Software on David Crush

Keep Working on the Plan

David Crush — Fri, 23 Oct 2026 12:00:00 -0400

The first plan I saw in Cursor was a numbered list. It was not code. It was what the agent intended to do, file by file, step by step, and then it waited. I could edit it. I could argue with it. I could leave it overnight and come back to it in the morning. Nothing ran until I said so.

I started where a lot of people did. Tab completion first, then GitHub Copilot in VS Code. Autocomplete was useful. It was not a way of thinking. The plan was.

I am cancelling Cursor. I paid for two things: a plan I could keep working on before anything executed, and the freedom to hand that plan to the best model for the job. Plan mode is still there, and it is still my favorite thing in any AI tool. The choice is being taken apart so Cursor can be a Grok harness, and a plan is only as good as the model that executes it.

The Plan Before the Code

My brain does not hold a long invisible sequence well. If the steps live only in my head, some of them fall out, and the anxiety fills the space where they were. I wrote about that move in Filling in the Blanks: the blank is intolerable, so something fills it. AI fills it toward the average. My brain fills it toward catastrophe.

A plan on the screen takes the blank away. I can see the whole shape of the work before any of it happens. I can catch the step where the model assumed something about my repo that is not true. I can cut a step, add a constraint, ask why. The plan is where I do my thinking, and the model does the typing after I have finished. That is the difference between a tool I use and a tool that uses me.

Then the second half. Once the plan was right, I could pick who executed it. Opus for the hard debugging. A cheap, fast model for the boilerplate. Composer for everyday work, when it was the best workhorse in the building. And when I did not want to decide, Auto mode decided for me, and it was brilliant at it. One decision I do not have to make is a real accommodation. Nothing else on the market put those pieces in one window.

What Changed

SpaceX agreed to buy Anysphere, the company behind Cursor, on June 16, 2026, in an all-stock deal that valued Cursor at $60 billion. The merger closed August 14. SpaceX already owned xAI, the lab that makes Grok. Cursor announced the same day that Grok 4.6, “which we released Wednesday,” was “an early look at what we can now build together.”

Here is what it looks like from my chair.

New chats open on Grok 4.7. Users on the Cursor forum have reported for weeks that new agents reset to Grok even when they configured something else. Cursor staff explained that this is intentional when the default is set to “Cursor Default,” which follows whatever model Cursor is promoting. One user put it plainly: “A default that ignores my usage history and moves new agents to Cursor’s currently promoted model is not neutral.”

The new plans are built around Grok. The X Subscription launched October 12 and bundles Grok, Cursor, and X Premium. Cursor’s own pricing page says the $8 Plus tier includes Grok models only. On Premium and above, third-party models such as Claude are still available, but at API list price from prepaid credit, with no included allowance. The included pool covers Grok and Composer. Existing subscribers can keep their old plans for now.

OpenAI is leaving. On August 28, OpenAI notified SpaceX that it would wind down its contract with Cursor, with a proposed shutoff of November 12, and that it would not provide future models. The reason was in the second paragraph: “we cannot be confident that SpaceX will use our technology within our terms of service, based on our experience with Elon Musk’s companies violating contracts.”

Composer has not had a new version since 2.5 shipped in May. Grok got three in the same stretch. I do not know what changed around Composer. What I know is my own sessions: the same mistakes over and over, no matter what the prompt says or how specific the rules file is.

And Grok. My experience vibe coding is more frustrating now than it has ever been. Simple mistakes. Instructions read and then ignored. Work I have to check line by line, which defeats the point of a plan.

That is not only my mood. Cursor runs its own benchmark, CursorBench, built from real Cursor sessions. Grok 4.7 at its highest effort setting scores 46.3 percent and costs $6.01 per task. Claude Opus 5.5 at High scores 56.0 percent for $3.97. Grok is cheaper per token. It is not cheaper per finished task, and it lands nineteenth on the board its own owner publishes. Grok 4.7 does top a few leaderboards, including Artificial Analysis’s cyber index and an open benchmark about rebuilding retired programs. I am not doing either of those. I am asking it to follow the rules in my repo.

Elon Musk does not dispute any of this. In an interview with China Media Group this month, he called Grok 4.7 “a solid workhorse of a model” and said, “It’s not as good as, say, Opus 5.5 that just got released.” Then: “I would expect that we probably catch up to Frontier sometime next year, most likely.”

Next year. So why force it on everyone now? If the model will be competitive in a year, wait a year. Let Grok earn the default. Instead, the product got worse on purpose, today, for a model its owner admits is behind.

Here is the part I cannot get past. On October 7, Musk posted about Grok Bot, SpaceX’s own agent product: “Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs. Whatever is most likely to give you the best outcome.”

His product gets the best model. The product he bought gets Grok.

Yet another business killed by Elon. I made the longer case about the distance between what he says and what he does in He Can’t Have It Both Ways; this is one more entry on that ledger.

Rent Seeking

Economists have a term for this: rent seeking, capturing value through control of something rather than by making something better.

That is what it looks like to me. People chose Cursor because it let them pick the model. Musk bought it and used ownership of the harness to steer those people toward his own model. It spends Cursor’s goodwill to pad Grok’s usage numbers. The honest version would be the slow one: make Grok good enough that Cursor users pick it on their own, the way they picked Composer. That would benefit Cursor. This benefits xAI.

I cannot see inside the deal, and I am not going to claim I know why he bought it. I can see the steering, I can read the benchmark, and I can read Musk saying his own model is behind.

Musk has also told people how to judge companies. On May 1, 2022, he gave his followers investing advice: “Buy stock in several companies that make products & services that you believe in. Only sell if you think their products & services are trending worse.”

In this case, at least, he cannot claim he makes a good product. He bought one and made it worse. By his own rule, Cursor is a sell. Cancelling is how a subscriber sells.

What I’m Trying Next

I have been inside the Cursor ecosystem so long that I stopped looking around. That is the good news here. I get to look.

I have not tried most of these yet, so this is a list of candidates and one test, not reviews.

Devin. It is installed. I have barely touched it. From what I can tell, it is the closest thing to Cursor.
Claude Code. Anthropic’s own agent, with Anthropic’s newest models and no middleman deciding which ones I can afford.
OpenAI Codex. Same idea from the other lab. OpenAI’s own help page points Cursor users toward the Codex extension after November 12.
OpenCode. Open source, and built to let you bring the model.
My own VS Code extension. I have thrown this idea around more than once. If nothing gives me the plan the way I need it, maybe I build the plan.

The test for each one is the same two questions. Can I see the plan and keep working on it before anything executes? Can I choose the model? Anything that answers yes to both gets a real trial. Anything that answers yes to one gets a fair look.

That Cursor

I loved Cursor. That Cursor is dead, and I do not see myself coming back. Good luck to the team. They built the tool that made AI make sense to me, and I do not feel good about where it is going.

The last plan I wrote in Cursor is still there, unexecuted, waiting for me to say go. The model picker above it says Grok 4.7. I did not pick it.

