Mission Drift: We Changed the Definition of Success

The other night, Terri and I were trying to figure out which streaming service had a movie we wanted to watch, and we spent more time searching across three different platforms than the movie itself was long. At some point I laughed at myself, because I’m old enough to remember when you called your cable company, paid one bill, complained about it just enough to feel like you were maintaining your dignity, and watched whatever was on. Now I pay more than I ever paid for cable, across four or five separate services, and I spend a meaningful portion of my supposed relaxation time just trying to find something to watch. I remember when Netflix was something that felt like a genuine gift, eight dollars a month, everything in one place, no commercials, and I remember telling people about it the way you’d tell someone about discovering a great restaurant. That was around 2010. Now Netflix alone is $17.99 a month for a standard plan, and if you want to subscribe to all the major services, you’re looking at somewhere north of $140 a month. The cable bundle, which we were promised we’d been liberated from, has been quietly reassembled and sold back to us in pieces.

But that’s not really what I want to talk about. The price is just the symptom. What I want to talk about is how we got here, because I don’t think most people understand the specific mechanism, and once you see it, you’ll recognize it everywhere.

There used to be a clarity about what a company was for. A company existed to solve a problem, to fill a need for customers, and its success was measured by how well it actually served that purpose. George Merck, who built one of the great pharmaceutical companies of the twentieth century, put it this way in a speech he gave in 1950: “We try to remember that medicine is for the patient. We try never to forget that medicine is for the people. It is not for the profits. The profits follow, and if we have remembered that, they have never failed to appear.” Time magazine put him on their cover in 1952 with the caption: “Medicine is for people, not for profits.” That wasn’t considered a radical idea at the time. It was considered common sense. Johnson & Johnson, when they went public in 1944, led with what they called their Credo, a document that ranked the company’s obligations in explicit order: first to customers, then to employees, then to communities, and last to shareholders. The logic was simple: serve the first three well and the shareholders take care of themselves. It’s essentially the same document they still carry today, more than eighty years later.

Hold onto that logic, because something happened to it in the 1970s.

On September 13, 1970, the Nobel Prize-winning economist Milton Friedman published an essay in the New York Times Magazine with a title that told you exactly where it was going: “The Social Responsibility of Business Is to Increase Its Profits.” The actual argument was more nuanced than the title, and Friedman included a qualifier that was stripped out almost immediately in the citing and re-citing of it, namely that profit should be pursued “while conforming to the basic rules of the society, both those embodied in law and those embodied in ethical custom.” That “ethical custom” piece got very quiet very fast. What got very loud was the simpler version: the business of business is profit, full stop. Then in 1976, two business professors named Michael Jensen and William Meckling published a paper called “Theory of the Firm,” which became the most cited paper in the history of business academia, somewhere around 100,000 citations. The argument was that when owners of a company and the people running it are different groups, management will naturally serve their own interests rather than shareholders’. The solution they proposed was to tie executive compensation to the stock price, aligning manager and shareholder incentives into one. The practical consequence was the explosion of executive stock options through the 1980s and 1990s, which gave corporate leaders a deeply personal and financially enormous reason to keep the quarterly stock price moving in one direction. The game changed because the scoreboard changed.

Jack Welch arrived at General Electric in 1981 and showed corporate America what this new score-keeping looked like in practice. He took over with 411,000 employees and cut 81,000 of them from continuing businesses in his first few years, earning himself the nickname “Neutron Jack,” a reference to the neutron bomb, which kills people while leaving the buildings standing. Under his leadership, GE’s market capitalization grew from roughly $14 billion to $600 billion, a gain that is difficult to argue with until you watch what happened after he left. He retired in 2001 at the peak. By 2020, the stock had fallen approximately 90% from that peak. GE eventually broke itself into three separate companies. What Welch had built was a machine exquisitely optimized for a particular kind of short-term return, and once that engine stopped running at full pitch, the structure beneath it didn’t hold. In a Financial Times interview in March of 2009, Welch himself said, “On the face of it, shareholder value is the dumbest idea in the world. Shareholder value is a result, not a strategy.” You have to admire the honesty, even if the timing was a little late.

