified? What would people do? So the big problem with AI is that it's just not real, right? No one's ever lost as much money as they have on AI. AI is the money losingest proposition in business in the history of the world, right? And it doesn't make any money. So, you know, the companies have already done like, okay, yet. But let's talk about this. Let's talk about what we mean when we say yet. Six to $700 billion in capital expenditure to date, right? Two trillion to three trillion committed in capital expenditure between now and 2030. It's going to be hard to make up. Right? Gross revenues from the industry by its own account globally, $60 billion. Right? $600 billion in CapEx, $60 billion gross revenues. $60 billion is incredibly inflated. Because in that, just to give you one line item, is the $10 billion that Microsoft gives to OpenAI and OpenAI gives back to Microsoft, right? This is like you walk past the Starbucks and a kid in a green apron gives you a voucher for a $7 latte. You go and you redeem it and they book that as $7 revenue. They're poorer than they were when you redeemed it. They're not richer than they were. They're out all the materials needed to make the latte and the wages of the barista who mixed it, right? So like this is, to call this accounting fraud is to do violence to the noble accounting fraud, right? So you have this giant money losing enterprise. There are like seven companies that are like 33% of the stock market, 33% of the S&P 500. They're just passing around the same $100 billion IOU and pretending it's in everyone's bank account at once. They're insolvent for the most part. When they implode, the market's going to just because you can't vaporize a third of the market overnight. Jon Favreau Well, aren't they in the, to take in certification theory, aren't they in the stage where they are trying to get so many people hooked on the LLMs at this point? And there'll probably be other products, I'm guessing as well. But for now, the LLMs, and then especially you get businesses hooked on it, and then suddenly they're paying a lot more money, and then they're turning the screws. Cory Doctorow Sure. Yeah, yeah. No, that's the theory, right? But the values involved, the dollar prices involved are so high, right? You have to go from $20 a month to $2,000 a month. And for everyone who's nopes out of that, the $,000 figure goes up because their fixed costs are so high, right? So like, I mean, the thing here, I sometimes feel like I'm on crazy pills, right? Because like the people who've lost the least money on GPUs are the ones who forgot that they needed electricity for them. And so they're just sitting in a warehouse. The minute you plug in the GPU, you start losing money. The more you use the GPU, the more customers you have, and the more they use your product, the more money you lose. And it's a lot of money, like a shocking amount of money. They say in finance, Stein's law, anything that can't go on forever eventually stops, right? Eventually, you run into suckers who are going to bankroll this adventure. ([Time 0:54:00](https://share.snipd.com/snip/07857130-5045-49e8-bd92-0d849b048321)) - **Prefer Local, Efficient AI Over Centralized Cloud** - Favor AI that runs locally or as efficient plugins rather than massive cloud monopolies. - Build tools that run on cheap hardware and empower users instead of centralizing control. Transcript: Cory Doctorow Like, I would be really surprised if when you're editing this, someone wasn't running Whisper, which is OpenAI's open source transcription model, to help you build the EDL so that You can search for specific phrases and also maybe do closed captioning and so on. I use it all the time. I had to find a quote for an article I was writing, and I didn't remember which podcast I heard it in. It runs on my laptop, Whisper. It's like, it's not in the cloud. I just downloaded a copy of Whisper. I threw 30 hours of podcasts at it, and then it transcribed all of them. And then I just searched the text, and I found the podcast I was thinking of. Jon Favreau It was great, right? Cory Doctorow That plugin. Jon Favreau What do you mean plugin for people who don't? Cory Doctorow It's just an add-on to your product. If you're sitting there at your Avid suite down the road at Warner's or whatever, and the director comes in and says, God, I wish I'd had those 300 extras look over there instead of over Here when we were shooting that. And you say, oh yeah, no, I got this new feature for my edit suite. I can just move their eyes, right? Well, yes, that is deep faking AI, blah, blah, blah, but it's just a plug-in. ([Time 0:56:50](https://share.snipd.com/snip/9ab0ef52-1d20-4772-aa12-32f017b55c8b)) - **User Control Was The Web's Missing Ingredient** - The early internet's key virtue was user control: users or skilled others could modify or fix services. - Restoring that control is central to rebuilding a better internet that resists extraction. Transcript: Jon Favreau You started working there back in, I think, 2002. What's that work like today? And are you as hopeful about it as you were when you started? Cory Doctorow Hope is the right word. So I'm on the record as thinking optimism is for suckers. Optimism is fatalism. It's the, you know, sunny twin brother of pessimism. The idea that things are going to get worse or better no matter what we do, that like the future is arriving on rails, you can't steer it. And I think hope is the right thing. And hope is the idea that if you materially improve your circumstances, even though you don't know how to get from here to the peak you want to be at, that if you can ascend the gradient Just by one step, you can attain a new vantage point. And from there, you can see terrain that was occulted when you were down here at this lower elevation and new ways up the hill. And the fact that you can't see from here how you get to there doesn't mean there's no path. In fact, there almost always is. It's just going to require a lot of back and forth and stepwise motion. So I have all the hope in the world. I'm not happy, right? But I have all the hope in the world. And when I started at EFF, I think a lot of our mission was about preserving all the good things on the internet and preventing people from undertaking these in shittificatory gambits, Right? Like I started at EFF just like two or three years after the Digital Millennium Copyright Act was passed. No one knew how far it would go, how far this anti-circumvention would go. And a lot of our work back then was like warning people about how bad this was and trying to get Congress to rein it in and trying to get the Copyright Office, who often adjudicate questions Related to it, to rule in a narrow and conservative way rather than the expansive way industry wanted and so on. And so it was about preserving what we had. And we lost a lot of those fights. I mean, not all of them, but we were overmatched. We won critical ones, but we lost so many of them. And now I think that our mission is more about like, what does a new good internet look like? And there were lots of problems with the old good internet. It was hard for normal people to use. A lot of my friends are normal people. I like hanging out with them on the internet. But the part that wasn't bad about the old good internet was that users got to decide how it worked. And if someone tried to screw you, you didn't have to put up with it. Either you or someone who understood the technology better than you could take that 10-foot pile of shit that someone had parachuted into your path and make an 11-foot ladder out of Code that would let you just go straight over it. ([Time 1:01:13](https://share.snipd.com/snip/f451dbff-b50e-47e2-8d5e-24f5f255bb69))