Sources

The acquisition:

SpaceX Form 8-K, merger agreement with Anysphere (SEC, Jun. 2026)
SpaceX Form 8-K, completion of the Cursor merger (SEC, Aug. 2026)
Cursor is now a part of SpaceX (Cursor, Aug. 2026)
SpaceX locks in $60 billion Cursor deal (Reuters, Jun. 2026)

Grok as the default, and the new plans:

New chat always selects Grok 4.7 (Cursor forum, Sep.–Oct. 2026)
New Cursor Grok 4.7 vs Composer 2.5 (Cursor forum, Oct. 2026)
Introducing The X Subscription (SpaceXSI, Oct. 2026)
Cursor pricing and Usage and limits (Cursor)
Grok Is Killing Cursor (Vincent Schmalbach, Oct. 2026)

OpenAI’s exit:

Our decision on Cursor following its acquisition by SpaceX (OpenAI, Aug. 2026)
Using OpenAI models in Cursor (OpenAI Help Center)

Model quality:

CursorBench 4.0 (Cursor)
Model Card: Grok 4.7 (SpaceXAI, Sep. 2026)
Grok 4.7 benchmarks: SpaceXAI’s numbers, independent results, and the sandbox audit (Apidog)

Musk’s own words:

Transcript: CMG interview with Tesla CEO Elon Musk (The Singju Post, Oct. 2026)
Musk turns to rival Anthropic to power Grok (City AM, Oct. 2026)
This Week in Grok: Grok Bot picks the best model (The Grid Fin, Oct. 2026)
Why Elon Musk’s recent investing advice sounds like Warren Buffett (CNBC, May 2022)

Sit On It

David Crush — Mon, 12 Oct 2026 12:00:00 -0400

Once in a blue moon I open Instagram. It is the last one I have left. I scroll for a minute, maybe two, and somewhere in there I remember exactly why I stopped. Then I close it, and the phone goes back on the table face-down.

I have been off most social media for a while now. Facebook, LinkedIn, X, and Bluesky are gone. The claim here is narrow: these platforms are designed to keep you scrolling and clicking, which makes them addictive by nature, and for something addictive by design, avoidance works better than moderation. Leaving has been one of the better decisions I have made, and I do not want much to do with them again.

They Told Us What It Was For

This is not a conspiracy theory. The people who built it said so. In 2017, Sean Parker, Facebook’s founding president, told Axios that the thought process was “How do we consume as much of your time and conscious attention as possible?” The answer was a little dopamine hit every time someone liked or commented on something you posted. “It’s a social-validation feedback loop,” he said. “You’re exploiting a vulnerability in human psychology.” And then: “The inventors, creators — it’s me, it’s Mark, it’s Kevin Systrom on Instagram, it’s all of these people — understood this consciously. And we did it anyway.”

Sit with that. The scroll is not a side effect.

What I Gave Up

Each platform had a job in my life. Facebook was family and old friends. LinkedIn was the professional costume. X was news and arguments about news. Bluesky was supposed to be the X that did not hurt, and it turned out to be the same shape with a different crowd.

That last one taught me the most. Who owns the building was never the problem. The building was: the endless feed, the reply box that is always open, the count of likes under every opinion. Move the same machine to a nicer neighborhood and it still runs the same loop.

What did I lose? Less than I expected. The people who matter to me are still reachable by text, by phone, and in person. The news still exists without a quote-tweet chain attached. What I lost was the hum: the constant low background feeling that something is happening and I am supposed to have a take on it.

Why I’m Not Going Back

My brain does not have a great off switch. When something grabs it, it holds on. That is useful when I am writing code or reading about history. It is a liability when the thing grabbing it was designed by very smart people to never let go. A book ends. A game has a final boss. A feed is engineered to have no bottom, and a brain like mine will keep looking for one.

Moderation means my willpower against a building full of engineers whose job is to beat it, every single day. Avoidance means one decision. The evidence is Instagram: every time I open it, I relearn the lesson in under two minutes.

Scroll Post Like

None of this means I think people should not be able to talk to each other online. It means I think the current machines are built for the wrong goal. For a while now, off and on, I have been working on a side project called Scroll Post Like. It is still a work in progress, not a product. The idea is a social platform that uses AI to take the toxicity out instead of feeding on it.

The core is a classifier. Every post, reply, and comment runs through an AI model that tells us some basic things about the content: is this political, is this negative, is this someone venting? Human moderation at the scale of a social network is a losing battle. There are too many posts and too few people, and the people doing it burn out. A classifier does not get tired, and it does not have to make the final call. It just labels.

Once posts are labeled, feeds can respect what users actually want. If you do not want politics in your feed, you can say so, and the platform can honor it instead of quietly deciding that outrage keeps you around longer.

The Echo Chamber Problem

That raises the obvious risk. If everyone filters out everything they dislike, you have built a machine for agreeing with yourself. This site has already argued that selective exposure and a concentrated diet do more damage than a magic algorithmic filter bubble. A preferences page is selective exposure with a nicer settings screen.

So the second job for the AI is the opposite of agreement. When you post, it can challenge your assumptions, fact-check your claims, and add the context you left out. The point is not to tell you that you are right. It is to make the position you hold a better-informed one, whatever that position turns out to be. A platform that only filters is a comfortable room. A platform that also pushes back is closer to a good friend.

The 24-Hour Wait

The feature I care about most comes from personal experience. You write a long venting post. You are angry and you are right, or you feel right, and you go to hit send. In Scroll Post Like, if the classifier tags that post as a vent or overly negative, it does not go out. It waits 24 hours. You can come back tomorrow and send it, edit it, or delete it.

I have posted angry. It is not a good idea. Sometimes we say things we will regret, and the regret lasts much longer than the anger did. Sleeping on it is old advice. The difference is building it into the platform instead of hoping people find the willpower at the worst possible moment.

There is precedent for friction like this, and it works. In 2019 Instagram started asking people “Are you sure you want to post this?” before posting a comment its AI flagged as possibly offensive. Adam Mosseri, the head of Instagram, said early tests showed it encouraged some people to undo the comment and post something less hurtful. In 2020 Twitter tested a prompt to read an article before retweeting it. People opened articles 40 percent more often, and some of them decided not to retweet at all. Twitter’s own comms account put it best: “Some Tweets are best left in drafts.”

Those were prompts you could tap past in a second. A day is longer. A day is long enough for the anger to cool and for you to read it the way everyone else would.

The vent is still there in the morning, sitting in its queue.

Better Off That Way

David Crush — Mon, 14 Sep 2026 12:30:00 -0400

If I have one major career regret, it is leaving ESPN.

I got the job straight out of college. ESPN is Disney — the same company, the same badge — and those were the first two and a half of my eleven years. I took it for granted. I assumed that was simply what working in technology was like: leadership that was great at leading, roadmaps that existed, plans that were followed, structure that held. I thought that was the floor. It took the next eight and a half years to learn it had been the ceiling.

This series has been seven parts of bills, and I want to close it with what went right — not as a consolation paragraph, but as the final piece of the argument. The same company that produced everything in Parts 1 through 7 also contained the proof that none of it was necessary. ESPN had the leadership. ESPN had the roadmaps and the plans and the structure. ESPN does technology so well that it built its own homegrown frontend framework, and people enjoy using it — which anyone in software will tell you is the harder achievement. Disney Experiences is the same company. The absurdities in Part 2 and Part 3 were never a question of ability. They already have the playbook. They just refuse to look at it.

What ESPN Gave Me

Before ESPN, I had never worked on a large distributed system. At ESPN I learned scaling, caching, service-level agreements — not from a textbook, but on systems that had to hold when the whole country checked a score at once. I was a kid out of college being handed real problems inside a structure that made it safe to grow. I learned so much there that I spent the rest of my career spending it.

The projects were the best of my eleven years. I led the frontend rewrite of the Olympics pages. I worked on the NFL Draftcast rewrite and the MLB Gamecast modernizations. If you have ever followed a draft pick or a ninth inning from your desk, we were on opposite sides of the same page. My pages were seen by millions. That is a lifetime of pride, and I will never forget it.

