Cory Doctorow is the author of over 20 books, including The Internet Con, How to Destroy Surveillance Capitalism, and the science fiction novels Unauthorized Bread and Down and Out in the Magic Kingdom. He’s also an activist for civil liberties on the internet with the Electronic Frontier Foundation, and a prolific blogger at his website Pluralistic. Doctorow joined Current Affairs editor-in-chief Nathan J. Robinson and associate editor Alex Skopic to discuss his newest book, Enshittification: Why Everything Suddenly Got Worse and What to Do About It.

(This interview has been edited for length and clarity.)

Nathan J. Robinson

People might have, if they’ve used the internet and various platforms, a vague ambient sense that things they once loved—things that seemed to work—no longer seem to work as well. And you are here to tell them that this feeling is not an illusion, but is, in fact, a real phenomenon, which you have christened “enshittification.” So what’s going on?

Cory Doctorow

Enshittification makes an observation about the pattern through which platforms decay. It hypothesizes a reason why that decay is happening now, and it proposes a solution. So the pattern, I think, is very recognizable: a platform in the beginning is good to its end users; it finds a way to lock those end users in; once it’s hard for those end users to depart, it makes things worse for everyone to make things better for the shareholders and for the executives at the firm.

Robinson

So you’re describing a process by which—you give case studies of Amazon, Google, Facebook, and Apple—they deteriorate or get worse. Now, can you explain a little more what you mean by something getting worse?

Doctorow

So I’ll give you a very concrete example from Google, one that we have hard evidence for. So you may have felt when you use Google that it feels like the search results aren’t as good. There’s actually a reason for this. Last year, the Department of Justice sued Google for antitrust violations; they served discovery notices on Google executives’ memos, and then they published a bunch of them, and in that evidentiary record, we see an epic struggle that played out at Google about the quality of Google Search. 

In 2019, Google had hit a wall. Their search growth had stalled out, for the very good reason that they had a 90 percent search market share. It wasn’t like they could convince more people to use search, and it wasn’t like they could convince us to use search more. We were already typing every fool thing that came into our head into a Google search box. And so how are you going to grow? Well, you could breed a billion humans to maturity and hope they become Google Search customers. That is Google Classroom. It takes a minute for that to mature into a viable growth strategy. In the meantime, they want to keep [Wall] Street happy. They want to show growth. 

And so this guy, Prabhakar Raghavan, who’s the head of Google Search revenue—he’s an ex-Yahoo and ex-McKinsey guy—has a solution. He says, Why don’t we make search worse? So what we’ll do is we’ll take all the things we use to one-shot your query—so you search for trousers, and we also search for pants. You make a spelling error, and we try and correct it. You make a search that relates to something in Current Affairs—someone throws a submarine sandwich at a National Guardsman, you search for submarine sandwich, and we put something about that story near the top. So we’re very likely to get you your response in one go. We take that stuff out. So now you have to search twice. You type “submarine sandwich” and you don’t get your query. You have to type “submarine sandwich National Guardsman DC.” Well, now you’ve done two queries. There have been two chances to show you ads. 

Now, if Google had viable competitors—if it hadn’t been bribing Apple to the tune of $20 billion a year not to make a rival search engine; bought the default search placement on every browser, every mobile carrier, every OS; and bought the shelf space, there might have been some viable competitors to Google. There aren’t any. They’re not just too big to fail now. They’re too big to care. And so when Ben Gomes says, “I would feel icky if we made our search worse,” Prabhupada Raghavan says, “But we will be richer if we make the search worse.” Now, there was a time when Ben Gomes could have said, “I will feel icky, and we will be poorer if we make the search worse.” But once that argument goes out the window, he loses the argument, and that’s why Google Search sucks.

Robinson

Well, they did take “Don’t be evil” out of the corporate charters.

Doctorow

And I have had my fellow leftists say things like, Isn’t this just capitalism? Well, here’s the thing: Google Search used to be better. So if it’s just capitalism, you need a theory about why it was better. And given that we can point to specific policy choices that were made that allowed Google Search to get worse, that allowed them to get worse without losing money, you have to explain why those policy choices were inevitable because we have had different policy environments. This policy environment, historically, is very weird. It is quite enshittogenic. It encourages firms to make bad choices.