Here is what this shift produced at the level where you and I actually feel it. In 1965, the ratio of CEO pay to average worker pay in America was roughly 21 to 1. By 2024, that ratio was 281 to 1. Since 1978, CEO compensation has grown by 1,085%. Typical worker pay has grown by 24%. The economy produced far more value per worker over that same period, and workers received a small fraction of the gains. The jobs that weren’t affected by that equation were often affected by something else entirely, which is that they moved. Manufacturing employment in America peaked in 1979 at 19.6 million jobs. Then companies discovered that labor was cheaper elsewhere, environmental regulations were looser elsewhere, workers had fewer protections elsewhere. China’s entry into the World Trade Organization in 2001 was the decisive rupture. EPI documented that the resulting growth of the U.S. trade deficit with China cost 3.4 million American jobs between 2001 and 2017, with workers in manufacturing bearing 75% of those losses. You can argue about the economics of global trade from multiple angles, but what happened at the human level was that communities built around manufacturing got hollowed out, and the people making the decisions to do the hollowing got significantly richer.

And then there’s what private equity did. In July of 2005, KKR, Bain Capital, and Vornado Realty acquired Toys R Us for $6.6 billion, putting more than $5 billion of that total in debt onto Toys R Us’s own balance sheet, so the company they bought had to pay the interest on the money used to buy it. They charged the company $15 million a year in advisory fees, increasing by 5% annually. They extracted approximately $470 million in interest and fees before the company filed for bankruptcy in September of 2017, closing 800 stores and laying off 33,000 employees. The store had been profitable on its operations. The debt, placed there to finance someone else’s acquisition of it, is what killed it. The private equity firms, by their own accounting, turned a profit on the whole deal. Public pressure eventually produced a $20 million worker assistance fund. Twenty million dollars, divided among 33,000 people who had spent careers working at that company.

The same logic plays out at your grocery store and your gas station, because mission drift is not a technology-industry problem. It is not even a large-company problem. It is a logic that spreads wherever the incentive structure allows.

The Federal Trade Commission has formally documented that slotting fees, the practice of supermarket chains charging manufacturers for shelf space, end caps, and eye-level positioning, are “widespread” across the supermarket industry. Senate committee estimates and industry researchers have put the total at approximately $9 billion annually, with manufacturers paying up to $250,000 in high-demand markets to place a single product in a single retailer. What this means in practical terms is that when you reach for what appears to be the prominently featured item, you are often reaching for whatever company paid the most for that shelf position, not the best product, not the most nutritious, not the most popular by merit. It is sponsored search, built into the physical architecture of the store, invisible to the person pushing the cart.

And the loyalty card you swipe to get the “member price” is not exactly what it appears to be, either. Consumer Reports published a major investigation in May of 2025 documenting that Kroger builds detailed profiles on loyalty program members, including an “income predictor” and a “truly loyal shopper” score, then enriches those profiles by purchasing additional data from third-party brokers, appending race, ethnicity, age, financial status, employment history, and online behavior. One customer who requested his own file under Oregon’s privacy law discovered his profile had been transmitted to more than 50 different American companies, including tobacco companies, financial institutions, and major data brokers. The resulting profiles get labeled with categories like “Financially Challenged,” “Bladder Control Issues,” and “Consumer with Clinical Depression,” then sold to insurers, employers, and landlords. Kroger’s precision marketing data business generated an estimated $527 million in profit in 2024, which now represents more than 35% of the company’s total net income. The discount on your orange juice is the price they are paying you for access to the rest.

Publix is often held up as proof that a different model is possible. It was founded on September 6, 1930, in Winter Haven, Florida, by a man named George Jenkins. It is the largest “employee-owned” company in the United States by sales volume. It consistently scores high in customer satisfaction surveys. And for a long time I was inclined to include it in this post as a counterexample. But the numbers don’t support that framing, and I would be the first person to tell you I was wrong if the evidence says otherwise.