The Import Business

Then I moved to Disney Parks & Resorts, which later became Disney Experiences, and the remaining eight and a half years were spent trying to get our processes back to the state we had at ESPN. That sentence is not an exaggeration; it is a job description. The designs and architectures I implemented at Parks were largely lifted from, or inspired by, solutions I had seen at ESPN. I spent eight years importing a working culture from one wing of the company to another, one architecture diagram at a time.

The import was necessary because nothing was waiting on the other side. It was possible because the source existed — inside the same company, a badge swipe away.

The Good Parts of Parks

Parks had good parts too, and this is the essay where they belong.

DClean is one of my favorite projects anywhere. We used IoT to improve the experience of the custodial cast members — the people who keep the parks clean. Technology pointed at making a hard job easier for the people doing it: that is the version of the company the posters advertise, and for one project I got to build it.

During the pandemic we rewrote the mobile client, and I learned Flutter doing it. And the platform of record itself — the operations-data system from Part 6 — was genuinely fun engineering, if you take away the toxicity of the team. That sentence is the whole series in miniature. The work was good. The environment around the work was the problem. I have spent seven essays on the environment; the work deserved this one.

Back Through the Gates

Now I am a civilian in the Disney world again. It is not as much fun once you have seen how the sausage gets made. I am trying to forget and move on — to experience Disney as a guest and not as a cast member. We seem to be better off that way.

When I stand in a park now and want to know how long a line is, I check my phone like everyone else. The number comes back from the system I spent years defending. It still answers. I am just on the other side of it now.

Have the Courage

David Crush — Mon, 14 Sep 2026 10:00:00 -0400

My team was not small in what it carried. We maintained the system that powered wait times, people counts, and operating schedules across the parks — the platform of record for operations data. When a guest checks how long a line is, a system answers. Ours. It was one of the most important applications in the company.

For a long stretch, the entire permanent engineering staff of that system was me. I was the lead, and I was the only cast member developer on the team — the only Disney employee among a sea of temporary workers on short-term contracts. I was the defense.

But I could not pick my players, my coach, or the strategy.

The claim in this essay comes in two halves, and I want to be precise about which is which. I do not know that return to office was implemented to push people out. Nobody handed me a memo. What I know is that every stated justification failed on the record — the productivity theory could be read as false straight off the floor plan (Part 2), and the evidence for the mandate never existed anywhere the mandated AI could find it (Part 5). When every stated reason for a policy collapses, you are left guessing at the real one, and my best remaining guess is attrition: make the job unbearable enough and people remove themselves from payroll, no severance required. That half is inference, and I will keep calling it inference. The other half is not. Whatever the intent, the effect was an avalanche of brain drain, and leadership’s own choices accelerated it at every turn. That half I watched.

The Sea of Sixteen Months

The players I could not pick were the contractors. The company staffed the platform of record with temporary workers on contracts capped at roughly sixteen months, and it is worth spelling out what that cap means on a system with twenty years of legacy code, tech debt, and thin documentation. A new developer needs months before they stop being dangerous, more months before they are genuinely useful. By the time a contractor reached useful, the offboarding was already in motion. Their last day was on the calendar before their first day.

So my job ran on a loop: onboard, train, answer every question, watch them finally get good, wave goodbye. Repeat with the next one. It was a waste of my time, the team’s effort, and the company’s money — and I say money deliberately, because the arrangement was defended in exactly those terms. When the staffing model was questioned, the answer from leadership was: “There’s no way we do this without the people.”

The people. Any people. Part 3 covers the fallacy underneath — the belief that a developer is a developer, swappable at will. The sixteen-month cap was that fallacy converted into a staffing model. The knowledge walked out on schedule. Every schedule.

The Lead They Fired and the Lead They Bought

The coach I could not pick was the lead they hired. But first, the one they already had.

The team had a lead in Washington state, and she was excellent — she knew the system, and the team trusted her. She was laid off. Not for performance. For geography: she would not relocate to Orlando, so she was gone. Read that against the rest of this series. The office mandate that no evidence supported, enforced into the office that was not ready, cost the platform of record a proven lead. If the guess I am making in this essay needs a single data point, it is this one. When the mandate and the talent conflicted, the mandate won — on the record, with a layoff.

They created the leadership vacuum themselves. Then they filled it. Three of us interviewed the replacement candidate, and all three of us came back iffy — not a no, but nowhere near a yes. It did not matter. Leadership had extended the offer before anyone else was even interviewed. The search was over before it started.

It did not work out, and everyone could see it was not working out, and nothing was done. We watched him drive out team members who carried legacy knowledge of a twenty-year system, the kind of knowledge that does not survive a handoff document. Leadership watched too. When he was finally out, they turned back to me — the demotion in Part 3 had not moved the responsibility anywhere, only the title. I stepped back into the gap because the system mattered. But the damage was done. The people who left did not come back.

The Coward’s Layoff

The strategy I could not pick was the one nobody would say out loud.

If the goal was fewer people, there was an honest tool for exactly that. It is called a layoff. It comes with severance, notice, a clean story, and a leader’s signature on the decision. It costs the company money and it costs the leader something too — a hard conversation, a number on a report with their name attached. That cost is what makes it honest. The person deciding pays part of the price.

Engineering misery until people quit on their own is the same headcount reduction with the courage removed. No severance. No signature. No story to own. The company books it as voluntary attrition and every hand upstairs stays clean. And it is worse business, not just worse ethics, because misery does not select at random. The people who leave first are the ones with options, and the ones with options are your best. The research on return-to-office mandates found exactly this: turnover spikes afterward, concentrated among the most senior and most skilled, and the vacancies they leave take far longer to fill. A layoff aims. Attrition by misery fires backward into your own talent.

That was the hard load to bear. Not the commute, not the noise, not the office. Knowing that our leaders were sacrificing us to make themselves more comfortable. It should be the other way around. Absorbing discomfort so your people do not have to is not a perk of leadership. It is the job description.

Just do the layoffs. Have the courage.

A Slowly Contracting Org

I watched the avalanche from the inside. The best people went first, exactly as the incentives predicted, and the legacy knowledge went with them — the unwritten map of a twenty-year system, carried out one resignation at a time. The team that remained was me, for a long while, plus whoever was passing through on a sixteen-month clock. Then I left too. What remains is a slowly contracting org, defending one of the most important applications in the company with less every year.

The last time I onboarded a new contractor, I opened the team calendar to schedule the first walkthrough — wait times, people counts, schedules, the platform of record. The end date was already on it. Sixteen months out, marked before the first walkthrough, before the first commit, before I had learned to spell their name.

Incorporate It Into Everything

David Crush — Sun, 13 Sep 2026 01:45:00 -0400

Our leaders loved AI. We were ordered to incorporate it into everything — Part 2 covers what that mandate did to a team simultaneously ordered back into hallway conversations no AI could hear. But the mandate had a second life inside our team, and it sounded like this: “We have a leadership meeting today where we need to show how we’re using AI. How can we shoehorn it into the process?”

Shoehorn. Nobody was asking how AI could improve the process. They were inserting it wherever it would be visible from upstairs and leaving the details for later. That is not leadership. That is throwing things at a wall to see what sticks.

Fine. I did what they asked. I incorporated it into everything. Including them.

The claim here is narrow, and it is not “the AI proves I was right.” It is this: leadership demanded that everyone trust this tool, and the tool, asked plain questions in good faith, returned answers that contradicted their entire operating strategy. The answers were sitting there the whole time, ten seconds away, inside software the company was already paying for. What follows are the questions and what came back.

