4 Questions Essay
I was diagnosed with type 1 diabetes in 1988. I was nine years old and a person in a lab coat told me not to worry, we will have a cure in about five years. To a nine year old the lab coat is the authority and who would argue with science?
The five years came and went. Then another five, and another. The institution behind the lab coat was everything you are supposed to respect. Academia. Research grants. Peer reviewed papers. Double blind studies. A giant list of trust building mechanisms. But the innovations from those mechanisms yielded something different. Slightly better insulin at a much better profitability. A meter that needed strips, then a better meter that needed different strips. A pump, then supplies for the pump, supplies upon supplies, refilled monthly, forever. Every few years there would be news about a real cure. Transplanting islet cells. Encapsulation. Something promising in mice. I learned to read those articles the way you learn to read horoscopes. Nothing materialized, and nothing materialized, and nothing materialized.
So as a kid I did the natural thing. I personified the whole institution. I personified the medical device market too. I gave them faces, motives, a story. They knew about the cure and they were sitting on it, because a cured patient is a lost customer. I made them into characters in my head, and I used those characters to not trust a lot of science. Something went wrong? But what is it?
What got smuggled in
In January 2010 the Supreme Court ruled 5 to 4 that the government may not restrict independent political spending by corporations and unions. The reasoning was that political speech does not lose First Amendment protection just because the speaker happens to be a corporation. The immediate consequence everyone talks about is the money. Super PACs, billionaire donors, spending without ceilings. That happened. But something else came in along with the money, and I think it did a lot of damage.
It is worth understanding what the ruling actually argued, because almost nobody does, and the gap between what it argued and what we heard is the whole story. The argument went like this. A corporation is a tool people use to speak together. That is what a union is. That is what a newspaper is. People do not lose their speech rights when they pool them. During oral argument in March 2009, Justice Alito asked the government's lawyer whether its position would allow banning a book published by a corporation. Roberts pressed harder: a five hundred page book that ends with "so vote for X." The answer was that the government could prohibit publication using corporate treasury funds. Alito called it incredible. The Court ordered the case reargued that September on a much wider question, and the exchange is widely credited with the shift, though serious people dispute that it was the cause. It is also narrower than the "government defends book banning" version that circulated afterward: the concession was about which funds a corporation could publish with, not a general power to ban books. Even the story about the story got compressed.
Past that, the case was about machinery. What are the right mechanics for detecting when spending money becomes corruption? The Court needed a definition of corruption that a judge could actually administer, so it picked the narrowest one, an explicit exchange, a bribe. Broader ideas like access and influence and gratitude were rejected, not because the justices thought influence was fine, but because no court can consistently tell corrupt influence apart from ordinary democratic responsiveness. Then there was line drawing. If the New York Times is a corporation and it can editorialize about candidates, what principled rule lets NBC speak but not anyone else with a corporate charter? Nobody could produce one.
You can think the Court got it wrong. Plenty of serious people do. But notice what kind of argument it was. Rules, definitions, administrability, line drawing. It was technical and it was boring. And as a fitting consequence, the way the ruling was received was not technical or boring at all, and the reception is what lasted.
Culturally, one phrase started gaining traffic as the shorthand for why the ruling mattered: corporations are people. Mitt Romney stood on a hay bale at the Iowa State Fair in 2011, got heckled about corporate taxes, and said "Corporations are people, my friend." It was treated as a gaffe, and the headlines about it never really stopped. The Communications Workers of America told its members the Court had decided corporations are people too. Tucson's city council voted in 2012 to abolish corporate personhood, and that was the language of local resolutions all over the country. Stephen Colbert tried to put the question on a ballot in South Carolina, corporations are people versus only people are people, and he probably taught more Americans about the ruling than the news did.
Here is the strange part. Both sides of that fight were saying the same thing. Corporations are people. Corporations are not people. Either way, the frame was personhood. The Court had written an opinion about rules, and the country heard a claim about souls. The ruling was about money and speech. What culture received was that we now live in a world where corporations should be thought of as people. And once you think of something as a person, you start trusting it like one. That is the part that got smuggled in.
The outrage machine finds its fuel
The timing could not have been better, or worse, depending on where you sit. This all happened right as social media was reaching full strength, and with it we discovered how powerful the algorithms that kept users on the platforms could be. Bot accounts pushed the first waves of manufactured news from propaganda arms in Russia. The platforms ran their experiments, mostly on us, and the results came back clear. Lots of things hold attention. One thing worked the best of all. Outrage.