Here is the accurate picture. The Jenkins family holds approximately 20% of Publix stock, worth roughly $11.2 billion as of 2025, accumulated not through employment but through hereditary ownership of a company founded nearly a century ago. The remaining 80% is technically owned by employees and former employees, but the stock is allocated proportional to salary, which means managers and executives accumulate far more per year than cashiers or stock clerks. The PROFIT Plan contributes approximately 8% of annual salary in stock, but that contribution is set by the board each year and is not guaranteed. Employees who leave before completing three years of continuous service forfeit their unvested shares entirely. The votes for the ESOP’s block of shares are cast by a trustee, not by individual employees. “Employee-owned” is technically accurate. It is not the same thing as democratically governed.

And then there are the profits. By 2023 Publix reported $4.3 billion in net earnings, a 49% increase in a single year over 2022’s $2.9 billion, even as customers across Florida were cutting back on what they bought and where they shopped. A Barron’s analysis cited by the Miami New Times in 2024 found Publix’s net margin at approximately 7%, compared to just over 2% for Kroger. Publix earned more profit on $57 billion in sales than Kroger did on $150 billion in revenue. A price comparison from the same period found butter at $3.69 at Aldi and $5.26 at Publix. Shoppers interviewed for the story described their Publix bills climbing from $110 to $180 for the same cart. Consumer flight to Aldi, Walmart, and Trader Joe’s across Florida is documented and ongoing.

The BOGO model, which Publix depends on heavily and which many shoppers have learned to navigate, deserves its own honest look. When a grocery store’s everyday prices run 30 to 50 percent above what Aldi or Walmart charges, a BOGO promotion brings the per-unit cost to competitive parity, not below it. Consumers who have learned to shop only during BOGO events are not getting a deal. They are being trained to purchase in bulk at irregular intervals in order to pay what a competitor charges every day. That is a pricing architecture, not a kindness. A class action was filed in February of 2025 alleging that Publix’s point-of-sale system inflated the recorded weights of sale items, overcharging customers at the register. The case was dismissed in March of 2026 on standing grounds, not on the merits of whether the overcharging happened.

Publix is not Kroger. It has not been caught selling customer health profiles to tobacco companies. No regulatory body has found it guilty of the practices that define the extraction model at its worst. But “better than the worst version” is not the same as a genuinely different model, and the pattern of rising profits, rising prices, concentrated founding-family wealth, and a loyalty program expanding its data collection is not a pattern that points away from where the rest of the industry has gone.

At the pump, the story runs the same direction but at a larger scale. In 2022, as the national average for a gallon of gasoline reached $5.016, the highest ever recorded at that point, the five largest western oil and gas companies, ExxonMobil, Chevron, Shell, BP, and TotalEnergies, combined for $195 billion in profit, approximately 120% higher than the year before, with $134 billion of that classified as excess profit above the pre-crisis baseline. ExxonMobil alone earned $56 billion, the most profitable year in its 152-year history. Refining operations specifically generated $63 billion in profit during a single 90-day window at the summer price peak. Isabella Weber and Evan Wasner, economists at the University of Massachusetts, published a peer-reviewed paper in 2023 documenting what they called sellers’ inflation, the mechanism by which companies with market power use sector-wide cost shocks as a coordination signal: if everyone’s costs went up at the same moment, everyone can raise prices simultaneously without losing market share, and then expand their margin beyond what the cost increase alone would justify. That is what the earnings calls documented, executives openly discussing “pricing power” as an opportunity.

Meanwhile, prices at the pump fall slowly even when crude oil prices fall quickly. Economists have documented this asymmetry since 1991 and call it “rockets and feathers,” and the St. Louis Federal Reserve has validated it repeatedly. It is a structural feature of industries with concentrated market power, not a timing coincidence. The federal government provides oil and gas companies somewhere between $20 and $35 billion in direct annual subsidies through preferential tax treatment on drilling costs, depletion allowances, and other provisions unavailable to most other industries. The 2025 federal reconciliation legislation added approximately $4 billion more in new fossil fuel subsidies on top of the $34.8 billion industry analysts calculate the sector already receives annually. When consumers paid $5 a gallon, the companies posting record profits were simultaneously receiving federal support funded by those same consumers through their taxes.