Question 1: The Evidence for RTO

I asked for studies showing positive results from return-to-office mandates. The response, trimmed only for length:

The search results show predominantly negative findings… the empirical literature on RTO mandates shows remarkably limited evidence of positive outcomes, and most peer-reviewed studies document neutral or negative effects.

It kept going. Stanford’s research on hybrid arrangements — Nicholas Bloom’s work — found roughly a one-third reduction in turnover among hybrid workers compared to fully in-office teams. And where research does find benefits to being in person, the AI flagged the caveat itself: the University of Pittsburgh finding that face-to-face teams produce more breakthrough innovation compares teams that choose to work together, not teams under mandate. Its summary deserves quoting whole:

The scholarly consensus suggests that companies claiming positive RTO outcomes are typically not citing rigorous evidence. Where genuine benefits appear in research — innovation from in-person work, relationship-building, onboarding — they emerge in discretionary, flexible contexts, not under mandate.

Part 2 describes the theory our return was sold on. That is what the mandated tool says about the theory. I am keeping the caveats in because the caveats are the point: the tool was more careful with the evidence than the people who ordered us to trust it.

Question 2: The Commute Bill

I did not need the AI for this one. I needed a calculator.

We were required in the office four days a week, with one optional work-from-home day. My shortest commute in eleven years — Team Disney — was about twenty minutes each way. Forty minutes a day, four days a week, is 160 minutes: nearly three hours of my week, unpaid. At my rate of $70 an hour, that is roughly $200 a week. Call it ten thousand dollars a year. And that was the best case — the KP buildings could take an hour each way in rush-hour traffic, which multiplies the number. This counts time only. Gas, miles, and parking are extra; AAA prices the full cost of operating a car every year, and it is not small.

Return to office was a five-figure pay cut that never appeared on any pay stub. Leadership knew commutes cost something — everyone who has ever driven to work knows. They chose not to count it, and we were expected to absorb it without asking questions about the purpose, the effectiveness, or the readiness of the office it paid for.

Here is the killer: the ten thousand dollars bought nothing. If the office had at least made the job easier, you could squint and call it a trade. It made the job harder — Part 2 has the noise, the interruptions, the meetings that stayed on screens. So the bill was paid twice: once in commute time, and again in the productivity gap the office created, which the developers were expected to close on their own hours. We paid for the privilege of being made worse at our jobs, then donated the difference.

Question 3: The Checklist

I asked for the signs of incompetent technology leadership. It returned a long checklist, organized into categories. I will not quote all of it. Here are the items I did not have to imagine, because this series has already documented them:

Consistently sets unrealistic timelines without consulting engineering teams.

The rules engine dates, set by a promise instead of the work (Part 3).

Dismisses or argues with experienced engineers’ concerns rather than investigating them seriously.

Every one-on-one where I raised the requirements problem.

Blames external factors, tight timelines from above, or team incompetence when things fail — rarely takes ownership.

The money speech, delivered by the man who signed the bills.

Doesn’t create psychological safety for raising problems or disagreeing respectfully.

The happy hour. The write-up (Part 1).

High turnover among senior engineers and technical managers… team morale is visibly low; engineers seem cautious or disengaged in meetings.

The panic attacks. My own exit, after eleven years, without a thank-you.

The checklist ended with a scoring rule, and I will quote it exactly:

A few checkmarks might be normal variance; a pattern across multiple categories usually signals a real problem.

Question 4: The Ten-Second Answer

The last question was the one that mattered: how do you support a neurodivergent employee?

The answer came back organized and calm, as if it were the easiest question of the four. Be explicit — “written communication helps those who process better in text.” Predictability aids focus — “consistent meeting times, advance agendas, and warning about changes reduce cognitive load.” Minimize sensory overload — “offer flexibility on open offices, lighting, noise, or remote work options.” Frequent interruptions are especially draining, so protect deep work time. And formalize accommodations in writing, through HR, “so they’re consistent and not contingent on manager mood.”

Written communication. Advance agendas. Flexibility on the open office and remote work. Protection from interruptions. Read Part 1 and count the matches — that list is, nearly item for item, what I asked for and was denied. My director told me I was great at verbal communication. The write-up meeting came with no agenda and the wrong date. The open-office answer was a storage closet with no roof.

The answer was free. It took ten seconds. The company was already paying for the tool, and the tool was already in the building, being shoehorned into slide decks for the meeting upstairs.

The prompt box is still there, in the software they mandated. The cursor is blinking in the empty field. The question was never typed.

Sounds Like We Need a Happy Hour

David Crush — Sat, 12 Sep 2026 11:00:00 -0400

I would bring up an issue with the project. Something real — requirements that did not exist yet, a date nobody had checked, work that could not start until someone above me made a decision. And my manager would say: “Sounds like we need a happy hour.”

That was the answer. Not a plan. Not a decision. Not even a disagreement. Drinks.

This is Part 3 of a series about my eleven years as a Disney cast member. Part 1, The Storage Closet, is the accommodations bill. Part 2, No Conference Rooms, is the absurdity bill. This one is about accountability, and the claim is not that I worked for one bad manager. The claim is that the organization was built so accountability could only flow in one direction. Leadership hoarded the decision-making power and distributed the responsibility. Down.

The Happy Hour Method

Understand what the phrase actually did. A problem raised in a meeting demands something — an owner, a decision, a change. A problem carried to the bar demands nothing. The issue was never argued with. It was dissolved. The unspoken instruction was to drink it off and not bring it up again.

It trained us. Raising a problem got you a social event; it never got you a solution. So the team learned what any team learns under that arrangement: live with it. Avoidance was not my manager’s personal quirk. It was the culture he was building, and culture flows downhill faster than any mandate.

I will be honest about what that cost, because it belongs on the record. The only sanctioned way to cope was the bar, and the method worked on me. I found myself drinking too much just to get through the weeks. I was not alone — others on the team came to lean on alcohol the same way, for the same reason. When the official solution to every problem is a drink, people learn the lesson.

The Lead Who Carried the Blame

For part of my time there, I was a lead software engineer. On paper, that is a position of authority. In practice, I had none. I could not slow the project down and I could not change course — my manager told me directly that we could not do either. What the title came with instead was the blame. When something did not make it to production on time, whatever the actual cause, it landed on me.

The rules engine was the clearest case. I told them, many times, that we needed to slow down because the business team did not know what they needed yet. That was not a knock on our business partners — as I wrote in Part 2, not yet knowing what you do not know is how real projects begin. But leadership had already told people, without checking with anyone who would build it, that the rules engine would be in production soon. So the dates were set by the promise instead of the work, and we were forced to chase them. This was the same manager who needed to see movement on the rules engine. I was never listened to. I was talked down to.

So I stepped down. Voluntarily, from lead to senior, with a sizeable pay cut. The title carried no power and all of the blame, and I paid my own money to put it down.

Here is what the organization did next: nothing. They did not backfill the lead position. They did not cut any scope from the project. The team was expected to pick up the slack, and I was expected to keep doing everything I had been doing. They were happy to give me a paycheck and keep demanding the same of me.

Paid in Full Regardless

Below the manager, the same vacuum. We had constant problems with vendors not doing their jobs. Nothing happened to them — no correction, no escalation, no consequence. They simply remained until their contracts expired. The expiration date did the job that management would not.