Robinson

Another Cory Doctorow coinage.

Doctrow

Oh, I have coinage. We are living in the enshittocene. I’m in favor of disenshittification. My adversaries are anti-disenshittifiers. I can do this all day like a Latin student.

But the fact is that Google, once upon a time, had better incentives, and they didn’t have better incentives because they were better people. They had better incentives because our policymakers attended to the likely outcomes of bad policies, and they made better ones. 

What we did with Google was, we said, No, you can tap the capital markets and buy anyone you want to buy. So Google is a company that has had one successful in-house product launch. It was a search engine that they made a millennium ago, and in the current millennium, virtually everything they’ve made in-house was a failure, and virtually every success they’ve had was someone else’s company that they bought and operationalized. 

They are not Willy Wonka’s Idea Factory. They’re rich Uncle Pennybags, and they bought all the other kids’ toys. And we could have just stopped them from doing that. We did stop many firms from doing that. That is how they became too big to care. 

Robinson

I was trying to remember what else they had made, and I was like, the glasses!

Doctorow

Google Video. Orkut. Google Plus.

Alex Skopic

Google Plus was their Facebook imitator that never went anywhere.

Doctorow

Yes. They’ve had lots of failures, and actually, where they’ve had successes in sectors, it’s often after their own effort failed. So Google Video failed, and YouTube succeeded. That’s someone else’s company. So you see this up and down the stack, from A to Z, as they say in the whole alphabet. 

And it’s not just Google. And the mergers that were waved through were waved through even when the firms in question explicitly said they were for anti-competitive purposes. So Mark Zuckerberg is a guy who’s never had a bad idea he didn’t put in writing, and this should nominally make it very easy to stop him from doing bad things. 

Because this is a guy who’s got a folder on his desktop called "Mens Rea." When he was buying Instagram, his CFO sent him a memo and said, “Mark, why are we spending a billion dollars on a company with 12 employees?” And he said, “Well, I think you’ll find that people like Instagram better than Facebook. They leave Facebook, they go to Instagram, and they don’t come back. We should buy Instagram so that we can recapture those users.” This is a guy who also periodically and frequently put into writing his motto, It is better to buy than to compete. 

So this goes in front of the Obama DOJ, and they’re like, We see nothing wrong with this. So this is like if he’d sent him an email that said, Bob, you know that guy we’re going to kill? For the record, I want you to know this is a murder. And as I type these words, I am premeditating it. And if you cannot stop a merger on that basis, you cannot stop any mergers. And so no one came down from a mount with two stone tablets and said, We’re going to let these companies buy each other. 

Skopic

You mentioned that it’s a long-standing problem. It’s a bipartisan problem. 

And you mentioned in your book that as you were writing it, Trump got elected for his second term and reshuffled a lot of things with antitrust policy, getting rid of [former Federal Trade Commission leader] Lina Khan. And then, of course, all the tech billionaires were there at his inauguration. You had Zuckerberg and his crew just lined up. So how much worse has it gotten since you wrote the book and since Trump came back?

Doctorow

It’s gotten much worse in America, I’ll tell you that much. So I predicted at the time, and I think I was right, that he would practice what I call boss politics antitrust. 

So I don’t think Biden became an anti-corporate guy in 2020 after a lifetime in service of corporations, but I do think that the party had a faction—the Sanders-Warren wing—that was powerful enough that [Biden] had to give these appointments to good people in the competition brief. And what they did was they took a list of all the corporations in America that had violated antitrust law, which is pretty much all of them, and they sorted it by which companies were doing the worst things to Americans. And they started at the top. 

Trump did the same thing, except he sorted it by the people who made him angriest, and they started at the top, and he has been a remarkably cheap date. So when he took office, Juniper and Cisco had proposed a $14 billion merger. That’s effectively all the industrial Wi-Fi access points, all the industrial networking stuff. So if you’re in an institutional building and you look at the ceiling, you’ll probably see a beige disk that’s doing your Wi-Fi; it’s probably either a Cisco or Juniper gadget. 