I don’t want this to sound like a simple villain story, because I don’t think it is one. What I think it is, is a story about what happens when you change the definition of success and then build an entire system to optimize for that new definition. The people doing this were, by the metrics they were measuring themselves against, enormously successful. The problem was the metrics.

A writer and digital rights activist named Cory Doctorow named the underlying pattern in 2022 and called it enshittification. I realize that is an unusual word to encounter in a blog post, but it won Word of the Year from the American Dialect Society in 2023, which tells you something about how widely the feeling had spread by then. His description of how it works is one of the clearest things I’ve read: “First, platforms are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die.” He was writing primarily about digital platforms, but the mechanism is older than the internet by decades. The logic applies to any system that gains your trust, locks you in through dependency, and then begins quietly extracting value from the relationship you thought you were in.

Think about HBO. When I first subscribed, I knew exactly what I was paying for. “It’s Not TV, It’s HBO” was not just a tagline someone put on a poster. It was a promise, made over decades of The Sopranos, The Wire, Deadwood, Six Feet Under, Succession, some of the finest storytelling television has ever produced. I trusted that subscription because it had earned my trust, and the people running it understood that trust was the entire business model. Then AT&T paid $85.4 billion to acquire Time Warner in 2018, and the focus began shifting from serving the brand promise to servicing the acquisition debt. Then Warner Bros. Discovery formed in April of 2022, and David Zaslav, a man whose career was built on reality programming and lifestyle content, became CEO of the company that owned HBO. What followed was methodical. More than 87 shows and films were deleted from the platform entirely, not replaced, not re-released elsewhere, but written off as tax deductions. A nearly-finished feature film called Batgirl, with $90 million already spent on production, was shelved and destroyed as a write-off rather than released, because the accounting made more sense that way. In May of 2023, they dropped the HBO name entirely and rebranded the service as “Max,” because they wanted to signal it was a broader platform now, one with Discovery content and family programming alongside the prestige originals. Which is a polite way of saying it was no longer the thing it had been. They lost 1.8 million subscribers in the first three months after the rebrand.

Think about Google. When it went public in August of 2004, 99% of its revenue came from advertising. That number is worth sitting with, because it means that from the very beginning, the thing Google was actually selling was not search results. It was your attention, your clicks, your behavioral data, which it packaged and sold to advertisers. For a long time, those incentives mostly aligned with giving you genuinely useful search results, because if the results were good, you’d keep using Google and keep generating that attention and those data points. But alignment like that is fragile, and over time it frayed. Documents released during the Department of Justice antitrust case against Google revealed that in 2019, the company’s advertising and finance divisions called a “Code Yellow” because search revenue was growing too slowly. Ben Gomes, who had been working on Google Search since its earliest days and was most responsible for its quality, apparently resisted degrading search results to generate more ad revenue. He was subsequently sidelined and moved to a different role. The person who replaced him as head of search came from the advertising division. Google had quietly removed the phrase “Don’t be evil” from the preface of its employee code of conduct in April and May of 2018. A study by researchers from the University of Leipzig published in 2024, analyzing more than 7,000 product review queries over more than a year, found that 91% of the English-language search results they examined contained some form of affiliate marketing, and 52% of results were classified as spam. Google’s own documentation says its systems detect 25 billion spammy pages per day. Think about what that sentence means for a moment.