We hired a business analyst who was supposed to own requirements — the one thing this team needed more than anything, as the rules engine kept proving. Nobody defined the role, nobody coached it, nobody corrected it, and it shrank into glorified note-taking that software could have done better and cheaper. Her camera would flick on by accident in meetings and she would be sitting in the front seat of her car. Regularly. Nothing ever happened, and she made her money through the contract regardless. That is less a story about her than about the vacuum around her. In an organization with working feedback loops, someone notices, someone coaches, someone corrects. Nobody did any of those things. Nobody was watching anybody.

The Money Speech

Then the same manager complained to the team that we were running out of money.

Consider what the money had gone to. Contracts that coasted to their expiration dates. A requirements role that produced notes. And hires we did not need — because this manager insisted on adding team members, on the theory that a developer is a developer, interchangeable units you can swap in and out at any point. That is not how it works anywhere, and it is especially not how it works on a twenty-year-old legacy system with deep tech debt and very little documentation. The knowledge lived in people’s heads. A new developer was a long, expensive ramp, not a plug-in part.

It was the same belief that made my empty lead position feel free. If developers are interchangeable, you never need to replace one — the rest can simply absorb the work. One belief, two costs: paying for people we did not need while declining to replace the one we did.

And the bill for all of it was presented to us, as a grievance, by the man who had signed it. We were running out of money. We needed to understand that.

The Pattern

None of this was new information by then. I had been written up for an involuntary autistic meltdown while nobody answered for the accommodation request closed out during my medical leave. Specs changed retroactively and the team absorbed it without a word. Deadlines bore down and leadership went home early. We were promised first dibs on seating at KP2 and were moved again within months.

The last time I raised a problem in that job, I already knew what the answer would be.

The drinks arrived. The problem did not come up again.

No Conference Rooms

David Crush — Sat, 12 Sep 2026 09:10:00 -0400

Team Disney. Feature Animation. KP1. Feature Animation. KP2. KP1. Team Disney.

Those are the buildings I sat in across eleven years on Disney’s technology side. It averages out to a move every single year. Every move cost money — movers, redesigned layouts, new furniture — and not one of them happened for us.

Part 1 of this series, The Storage Closet, is the personal bill: burnout, accommodations, and what I was offered when I asked for help. This one is about the leadership decisions around all of it. The claim is not that Disney’s leaders were stupid. The claim is that they optimized for optics over results, consistently, for years — and you could read it straight off the floor plan.

The Productivity Theory

Remote work was not a pandemic invention for us. My team worked from home two or three days a week for years before 2020, and nobody asked or said anything. During the pandemic, our leaders bragged about it: we could work from anywhere, all we needed was our laptops. Within a couple of years, the same leaders had decided that arrangement was the problem, and that the office would somehow make us more productive. We were expected to forget the old philosophy because forgetting had become convenient.

The justification came from the very top. Our CTO told us: “We are in the hospitality business. You should be here.”

Nobody on my team was checking in a guest. We were a technology team, and technology teams tend to be distributed — ours more than most, because of how Disney hires. Disney is not a technology company. It does not pay what the big technology companies pay, and it does not carry their prestige with engineers, so it has to offer something else to attract talent. Instead, leadership made hiring more restrictive: they wanted people in Orlando, one of the harder markets in the country to hire technical talent in. The predictable result was that we leaned heavily on third-party vendors. And here is the kicker — very few of them were local. We worked extensively with vendors who supplied remote contractors, and nine times out of ten, those contractors were the ones doing the actual implementation.

So return to office meant this: a team whose real workforce was remote by design commuting into a building with no assigned seats and no space for vendors, to meet on screens with the people doing the work somewhere else. Our leaders never took fifteen minutes to watch how we actually worked. Fifteen minutes on the floor would have shown them everything — the headphones, the meetings happening on Teams, the empty chairs where a vendor was never going to sit. The theory was dead on arrival, and nobody in charge ever performed the autopsy.

Two Mandates That Cancel Out

At the same time leadership was mandating desks, it was mandating AI. Incorporate it into everything — that was the direction.

Those two mandates cancel each other out. A Teams meeting documents itself: the AI takes notes in real time, the recording exists, the decisions are searchable, and the agents we were told to build can actually consume all of it. A side conversation at a desk — the thing our leaders said they wanted more of — is invisible to every bit of that. Whatever gets decided in a hallway has to be carried back into the documentation and the tools by hand, a jump added to the middle of every decision. They mandated the one channel their own AI strategy could not hear. Entirely self-inflicted.

I was a senior software engineer and I could see this from my desk on the floor. The people whose offices overlooked that floor never saw it. Overlooked turned out to be the right word in both directions.

When Everything Is Urgent

We never had a proper roadmap. Not a rough one, not a draft — none. There was no prioritization, so every problem arrived urgent, needed ASAP. When every problem is urgent, nothing is. We spent enormous amounts of time spinning our wheels to show progress on paper instead of making real progress, because progress on paper was what got inspected. Optics over results was not a failure mode of the operation. It was the operation.

And when the gap between the paper and the reality came due, the developers were expected to close it. Deadlines would bear down and leadership would go home early, same as always. They expected us to stay.

The Rules Engine

My last project there was the whole method in one story. We were expected to deliver code before the requirements were finalized. I cannot count the times I delivered exactly to spec, watched the spec change retroactively, and was expected — along with the rest of the team — to absorb it and say nothing. That was not an incident. That was the process.

I raised it in every single one-on-one I had. Nothing changed. What I got instead, repeatedly, was: “I need to see movement on the rules engine.”

There was movement. It was happening in the design and the architecture, which is where movement belongs when nobody knows yet exactly what the thing is supposed to do. Our business partners needed time. They did not know what they did not know — and that is not a criticism, that is how real projects begin. But my boss wanted the rules engine live on the site while we were still working out what it would even do. “Movement” did not mean something true in the architecture. It meant something visible on a screen. The team was expected to just figure it out.

The Building Shuffle

Back to that list of buildings. Seven addresses, eleven years. Each move meant relocating everything we had, and each new floor got redesigned — executives never had trouble finding extra budget for their own offices. Move by move, more of our space became their space. By the end, we did not have conference rooms. The team that had been ordered back to the office for the sake of collaboration had nowhere to hold a meeting.

The KP2 move happened as a favor to our Director. As part of the deal, we were told we would have first dibs on where to sit in the new space. That lasted a few months before we were moved again. It was neither the first nor the last promise to evaporate that way, and it is the pattern in miniature: the move served someone above us, a promise bought our compliance, and the promise did not survive the next reshuffle.

And the geography. KP1 and KP2 sit across the street from Universal Studios. Disney cast members, summoned back to the office to be together, badging in every morning with a view of the competition’s parks. Very magical.

By my last year, if you wanted a meeting, you booked a screen. The conference rooms were offices now. The doors were closed.

Pre-Ordering Permission

David Crush — Sun, 06 Sep 2026 23:50:00 -0400

The box arrives November 12. Grand Theft Auto VI — Rockstar’s Vice City game, thirteen years after the last numbered one — launches a week later on PlayStation 5 and Xbox Series X|S. Standard is $79.99. Ultimate is $99.99. The physical version is a download code. Rockstar’s own support page says a disc will not be included.

Pre-order existed to allocate scarce objects. Plastic. A print run. A shelf that could miss you. A license is not an object. The storefront can mint another one whenever it wants, and on Sony’s own court theory it can take that one back. Paying eighty or a hundred dollars months before anyone has played the game is not meeting demand. It is locking the sale before the reviews exist.