Cambridge Analytica belongs here, but not the way you remember it. The company sold the story that it could read your personality from your Facebook likes and aim messages at your private fears, and that story ran the world for a season. The research since has been unkind to it. Independent work has failed to find much advantage for psychographic targeting over ordinary generic advertising, operatives on both sides of American politics have said there was scant evidence the firm was effective, and one researcher pointed out the obvious thing nobody wanted to hear, which is that Cambridge Analytica was a consultancy selling services and its claims about its own power were sales material. So the most famous villain of the era may have been substantially a marketing deck. Sit with that, because it is the whole thesis in miniature. The outrage economy needed a mind control ray operated by a person with intentions, so it manufactured one out of a pitch, and we all still cite it. Including, until recently, me.
Outrage requires drama. And drama has requirements of its own. It needs a wrong, and it needs someone who committed the wrong. Betrayals are classic and familiar. Every culture has the same stories. The friend who turned. The leader who sold you out. The promise that was a lie the whole time. These stories are pre-installed. You do not have to teach anyone how to feel about them.
But institutions do not fit those stories, not naturally. An agency revising a guideline is not a story. A committee reaching a compromise is not a story. So if you have to explain the drama around an institutional problem, there is a shortcut, and the shortcut is irresistible. Treat the institution as a person. The moment you do, you get access to every trope in the library. Now the agency did not revise a guideline, it changed its story, like a suspect. The company did not follow its incentives, it betrayed you, like a friend. The anthropomorphism is a moralization shorthand. It compresses a structural situation into a character judgment, and character judgments are the native format of the human brain. This works exceptionally well on the internet.
What a wonderful luxury you are given when an institution can be framed as a villain. It requires no more thought. You are gifted moral clarity with no justification needed. What a wonderful simplification, so the existing narrative stays intact.
The outrage became its own economy, and markets refine what they reward. Fifteen years of refinement has bought us a race to the bottom in moral simplification. More polarization. Companies valued by the character of their megalomaniac leaders, because if institutions are people, the man at the top is the institution, so pick stocks by picking protagonists. News whose only reliable product is distrust, delivered daily, of leaders, companies, agencies, science, everything with a logo. And underneath all of it, a public that has been trained by ten thousand repetitions to run exactly one program when an institution enters the frame. The person program.
That sounds grim. It is somewhat grim. But there is a way through, and it lies in noticing precisely what got smuggled in. As culture got used to corporations being people, we ended up using the same trust techniques on institutions that we use on individuals. Read that again, because the problem is narrower than it looks. The problem is not that we trust too much or too little. The problem is the techniques. We are using the wrong instrument, and the wrong instrument does not just fail, it fails in a pattern. It condemns the healthy behavior and rewards the fake.
The program you are running
If I say "don't trust the CDC," you already know what the sentence means. You hear: they lied to you. They got caught. They are not who they said they were. You did not have to reason your way to any of that. The meaning arrived assembled, the way the meaning of "my brother in law is a liar" arrives assembled. That is the tell. The sentence works because your brain ran the program it uses for people. Someone looked you in the eye and said something false. That is the movie that plays, complete with a face in it.
But nobody looked you in the eye. There is no eye. The CDC is thousands of employees, an org chart, a budget process, and a stack of procedures, most of which contradict each other a little, because thousands of people wrote them across decades for different reasons. It cannot lie to you the way a person can, because it cannot do anything the way a person can. It has no story to keep straight. It also cannot love you, cannot be loyal to you, and cannot be sorry, no matter how much its press office insists otherwise. Not because it is evil. Because it is not a person. Expecting it to behave like one is not high standards. It is a category error, and the error has a body count of trust.
Here is the same story told twice, and I want you to actually feel the switch, because the switch is the entire skill this essay is trying to teach.
On February 29, 2020, the Surgeon General told the public to stop buying masks, that they were not effective for the general public, and that hoarding them put healthcare workers at risk. On April 3, thirty-four days later, the CDC recommended cloth face coverings in public. Those are the facts, agreed on by everyone.
Reading one. They told us one thing, then they told us the opposite. They changed their story. People who change their story are lying at least once. Therefore they lied to us, and you do not get fooled twice, so never trust them again. Notice the properties of this reading. It is fast. It is emotionally complete. It asks nothing further of you, no follow up questions, no homework. It also permanently closes the file.