So his Department of Justice challenged this merger, and they hired a MAGA podcaster influencer for a million bucks. That guy started hanging around the fifth floor of the DOJ building like a bad smell—that’s where the antitrust division is—just grabbing people on their way out of their offices and shouting at them about trying to stop this merger. Meanwhile, Laura Loomer took to Twitter, started @ing Trump, and condemned the number two and number three persons of the DOJ antitrust division. She managed to get them both fired. The merger went ahead; a million bucks to greenlight a $14 billion merger is a steal. 

However, this is a thing that I think a lot of leftists have not clocked—and it’s only gotten more interesting since the Trump years. It is a bedrock of political science. There’s a big, famous 2014 Princeton study where they studied 2,500 policy outcomes and concluded that public opinion has no outcome on policy preferences. Things billionaires don’t want don’t happen; things billionaires do want do happen. 

And yet, for this whole decade, we have seen massive antitrust surges in Canada, where I’m from. (Like all the best Americans, I’m Canadian.) Our Competition Bureau was the weakest in the world. In its whole history, it had challenged three mergers and been successful no times. In 2024, Justin Trudeau, hardly an enemy of corporate power, whipped his caucus to pass a bill making them the most powerful competition regulator in the world. In the UK, the Competition Markets Authority, under a series of completely shambolic Tory governments, did more work on antitrust, on competition, than their predecessors for 40 years. This happened in the EU with the Digital Markets Act and the Digital Services Act. In Singapore, South Korea, Japan, and China. The laws of political science have been repealed; water has started flowing uphill. 

And the thing is that everywhere that is not America has become very cognizant of the fact that U.S. tech companies are an arm of U.S. political power, that the U.S. considers its former trading partners to be rival nations, and that it will use the tech companies to attack them. And everywhere in the world, we have seen a rapid acceleration of muscular antitrust, muscular movements to break away from American tech, to weaken its power abroad. So this is very exciting. 

I wouldn’t wish Trump on anyone, but if life gives you SARS, you make sarsaparilla. 

Skopic

So do you think that that sort of sea change, and that surge in antitrust, is because the constituents of these politicians are demanding it organically? Or are the non-U.S. billionaires worried about their U.S. counterparts? Or is it both?

Doctorow

I think it’s a little of everything. So there’s this idea in finance, Stein’s law, that anything that can’t go on forever eventually stops. I think people are really sick to the back teeth of concentrated corporate power. And I think that while they haven’t put their finger on it, there is a sense that a lot of our problems—genocide, climate inaction, xenophobia, authoritarianism, worsening labor conditions—are all downstream of concentrated power. 

It’s just a sense that we’re living in a new Gilded Age, that something very bad is going on, and that even if you’re not an antitrust wonk, people are like, I think maybe anti-monopoly law has something to do with stopping power from being concentrated. They’ve at least played the board game. 

And so there’s some kind of inchoate but palpable tailwind for this stuff. You can think of politicians as just throwing out lots of stuff and listening for the echo, and so when they say stuff about antitrust, they get a roar. Those people are starting to push, and then I do think there’s a fair bit of billionaire-on-trillionaire violence. I think Daniel Ek, CEO of Spotify, is really angry that Apple and Google are taking 30 cents of every euro he makes. And I think that the owner of Epic Games is very angry that 30 cents out of every dollar they make is disappearing. I think Rupert Murdoch is very fucked off that 51 cents out of every ad dollar is disappearing into Meta and Google. So there is something about this odd coalition, this informal coalition, that has spurred action. I don’t think any of them on their own would have been enough, but I think all of them together are. 

Robinson

Do you see any possibility of solving the collective action problem? There have been efforts to create alternatives to Twitter, and it’s really difficult. People see Twitter as now absolutely just overflowing with Nazism and horrible, horrible slop. 

Doctrow

Slop, porn, scams, bots. Yes.

Robinson

Facebook’s full of AI slop. So there’s a lot of desire, but as you say, the lock-in, the collective action thing, is very difficult to surmount. Do you see all the action as needing to discipline the companies because we’re not going to get people onto something else?