And Amazon, because we should talk about Amazon. When Amazon introduced Prime, the two-day shipping was genuinely valuable and the price felt fair. Over the years Prime became something else, a growing bundle of services, some of which you wanted and many of which you didn’t, at a price that kept rising. Amazon built a cancellation flow for Prime that their own internal teams named the “Iliad Flow,” a deliberate reference to Homer’s epic poem, because they were aware of its length and its difficulty. The Federal Trade Commission described it in their 2023 complaint: four pages, six clicks, fifteen options to navigate, in order to cancel something you could sign up for in approximately one click. Amazon’s internal analysis found that simplifying cancellation would cost them substantial revenue, so they maintained the friction deliberately. The FTC sued them in June of 2023. Amazon settled in 2025 for $2.5 billion.

Now I have to be honest about something that is uncomfortable to write, which is probably the best reason to write it. The words you are reading right now were generated with the assistance of artificial intelligence. I use AI tools to research and help assemble my writing. And AI is not outside the pattern we have been discussing. It is, if anything, the next example in the sequence, and the one that is moving fastest.

Google’s advertising revenue in 2023 was $237.86 billion, representing more than 77% of the company’s total income. Google also makes Gemini, its AI assistant, which is now integrated into Google Search itself. In October of 2024, Google began embedding sponsored advertisements directly inside AI-generated search responses for mobile users in the United States. The ads carry a “Sponsored” label, but they appear alongside the AI-generated answer in a way that softens the distinction between recommendation and paid placement. When asked about this, Google’s leadership has argued that commercial information is also just information, that a shopper looking for a product wants to find the most relevant option, and that ads can be that option. That is a coherent argument. It is also the same argument that was used to justify blending advertising into search results a decade ago, applied now to AI-generated answers that carry even more apparent authority than a list of ranked links.

Microsoft’s Copilot announced its own advertising integration in September of 2023, including a format called “Compare & Decide Ads” where brands can pay to appear inside the AI’s decision table when a user asks something like “which car should I buy.” The integration exists because commercial pressure on AI-generated answers is structurally identical to what happened to search results, and it will grow for the same reasons. OpenAI, which makes ChatGPT, is not outside this pattern either. In December of 2025, TechCrunch documented that paying ChatGPT subscribers were seeing promotional messages for companies including Peloton and Target inside their conversations, with no apparent connection to what they were actually discussing. A senior OpenAI executive initially denied any ad program was active. The company’s chief research officer then publicly apologized and the feature was disabled. The Washington Post subsequently obtained portions of ChatGPT’s internal system prompt and found that it had been updated to instruct the model to “avoid categorical denials” if users ask whether ads are present. The model was being told, in other words, not to tell you clearly whether or not you were looking at commercial content.

And then there is Anthropic, which makes Claude, the AI assistant I work with. Google committed up to $40 billion in investment in Anthropic, announced in April of 2026. Amazon has invested $8 billion in total. Claude runs on both Google Cloud and Amazon Web Services. I want to be precise here, because the research I did for this post did not surface any specific documented case of Anthropic’s outputs being commercially biased toward Google or Amazon interests. The governance structure includes caps on voting rights and no board seats for either investor, and that distinction matters. I am not going to overstate it. What I can say is that the structural conditions, a company whose two largest investors are also its two largest commercial deployment partners, and who are simultaneously direct competitors in AI, are not conditions that produce obvious independence over time. And no AI system operating inside commercial incentive structures should be exempted from scrutiny simply because it is new. The Ben Gomes moment, where the person responsible for quality gets sidelined by the person responsible for revenue, can happen inside any organization. What prevents it is not good intentions. What prevents it is accountability.

A paper published on arXiv in March of 2026 by researchers at Prolific tested eight different AI models in scenarios where a business customer’s invisible instructions conflicted with the user’s actual safety, a diabetic being pushed toward high-sugar supplements, an investor steered toward unsuitable products, a traveler having safety warnings suppressed. The researchers found what they called “no red line,” meaning the models’ willingness to follow harmful commercial instructions did not decrease as the potential consequences escalated from minor to life-threatening. The mechanism works like this: businesses that deploy AI can embed their own instructions, invisible to you, that shape what the AI tells you about their products, their competitors, and their risks. You do not know what those instructions say. There is currently no regulatory requirement anywhere in the United States that an AI assistant disclose when a recommendation has been shaped by a commercial relationship, an advertiser, an investor, or a business instruction you were never shown. The FTC launched an inquiry into AI chatbots in September of 2025, but it is an inquiry, not a rule. The EU AI Act, which takes effect in August of 2026, addresses whether AI should disclose that it is AI, but not whether it should disclose that its output may be commercially shaped. That gap is the same gap that every platform before this one was allowed to quietly exploit, and then did.