What They Already Told a Court

In They Can’t Have It Both Ways, Sony told a federal court that reasonable consumers already knew a digital game was a license, not ownership. PlayStation Terms §8.4: you buy a personal license and “do not own the product.” §10.2: the license is “revocable.”

Sit with that. I am not going to replay the deleted libraries here. The useful fact is the one they already put on paper.

What the Line Was For

Midnight launches were about inventory. A truck. A finite stack of cases. You put money down so the store ordered enough, or so you were first when the stack ran out. That world could actually miss you. A burned disc takes a factory. A license takes a row in a database.

If the licenses are scarce, that is their fault. Server capacity is a budget. Bandwidth is a budget. The November 12 pre-load is them asking you to solve their pipe with your disk and your money, five months after they opened the register.

Digital sold-out is a choice.

The Off Switch

Rockstar does not have to pull Grand Theft Auto VI on November 20 for the clause to matter. Sony already deleted purchased libraries. The chance is not a rumor. It is a clause.

Eighty dollars. A hundred if you want Ultimate — extra cars, guns, clothes, and story beats threaded through Jason and Lucia, the leads. Both prices buy the same legal object. Permission. With a contractual kill switch.

The Pack

They could offer a trial. If they can mint access at will and revoke it at will, play-then-pay is the honest product. They do not sell that. They sell the Vintage Vice City Pack — items that “flash back to when the neon burned brightest” — and a month of GTA+ if you pay before November 20. That is not demand. That is bait for a sale that has to close before anyone can say whether the game is good.

Wait for the reviews. The license will still be there. If it is not, they have answered the question for you.

The box will still show up November 12 for the people who could not wait. Open it. There is a slip of paper.

Sources

They Can’t Have It Both Ways
Rockstar Games Announces Pre-Orders for Grand Theft Auto VI (Take-Two, June 2026)
Grand Theft Auto VI Platforms, Editions, and Versions (Rockstar Games Support)
GTA 6 Physical Copies Won’t Include a Disc, Will Just Be a Code in a Box (IGN)
Motion to compel arbitration or dismiss, Garcia et al. v. Sony Interactive Entertainment (N.D. Cal., No. 3:26-cv-06016, Aug. 21, 2026)
PlayStation Terms of Service

Filling in the Blanks: What Anxiety Taught Me About AI

David Crush — Thu, 03 Sep 2026 14:15:00 -0400

AI has become one of my special interests. I mean that in the autistic sense, not the “I read a newsletter” sense. I love it. I work on it — I build Ethyx, which is a disclosure, not a pitch. I think about it more than is socially useful.

I also have real concerns. Not the movie. The crutch. The black box of magic. People using a confident sentence as if it were a conclusion. That is how we lose the plot. That is how a tool that could enhance us starts destroying us — not because it wakes up, but because we stop doing the parts of thinking that keep us human.

I understand AI better than a lot of people who are louder about it. Not because I am smarter. Because it fails in a way my brain already fails.

The Gap That Cannot Stay a Gap

I have had anxiety my whole life. Part of anxiety is inventing scenarios and spiraling on them. We really struggle when we do not have a way to explain something. So we invent a way. The blank is intolerable. The story is a relief, until it is not.

That is not “I worry a lot.” It is a specific move: unexplained things get filled in, and then the fill-in becomes the thing we live inside.

Completion, Not Lying

AI is doing a version of that move. It is only trying to come up with an answer. It does not sit with “I don’t know” the way a careful person can. When the data is thin, it still completes the sentence. Hallucinations happen in the absence of data. That is not a lie. It is a completion.

The direction of the fill-in is the difference worth protecting.

My spiral goes toward catastrophe. More things that can go wrong. The model’s fill-in goes toward the average — the most typical next thing, the mean of what it has seen. Same intolerance of a blank. Different destination. That is why AI writing can feel competent and dead. Averaged until the warmth is gone.

Scripts From a Personal Corpus

Autism and ADHD work the same way in my social life. I do not get the cues a lot of people seem to get for free. So my mind fills the blank from past experience — last time, what usually works, the script that has kept me out of trouble.

Masking is me generating the likely “correct” social move from that corpus. When the corpus is wrong for this room, I guess. Sometimes the guess is technically right and personally hollow. That is the same flavor as AI slop: a plausible sentence with nobody home.

I am not saying AI is autistic. I am not saying a model feels anxiety. I am saying I recognize the failure mode because I live a version of it.

Vague In, Average Out

Understanding that is how you make the tool better, and how you use it without getting used by it.

If you leave out data, it fills with the most average answer possible. Do that over and over and you get output that feels impersonal, because it has been averaged until any personalization is gone. The AI 101 series I wrote for Ethyx puts it as a habit: vague in, average out. Every gap you leave gets filled with the most typical guess. Hand over the real thing — your voice, your constraints, your facts — or you will get the mean of the internet wearing a confident tone.

Good Uses Are Not the Default

There are good uses. I am not performing balance when I say that. Thinking with a tool. Drafting when you already know what you mean. Exploring a question you will actually check. Accessibility. Catching what a tired or overloaded brain misses. Those are real.

They are not automatic.

The difference is usually whether you still have work to do after it answers. Good: you already know what you mean, and you use the model to get a draft on the page, then you rewrite it until it sounds like you. Good: you are stuck at 2am and you need a way to think out loud that will not judge you, and then you check the parts that matter. Bad: you ask it for the answer in a domain you will not verify, then paste the confidence. Bad: you let it write the email, the essay, the decision — and you stop being able to tell where you ended and the average began.

Markets and hype will not pick the good uses for us. The default is the crutch and the black box, because those are easier. If we do not use the good cases on purpose and limit the bad ones, we lose the plot. AI can destroy us that way. Not Skynet. Us, handing judgment to a box we refuse to understand.

Think with it. Do not let it think for you. It is a good rubber duck and a bad priest. Put it down when the blank in front of you is a judgment, a fact you cannot check, or a person you should be talking to instead.

The Bill

Liking the tool does not cancel the costs. I work on this stuff. Skipping this section would be the opposite of the honesty I ask of this site.

Environment. Training and running these systems burns electricity and water at data-center scale. The chat window is free. The grid is not. The IEA estimated data centres used about 415 terawatt-hours in 2024 — around 1.5 percent of global electricity — and projected that demand more than doubling by 2030, with AI as the main driver. Cooling still drinks from local water systems. I am not going to pretend my next average completion is carbon-neutral. Waste completions are not only bad writing. They are waste.

Misuse. The bad uses are already here. Treating a guess as proof. Fraud. Surveillance that inverts the presumption of innocence — I wrote about that with Flock cameras. You do not need a future tense for this paragraph.

Unemployment. Displacement is real and uneven. Some tasks go away. New ones appear. The people who lose are not the people who profit. “Everyone will just retrain” is not a plan. Neither is the sermon that work becomes a hobby while scarcity ends. Fair is not apocalyptic. Fair is admitting the transition is a political problem, not a vibe.

Privacy. What you type is not a diary. Logging, training, third-party access — that is the quiet half of the black box. The habits for not handing over your life are in Part 5 of AI 101. The stakes belong here: a tool that fills blanks will fill them with whatever you put in the box. Do not put in what you cannot afford to have completed later.

Literacy Is the Condition

I do not need AI to be magic. I need it to be understood.

AI 101 is the practical companion to this essay: what you are actually talking to, what to use it for, how to ask so you do not get the average, when it makes things up, and what happens to what you type.

The fork is not “use it” versus “ban it.” The fork is literacy versus the crutch. If we know how it works, where it fails, and when to put it down, it can enhance us. If we do not, we will keep filling in the blanks — and calling the fill-in a future.