Doctorow

Oh, I think one of the things that disciplines the company is the threat of departure. If you have to worry about someone leaving... take a market that is full of monopolists and would-be monopolists. Mobile phones. And landline as well, although they’re less important these days. A couple of decades ago, we made a law that said mobile phones have to let you take your phone number with you. If you want to move from Sprint to T-Mobile, you do 10 seconds of administrative work and you jump ship. And that has been massively powerful at disciplining those firms. Now there’s no reason you couldn’t apply that to social media, and there are two ways you could apply it.

You could say, like the standard used for mobile phones to do number porting, here’s a standard for allowing people who leave Twitter and go to Bluesky or Mastodon to still see the messages on Twitter that matter to them. 

And a lot of people are focused on building alternatives, and rightly so. I get why they’re doing it, but that’s like saying, “Those people in East Germany have such low standards of housing, so we’re going to build some really good housing for them here in West Germany,” not thinking about how they’re going to get over the wall. And then you’re like, we built all this housing for the East Germans. Why haven’t they moved in?

Robinson

Yes, right.

Doctorow

Mr. Musk, tear down that wall.

Skopic

So another thing that I want to ask about is the enshittification of our technology and the downstream effect that has on our economy, which is also being enshittified. You had a blog post a couple of days ago on Pluralistic about the AI bubble. There was this report that came out a little while ago that said, basically, there has been virtually no economic growth in the U.S. other than AI. And we can see that a big part of enshittification is that these tech companies like Microsoft and Google are just shoehorning AI into things, in increasingly desperate ways, where nobody wanted AI. 

So how worried do you think we should be that enshittification is also going to lead to a sort of cataclysmic, 2008-style economic collapse?

Doctorow

I think this is a very serious concern. I actually write when I’m anxious, so I wrote another book over the summer about AI. It’s called The Reverse-Centaur’s Guide to Life After AI, and it’s about this bubble. And it’s about a fairly sui generis critique. I think I have a somewhat different critique than most people when we talk about AI, particularly about the labor and copyright issues, which I think are very important, but people get wrong. 

To understand the kind of material basis for the AI bubble, you have to understand the paradox of a monopoly and its relationship to the capital markets. So when companies are growing, the market values them very highly. So a growth stock trades at a much higher price-to-earnings ratio. So for every dollar you bring in, your share is worth, say, $10 or $20 or $30. That is your price-to-earnings ratio. The price-earnings ratio for a growing firm is much larger than for a mature firm. And the thing about growth stock is it’s very liquid, and people will accept it in lieu of dollars. And so if you want to acquire a rival firm to grow, or if you’re Mark Zuckerberg and you want to pay some AI scientist $100 million, you don’t use cash to hire those people or buy those firms. They use stock. 

If you make cash on the premises, the Treasury Department will break the door down and arrest you. But you make stock on the premises by typing zeros into a spreadsheet. And so if you imagine a mature firm bidding against a growth firm, the mature firm has to get dollars from creditors or from customers. The growth firm sits down at the spreadsheet and just types more zeros. They always win these bidding wars. 

So growth begets growth, but once growth ends, it’s very hard to start the machine again. And there is a natural end to growth when you’re a monopolist, because once you have conquered your market, you have nowhere to go. So Facebook does pivot to video, and then we have Metaverse, blockchain, augmented reality, extended reality, NFTs, shitcoins—we have all of those things, and then we get AI, and now we get super intelligence.

And I think it’s not necessarily that they’re so cynical that they don’t think that there’s a “there” there. They may believe there’s a “there” there, but that’s not why they're doing it. They’re doing it to find a plausible or credible adjacent market that the Street will continue to value them for. And so that’s the background radiation of AI. Twenty-nine to 31 percent, depending on the day of the S&P 500, is seven AI stocks. Those AI stocks have a collective capital expenditure (CapEx) up in the range of $700 billion—most of a trillion dollars. Their total revenue, by their own account, which is very, very shaky, is $45 billion a year. Nonprofit revenues—$45 billion a year. And that is a bad number. So to give you one example of why that number shouldn’t be believed, Microsoft gives $10 billion worth of compute tokens to OpenAI. These are vouchers that are redeemable for use on Microsoft’s cloud. OpenAI books that as $10 billion in investment capital, and then they give it back to Microsoft, who books it as $10 billion in revenue. So depending on whose books you're looking at, that’s either $10 or $20 billion worth of the $45 billion that the AI industry has earned this year.