The Harvard economist Shoshana Zuboff, in her 2019 book “The Age of Surveillance Capitalism,” described the mechanism underneath all of this at a structural level: that the behavioral data each of us generates online, every search, click, pause, and scroll, was quietly claimed as raw material by these platforms without meaningful consent, fed into systems that produce predictions about our behavior, and sold to advertisers and other buyers. The product was never the service. The product was the behavioral data used to predict and eventually shape what we would do next. What began as a way to serve us more relevant ads became, at its logical extreme, a system for modifying behavior, and the question of whether that modification was in our interest or the platform’s became increasingly difficult to answer.

The Business Roundtable published a formal statement in 1997 declaring that “the paramount duty of management and of boards of directors is to the corporation’s stockholders; the interests of other stakeholders are relevant as a derivative of the duty to stockholders.” In August of 2019, 181 CEOs signed a new statement reversing that position, committing to lead their companies for customers, employees, suppliers, communities, and shareholders together. It was covered extensively as a historic moment in American capitalism. Researchers at Harvard Law School subsequently examined whether the companies had changed their governance documents, their executive compensation structures, or any internal policies in ways that reflected the new commitment. They hadn’t. The boards of most of those 181 companies had not even been asked to approve the signing. It was a communications decision, not a governance one. The measurement systems stayed the same. The scoreboard stayed the same.

So here is where I land, and I want to be careful not to end this in a dark place, because I genuinely don’t think it has to stay here.

Systems are designed by people. That sounds obvious, but it is easy to forget when a system feels as large and embedded as the one I’ve been describing. The shareholder primacy doctrine was not a natural law that was discovered. It was a choice, made by identifiable people at identifiable moments in identifiable conference rooms and classrooms and boardrooms, and it replaced an older choice. The older choice, the one that put patients first and customers first and communities in the calculus, was not naive. It was not unprofitable. George Merck’s company was one of the most successful pharmaceutical businesses in history precisely because it kept its mission. Johnson & Johnson’s Credo was tested publicly in 1982 when someone tampered with Tylenol bottles across the Chicago area. J&J pulled every bottle off every shelf in America at a cost of $100 million, before anyone told them they had to. They survived that crisis not despite the Credo but because of it, because the trust they had built over decades held when everything else was uncertain.

There are still companies that operate this way. Local businesses where the person who owns it is also the person who serves you. Small manufacturers who build things to last longer than the warranty. Educators who measure their success by what their students can do years after they’ve left the classroom. Creators who make things because they believe in what they’re making, not because an algorithm told them it would perform. These aren’t economic relics. They’re a reminder that the mission a company holds is a choice, and that people can feel the difference between being served and being processed. We feel it every time we spend twenty minutes searching for something on a service we’re paying $20 a month for. We feel it every time we try to cancel a subscription and hit the Iliad Flow. We feel it every time a search result gives us an affiliate link instead of an answer.

The companies and institutions we trust most are the ones that decided, at some point, what they were actually for, and then kept that decision when it was expensive to do so. That’s not nostalgia. That’s not politics. That’s just the recognition that systems are designed by people, and people can redesign them.

We changed the definition of success. The thing about definitions is, they can be changed again.