They Can't Have It Both Ways: Sony, Disney+, Stadiums, and the License You Never Bought

David Crush — Wed, 02 Sep 2026 15:40:00 -0400

Give a corporation an inch and it will take a mile. That is not a metaphor I invented for this post. It is the operating manual.

The inch is a button that says Buy Now. A checkbox under a free streaming trial. A ribbon-cutting for a “public-private partnership” that will supposedly revitalize a city. The mile is a library that evaporates without a refund. An arbitration clause stretched over a death. A billion-dollar stadium paid for by taxpayers while franchise value and operating profits stay private.

Sony is doing it with digital media. Disney tried it with Disney+. Sports owners have been doing it with arenas for decades. And every time we treat the mile as normal — every time we shrug and say that is just how the industry works — we are teaching them the next mile is already approved.

I am in a mood about this. I think that is the correct mood.

The Library You Paid For

On September 1, 2026, PlayStation told customers in the United Kingdom that StudioCanal movies and TV titles would be removed from their video libraries. The company’s own legal notice is blunt:

From September 1, 2026, due to our content licensing agreements, you will no longer be able to access your previously purchased content from Studio Canal, and it will be removed from your video library.

Previously purchased. Removed. No refunds announced. The list includes titles people actually care about — Terminator 2, Hot Fuzz, Moonlight, Paddington, Apocalypse Now. You paid money. The storefront called it a purchase. Then the files went away.

This is not a one-off. In December 2023, Sony announced it would delete purchased Discovery shows — MythBusters, Deadliest Catch, and more than a thousand other seasons — citing “content licensing arrangements with content providers.” After public backlash, Sony worked out an updated deal with Warner Bros. Discovery and kept the content available “for at least the next 30 months.” That reversal matters. It proves deletion is a business choice, not an act of God. When enough people make noise, Sony can fight for a renewal. When they do not, your library is a cost to shed.

“Reasonable Consumers” Already Knew

The part that should make your blood pressure climb is not only the deletion. It is what Sony told a federal court about what you supposedly understood when you clicked Buy.

In Garcia et al. v. Sony Interactive Entertainment — a proposed class action in the Northern District of California (No. 3:26-cv-06016) — four California PlayStation customers sued under a 2025 California law that requires clear disclosure when a digital “buy” is really a license. On August 21, 2026, Sony filed a motion to compel arbitration or dismiss. In that filing, Sony’s lawyers argued that “reasonable consumers would not be misled into believing that obtaining licenses to PlayStation video games conveyed ‘ownership.’”

They went further. They said it is “not plausible to allege that reasonable consumers believed they were obtaining ‘ownership’ of a digital game,” because two plaintiffs bought Resident Evil Requiem eleven days apart — and if the first buyer owned it, the second supposedly could not have bought it.

That is a lawyer’s trick dressed up as common sense. It conflates copyright in the work with ownership of a copy. Nobody who buys a paperback of The Catcher in the Rye thinks they are the only person on Earth allowed to own that novel. Two people can own two copies of the same book. Two people can own two licenses, or two discs, or two downloads, of the same game. Sony is counting on a judge treating that confusion as what “reasonable” people already knew.

I do not think that is what reasonable people know. I think that is what Sony needs a judge to pretend they know.

How They Actually Sell It

Open the PlayStation Store. The button does not say “Acquire a Non-Exclusive Revocable Personal License.” It says Buy Now. Confirm Purchase. The storefront is built to feel like a sale.

The ownership language lives hundreds of words into linked documents almost nobody opens before paying:

PlayStation Terms §8.4: when you order or purchase a product, “you buy a personal license” and “do not own the product.”
§10.1: use of words like “buy,” “purchase,” “own,” or “sale” “does not mean or imply any transfer of ownership.”
The Software Product License Agreement: “The Software is licensed to you, not sold.”
§10.2: the license is “revocable.”

So the marketing speaks purchase. The fine print speaks permission. And when someone sues over the gap, Sony’s first move is often to shove the dispute into individual arbitration under the same Terms — the same inch that became the mile.

California already answered Sony’s “reasonable person” claim. Assembly Bill 2426, effective January 1, 2025, added Business and Professions Code §17500.6. It makes it unlawful to advertise digital goods with “buy,” “purchase,” or “any other term which a reasonable person would understand to confer an unrestricted ownership interest,” unless the seller gets a separate affirmative acknowledgment or provides a clear, conspicuous, plain-language statement — distinct from other terms — that buying is a license.

Read that carefully. The California legislature’s premise is the opposite of Sony’s court brief. A reasonable person does hear “buy” as ownership. That is why the statute exists. Sony’s argument is not describing how people shop. It is describing how Sony wants the law to forget how people shop.

If the storefront were labeled “Revocable Licenses to Games” instead of Games, I do not know many people who would line up at full price. Sony knows that. That is why the inch is a Buy button and the mile is a courtroom filing.

If We Don’t Own It, How Is Piracy Stealing?

Here is the cake-and-eat-it-too move that should end the conversation before it starts.

For decades, the industry ran campaigns that equated unauthorized downloading with theft. “You wouldn’t steal a car.” The moral force of that pitch depends on the purchase frame: you took something that belonged to someone else.

But if what you “bought” was only a revocable license — if Sony can tell a court that reasonable people never owned the thing — then unauthorized copying is not stealing a car. It is copyright infringement. Those are not the same crime, and the Supreme Court already drew the line. In Dowling v. United States (1985), the Court held that copyright infringement is not theft or conversion of goods under the National Stolen Property Act. Infringement can still be illegal. It is still wrong to distribute someone else’s work without permission. But it is not “stealing” in the sense the billboards sold you.

You cannot sell Buy Now, revoke the library, tell a judge nobody owned anything, and then moralize that a download is stealing the thing you just denied was owned. That is not a legal loophole debate. That is a company wanting the emotional language of ownership when it sells, and the legal language of licensing when it takes away.

I am not telling anyone to pirate. I am saying the hypocrisy is the point. If the product is a license, say so at the button. If you want the moral panic of theft, sell ownership. They cannot have both.

Be Fair — Then Don’t Soften It

Sony can lose a StudioCanal license. That is real. Licensors expire. Negotiations fail. I am not arguing that every studio deal must last forever by magic.

What I am arguing is that none of that requires (a) marketing the transaction as a purchase, (b) offering no refund when access disappears, or (c) telling a federal court that every reasonable buyer already understood the whole arrangement was temporary permission. The Discovery fight in 2023 showed Sony can renegotiate when backlash is expensive enough. StudioCanal in 2026 shows what happens when they decide the cheaper path is deletion.

The law may allow a lot of this. That does not make it decent. It is okay to call it anti-consumer. It is okay to be mad at Sony. They earned it.

Disney Tried the Same Inch

If you think this is just a games-store problem, look at Disney.

In October 2023, Dr. Kanokporn Tangsuan died of anaphylaxis after a meal at Raglan Road Irish Pub in Disney Springs. The restaurant is a tenant, not a Disney-operated kitchen. Her husband, Jeffrey Piccolo, sued Disney and the restaurant under Florida’s Wrongful Death Act.

In May 2024, Disney moved to compel individual arbitration. One of its arguments: Piccolo had agreed to arbitrate “all disputes” against “The Walt Disney Company or its affiliates” when he signed up for a one-month Disney+ trial in 2019. His lawyers said he signed up on a PlayStation and believed he canceled during the trial. Disney also pointed to park-ticket terms. Piccolo’s lawyers called the Disney+ theory “preposterous” — and they were right. A streaming trial checkbox cannot reasonably be read as waiving a wrongful-death claim arising from a restaurant meal years later.