Skopic

So they’re sort of passing the same money back and forth.

Doctorow

Yes, and it’s not even money. This is like your kid comes in and says, “Dad, can I have some lemons, sugar, and water?” They open a lemonade stand. No one buys any of the lemonade. So you go out and you buy all the lemonade from your kid, and your kid announces that they’ve made $10 today. This is the level of econ grift that we’re going for. It’s very funny because these are not sophisticated scams. They’re pushing on an open door here, to be fair. The market wants to be fooled.

So how can they possibly go from $45 billion to amortizing the $700 billion investment, which is just for starters? Mark Zuckerberg just said, I’m prepared to lose $200 or $300 billion on false starts in data center buildouts as we try and get there. It would be different if the amortization schedule on these assets was very long—if they had a long duty cycle. But the way that they sweat these assets, they run these GPUs like crazy; in a training run, thousands of $100,000 GPUs will burn out in 53 days. Amazingly enough, they are staking these GPUs as collateral for some of their loans. This is like staking a trout as collateral on your loans. So, yes, the loan is due in 10 years; you can repossess the trout then. This is insane.

So leaving aside all the questions about what AI can do and can’t do, and the election disinformation, deepfake porn, whatever it is, it’s economically terrifying. And I don’t think it’s going to be like 2008; I think it's going to make 2008 look like the good old days. It’s going to be fucking bad. And if you wanted to make a bet on what to do—I have a visiting professorship at Cornell. I was just up there, and one of the people doing their AI strategy said, “What should our AI strategy be?” And I said, you should be drawing up detailed plans for what you would do if GPUs cost 10 cents on the dollar. You could hire as many out-of-work applied statisticians as you wanted, and you had a ton of open-source AI models that did pretty cool stuff and had barely been optimized and had lots of room for improvement that you could probably accomplish on the cheap. That, I think, should be everyone’s AI strategy right now—that and investing in long poles to dig through rubble, looking for canned goods.

Robinson

Okay, encouraging there.

Skopic

Wow. So, as you said, there’s basically no situation you see where this doesn’t go south fairly quickly.

Doctorow

Yes, I don’t know when it’s going to be. The market can remain irrational, blah, blah, blah. But I don’t think this stuff has a future. Not as foundation models. I think the number of foundation models we will have in a few years could be zero, and it’s certainly not going to be the number of foundation models we have today. 

Skopic

I wrote the review of your book for the magazine, and then I was left with a question at the end. You lay out, I think very correctly, the diagnosis. You say all of this is due to the root cause of the companies having no more checks on them, and wanting as much money as possible, and just degrading the experience and raking in as much cash as possible. 

But then for the solution you come to, you lay out these four or five things. You talk about unionization in the tech sector, antitrust policy, and three or four others. And you use this metaphor of the lever: the capitalist goes into his office, and he yanks the shit lever as hard as possible, and all of these other things gum up the lever so it won’t move. And so it seems like the solution you come to is a sort of stalemate, where the capitalists are still there. They’re still in charge of their companies, but they have checks and balances. It’s regulated capitalism. 

We’re a socialist magazine, so I guess the question I have is, instead of a stalemate like that, why should we not be going for total victory? Why should we not just get rid of the capitalists and nationalize these platforms? They position themselves as the public square, so why should we not make them public property?

Doctorow 

So I think there is a role for public intervention here. But I don’t want Viktor Orbán to be my ISP. I don’t want Donald Trump to be my content moderator. And I think there is a way to have public intervention into this system such that you can have a pluralized base of power, some of which is provided through a purely public option, some of which is provided through worker-owned cooperatives, nonprofits, for-profit small firms, tinkerers, and even through large firms, but where the ability for any one of those to do something bad is quite curtailed.