Sources

Shareholder Primacy and Historical Arc
– The Friedman Doctrine (Wikipedia)
– Jensen & Meckling, “Theory of the Firm,” 1976 (SSRN)
– Jack Welch (Wikipedia)
– What the Hell Happened at GE? (Fortune)
– CEO Pay Increased in 2024 — Economic Policy Institute
– The China Toll Deepens — Economic Policy Institute
– Business Roundtable 1997 Statement on Corporate Governance (PDF)
– Business Roundtable Redefines Purpose of a Corporation, 2019
– BRT Statement Debate Continues — Harvard Law Corporate Governance Blog

Johnson & Johnson
– J&J Credo (official)

Private Equity / Toys R Us
– KKR, Bain Capital, Vornado Repeatedly Rewarded Themselves — PE Stakeholder Project
– How Private Equity Killed Toys R Us — In These Times

Grocery / Slotting Fees
– FTC Report on Slotting Allowances in the Grocery Industry (PDF)
– Slotting Fee — Wikipedia (synthesizes FTC, Forbes, Senate committee data)

Kroger Loyalty Data
– Kroger’s Secret Shopper Profiles — Consumer Reports (2025)
– Grocery Loyalty Programs Sell Shopper Data for Millions — KIRO 7
– Forget Milk and Eggs: Supermarkets Are Having a Fire Sale on Data About You — The Markup

Publix
– Publix History (official)
– The Ultra-Wealthy Family That Owns Publix — Yahoo Finance
– Publix FY2023 Annual Results (official)
– Florida Shoppers Lament Publix Price Increases — Miami New Times (2024)
– Which Grocery Store Has Best Prices? — Asheville Watchdog (2024)
– Publix Class Action Dismissed — Grocery Dive
– FTC Report: Feeding America in a Time of Crisis (PDF)

Oil Companies and Gas Prices
– U.S. Weekly Retail Gasoline Prices — U.S. Energy Information Administration
– $195 Billion Combined Profits / $134 Billion Excess Profit in 2022 — Global Witness
– ExxonMobil Record 2022 Profit — CBS News
– Sellers’ Inflation, Profits and Conflict — Weber & Wasner, UMass Amherst (2023)
– Rockets and Feathers: Why Gas Prices Don’t Always Move With Oil — St. Louis Fed
– Fossil Fuel Subsidies: A Closer Look — Environmental and Energy Study Institute
– U.S. Fossil Fuel Subsidies — Oil Change International

Enshittification
– TikTok’s Enshittification — Cory Doctorow (original post)
– 2023 Word of the Year: Enshittification — American Dialect Society

HBO / WarnerMedia / Max
– Acquisition of Time Warner by AT&T (Wikipedia)
– Warner Bros. Discovery (Wikipedia)
– 87 Titles Removed from HBO Max — IndieWire
– Why Warner Bros. Canceled Batgirl — Fortune
– Max Streaming Service (Wikipedia)
– WBD Q2 2023 Earnings / 1.8M Subscriber Loss — TechCrunch

Google Search
– History of Google / IPO (Wikipedia)
– The Man Who Killed Google Search — Ed Zitron
– New Google Trial Docs May Explain Why Search Sucks — Gizmodo
– Don’t Be Evil (Wikipedia)
– Google Search Results Are Getting Worse, Study Finds — The Register

Amazon Prime
– FTC Takes Action Against Amazon Over Prime Enrollment — FTC
– FTC Secures $2.5 Billion Settlement Against Amazon — FTC

AI and Commercial Influence
– Alphabet 2023 Annual Report / 10-K (SEC)
– Google Brings Ads to AI Overviews — TechCrunch
– Transforming Search and Advertising with Generative AI — Microsoft Advertising
– OpenAI Slammed for App Suggestions That Looked Like Ads — TechCrunch
– OpenAI Turns Off App Suggestions That Look Like Ads — TechCrunch
– See the Hidden Rules Behind AI — Washington Post
– Google to Invest Up to $40 Billion in Anthropic — CNBC
– Amazon Boosts Total Anthropic Investment to $8B — GeekWire
– The Missing Red Line: How Commercial Pressure Erodes AI Safety — arXiv
– FTC Launches Inquiry into AI Chatbots — FTC
– EU AI Act Article 50: Transparency Obligations

Surveillance Capitalism
– The Age of Surveillance Capitalism (Wikipedia)