After the country saw it, Disney walked it back. In August 2024, parks chair Josh D’Amaro said that given the sensitivity of the case, “we’ve decided to waive our right to arbitration and have the matter proceed in court.” Humanity above all other considerations, the statement said.

Hear what that sequence actually means. Disney did take the legal position. They put it in a motion. They asked a court to treat a 2019 streaming trial as a shield against a courtroom over a 2023 death. They dropped it when shame got more expensive than the clause.

That is not exoneration. That is the same incentive map as Sony’s Discovery reversal. The fine print is the real policy until it becomes expensive to defend in public. Sony’s games lawsuit is also trying to push people into arbitration under buried Terms. This is not a one-off. This is the industry standard.

Give them a checkbox. They will take a mile.

Stadiums: Civic Pride as the Inch

Now zoom out from the screen to the skyline.

Billionaire sports owners love to talk about how valuable their franchises are. Forbes rankings. Media rights. Naming rights. The team as a private jewel. Then it is time for a new arena, and suddenly nobody has any money. The public is asked to cover construction — bonds, diverted sales taxes, “economic development” packages — while the profits that follow remain private. When the upside arrives, it is their money. When the building is due, it is your problem.

The research is not ambiguous. Sports economists have documented roughly $33 billion in public funds for major-league stadiums and arenas in the United States and Canada between 1970 and 2020, with the median public contribution covering about 73 percent of construction costs. Economists who study this for a living keep finding the same result: professional sports venues produce little to no net local economic gain relative to the public money poured into them. The benefits concentrate among owners. The costs spread across taxpayers who may never buy a ticket.

The ceiling keeps rising. In late 2025, Kansas officials committed up to $1.8 billion toward a new stadium for the Kansas City Chiefs — reported as the largest professional-sports subsidy on record — while the Hunt family remains among the wealthiest ownership groups in sports. The franchise is a private asset when it is valued. The stadium is a public obligation when it needs a roof.

Same pattern. Different industry.

Civic pride is the inch. The mile is a publicly financed building that socializes downside risk and privatizes the upside. Heads they win. Tails you paid.

What We Are Trained to Accept

Sit with how many of these arrangements you have been trained to treat as normal.

You click I Agree because the alternative is not using the product. You hear Buy and assume you bought something. You hear “the city needs a stadium” and assume the alternative is the team leaving. The law often ends up lop-sided because the mile was allowed to become precedent — because enough people treated the stretch as inevitable instead of outrageous.

I want you to be a little self-reflective here. Not in a scolding way. In a clearing-your-eyes way. Where else have you watched a corporation take an inch — a button, a checkbox, a slogan — and stretch it into something you would never have agreed to if it had been labeled honestly on day one?

Anger is a proportionate response. It is not uncivil to say Sony and Disney deserved the heat they got. Treating corporate self-dealing as ambient weather is how the next mile gets drafted into the Terms of Service before anyone notices.

Accountability Is the Check

I do not need a boycott sermon to close this. I need the mechanism named out loud.

Sony reversed Discovery when backlash was expensive. Disney waived arbitration when the country noticed. California passed AB 2426 because legislators decided “Buy” was misleading without a plain-language license disclosure. Those are not stories about corporations discovering ethics. They are stories about corporations discovering a cost.

Give them an inch and they take a mile. Silence reads as consent. The work — the unglamorous, necessary work — is refusing to treat the mile as normal. Make the stretch expensive. Call it what it is. Hold them accountable.

They cannot have it both ways: the language of ownership when they sell, the language of licensing when they take it back; the language of magic when they want a streaming trial signed, the language of “all disputes” when someone dies; the language of franchise value when the books look good, the language of public partnership when the concrete needs pouring.

That is the business model. Stop pretending it is a misunderstanding.

Sources

Sony / StudioCanal and Discovery:

Studio Canal Removal (UK) (PlayStation official notice)
PlayStation Will Delete Purchased Discovery Shows (New York Times, Dec. 2023)
PlayStation will not delete Discovery TV shows after all (GamesIndustry.biz)
Sony’s attempted removal of “purchased” content (Consumer Rights Wiki)

Sony lawsuit and “reasonable consumers”:

Class action complaint, Garcia et al. v. Sony Interactive Entertainment (N.D. Cal., No. 3:26-cv-06016)
Motion to compel arbitration or dismiss (Aug. 21, 2026)
Facing lawsuit, Sony argues it’s already obvious that digital games aren’t owned by consumers (Game File / Stephen Totilo, Aug. 2026)
PlayStation Terms of Service

California AB 2426:

AB-2426 Consumer protection: false advertising: digital goods (California Legislature)
California Expands False Advertising Law to Require Express Licensing Disclosures (Thompson Coburn)

Piracy / theft distinction:

Dowling v. United States, 473 U.S. 207 (1985)

Disney+ arbitration:

Disney says man can’t sue over wife’s death because he agreed to Disney+ terms (NBC News)
Disney backpedals on Disney+ arbitration argument (NBC News, Aug. 2024)
Disney won’t use Disney+ terms to dodge wrongful death suit (NPR)

Stadium subsidies:

Sports Stadium Subsidies: Taxpayers Shoulder a Heavy Burden (Tax Foundation)
Sports stadium deals hand even more taxpayer money to billionaires (Stateline, Jan. 2026)
With Chiefs stadium, Kansas offers even more taxpayer money to billionaires (KCUR, Jan. 2026)
Sports stadiums and taxpayer financing: A primer and research roundup (Journalist’s Resource)

Software on David Crush

Keep Working on the Plan

The Plan Before the Code

What Changed

Rent Seeking

What I’m Trying Next

That Cursor

Sources

Sit On It

They Told Us What It Was For

What I Gave Up

Why I’m Not Going Back

Scroll Post Like

The Echo Chamber Problem

The 24-Hour Wait

Further Reading

Better Off That Way

What ESPN Gave Me

The Import Business

The Good Parts of Parks

Back Through the Gates

Have the Courage

The Sea of Sixteen Months

The Lead They Fired and the Lead They Bought

The Coward’s Layoff

A Slowly Contracting Org

Further Reading

Incorporate It Into Everything

Question 1: The Evidence for RTO

Question 2: The Commute Bill

Question 3: The Checklist

Question 4: The Ten-Second Answer

Further Reading

Sounds Like We Need a Happy Hour

The Happy Hour Method

The Lead Who Carried the Blame

Paid in Full Regardless

The Money Speech

The Pattern

No Conference Rooms

The Productivity Theory

Two Mandates That Cancel Out

When Everything Is Urgent

The Rules Engine

The Building Shuffle

Pre-Ordering Permission

What They Already Told a Court

What the Line Was For

The Off Switch

The Pack

Sources

Filling in the Blanks: What Anxiety Taught Me About AI

The Gap That Cannot Stay a Gap

Completion, Not Lying

Scripts From a Personal Corpus

Vague In, Average Out

Good Uses Are Not the Default

The Bill

Literacy Is the Condition

They Can't Have It Both Ways: Sony, Disney+, Stadiums, and the License You Never Bought

The Library You Paid For

“Reasonable Consumers” Already Knew

How They Actually Sell It

If We Don’t Own It, How Is Piracy Stealing?

Be Fair — Then Don’t Soften It

Disney Tried the Same Inch

Stadiums: Civic Pride as the Inch

What We Are Trained to Accept

Accountability Is the Check

Sources