Because we all know about nonprofits that are terrible to work for. We all know about governments that have gone bad. I don’t think there’s a static configuration in which we should trust that, but what we can do is create shared infrastructure. So let’s use, kind of as a metaphor: fiber in the ground. There is no such thing as fiber without government intervention. It doesn’t matter how Ayn-Rand-pilled you are; you cannot get a piece of fiber on the pole outside every person’s house with market clearing. Ask any economist, and they’ll tell you the person who controls the linchpin of that network is going to charge you the entire sum that you expect to bring in from it, minus one penny. Because they got you over a barrel. 

So you have to have some way to clear it through this thicket of property rights. And that’s governments. Governments come in and they create rights of way, and every phone company is a socialist enterprise in that regard. So fiber is very, very fast. Fiber is millions of times faster than satellite. Thousands of times faster than copper. And if you lay conduit, which is literally just a plastic tube that you stick in the ground, you can put more fiber through it, so every strand of fiber can be joined by more. So imagine if we have public investment, both in the rights-of-way and in the conduit, and then we have a public fiber network.

So that can be your ISP. You can sign up. Your library might administer it, or your town might administer it. Whatever. You can have something in addition to that, called essential facility sharing. So this used to be a feature of telecom policy in this country. If you remember, up until the middle of the first George W. Bush term, if you wanted to get DSL, which was the broadband of the day, there were dozens of companies that were provided to you because AT&T was required to share its infrastructure with them. It’s called essential facility sharing.

It meant that you could start an ISP, a co-op, a for-profit, whatever, and you could go into AT&T’s data center, and they would just rent you a shelf. You put a switch on it, and you plug into customers’ lines. You physically provide internet service to them. So we can have essential facility sharing in the data center.

Now, what you have is the system where co-ops could provide your internet service. They could be consumer co-ops or worker co-ops. Nonprofits could. A group of friends could be their ISP. You could have the city provide your internet, and here’s a cool thing about the city providing your internet: the First Amendment says that government agencies cannot discriminate on the basis of viewpoint. 

First of all, the government should not be using Twitter or Facebook to make announcements. They should be interacting with the public through a service that they own and operate, so if you’re on Facebook, you can talk to them, but they’re not part of Facebook. They’re running their own public service, but they can also provide it. You can have tribal governments doing it. You could have all kinds of governments doing it. It could be devolved to the city, the state, or the county. It could be a tribal government. It could be schools. You could have many different ways that this could be configured.

And there’s a reason that we operate at different scales, because the kind of institution that you need to adequately govern something at a national scale for 300 million people is very top-heavy. It’s creaky. It’s slow. It’s difficult. We all know what it’s like to try and make national changes. There are things you want devolved to the neighborhood level just because you don’t want to have to stand up an institution capable of mediating between 300 million people to decide what you’re going to do about whether this will be a road with a bus lane on it. 

And so it makes sense to have aspects of this digital world devolve down to the smallest level. The fact that Facebook is a global network is not good. So I guess maybe that’s my final objection. I’m thinking through this now as I speak to you. I think maybe my final objection is the presumption that Facebook should be a utility, which subsumes the assumption that Facebook should be. And Facebook shouldn’t be. It’s not a wieldy scale to do global community function.

Skopic 

So basically, instead of, say, nationalizing Facebook, you would want to devolve Facebook and maybe have a lot of regional Facebooks?

Doctorow 

Yes. Look, I don’t want the government to own all the bicycles. I do want the government to provide the bike lanes and the roads and the cops who will track down the people who hit and run you and do all kinds of other things.

I’m a socialist too, but I think it’s okay for me to have my own bicycle. And famously, Steve Jobs called a computer a bicycle for the mind. I think that we should have as much centralization and public management as we need, and not one bit more.

Robinson 

Let’s conclude it there, because that’s very positive. You gave us a very dark look at the future, but you [also] tell us that we do have the capacity to disenshittify, to truly innovate.

Doctorow

And exit the enshitticine and banish the enshitternet to the scrap heap of history.

Transcript edited by Patrick Farnsworth.