Reading two. An organization took a public position, the evidence changed underneath it, and it reversed itself in front of everyone, in five weeks, at obvious cost to its own standing. Now different questions open up. How did the evidence change? Who caught it? How long did the reversal take, and what was the delay made of? Is a fast public reversal a good sign or a bad one in a body that processes evidence for a living?
Same facts. Opposite conclusions. The difference is not the evidence, and it is not intelligence, and it is not politics. The difference is which program you ran. And almost all of us run the first one, every time, on everything, because it is the only program we were issued.
Part of the problem is that we are missing a word. English has one word for trust and it has to cover everything. Japanese has two. Shinrai is the bet you place on a person, on their character, their intentions, what kind of human they are. Anshin is different. Anshin is the security you feel because the structure makes betrayal a bad move. You do not need the bank teller to love you. You need the audit to exist, the insurance to be real, the camera to be on. A social psychologist named Toshio Yamagishi spent a career on this distinction and found that whole societies can run mostly on one or mostly on the other without noticing which one they are running. We have been demanding shinrai from things that can only ever produce anshin, and then acting surprised when the demand comes back unfilled, or worse, comes back filled with a performance.
Because here is what person trust actually checks. Four things, more or less. Is he sincere. Is he consistent. Is he loyal to me. Is he sorry when he screws up. These are excellent questions for a friend. Sincerity is hard to fake face to face. Consistency across years of dinners is real evidence. Loyalty gets tested and observed. An apology from someone who shares your life costs him something, and you can watch whether the behavior changes.
Point those same four questions at an institution and every one of them breaks. Not weakens. Breaks. Sincerity at institutional scale is a product with a marketing budget. The committee wrote the apology and legal reviewed it, which is why it sounds like a committee wrote it and legal reviewed it. Consistency, as we just saw, punishes exactly the institutions that correct themselves. Loyalty is the one thing a fair institution must refuse to give you, because an agency loyal to you is corrupt, by definition, and you would agree instantly if it were loyal to your neighbor instead. And the sorry of an institution costs the institution nothing. The spokesperson feels no shame. There is no one in there to feel it.
So the honest institution cannot produce the signals, and the fraudulent one can produce them cheaply and beautifully. Run the person program on institutions and you have built a detector that fires on the innocent and stays silent for the guilty. That is not a broken detector. That is worse. It is a detector installed backwards.
Replace the four questions. Here are the four I use instead. None of them are about anyone's character.
One. Can I identify the incentives?
Not "are these good people." You will never meet these people, and it does not matter. The people turn over. Whatever character you think you observed in an institution belongs to employees who may be gone next year, and the institution will still be there, behaving the same way, because the thing that makes it behave was never the people. The question that survives turnover is structural. What does this arrangement make profitable? What does it make punishable? What does it make invisible? And the simplest version of all: who actually pays them, and for what?
In 2008 the credit rating agencies became famous for stamping AAA, the highest grade there is, on mortgage bonds that turned out to be junk. Trillions of dollars of retirement money trusted those letters. When it all came down, the person frame asked its usual question. Were the analysts corrupt? Were they stupid? Congress held hearings looking for the villain. Wrong question, and years too late. The structural fact had been sitting in the open the whole time: the issuer pays model. The banks selling the bonds were the ones paying for the ratings. The agency that graded too hard lost the customer to the agency down the street that graded easier. No villain required. No conspiracy meeting. The incentive did all the work, quietly, through a thousand small accommodations, none of which felt like corruption to the people making them. Anyone who asked "who pays them" could have predicted the failure years early. Almost nobody asked, because the question is boring, and we were busy asking whether the people were good.
To be fair to the agencies, they made the opposite argument, and it is not stupid. Ratings were set by committees, not individual analysts, and analysts were not paid on the ratings they gave. The academic literature on whether issuer paid agencies systematically inflate is genuinely mixed, and at least one study found the investor paid alternative carries conflicts of its own. Notice that this is what a real argument about an institution sounds like. It is about structures, and it can be checked, and it might change my mind. Compare it to what we actually spent 2008 doing, which was looking for someone to hate.
The 737 MAX is the same lesson in a different industry. Two crashes, 346 people. The person frame went hunting for the bad engineers and the greedy executives, and it found some material, it always does. But the load bearing fact was structural. The FAA, underfunded and outmatched, had delegated much of the certification work back to Boeing itself. The company was grading its own homework, under schedule pressure to beat Airbus to market. And the warning was on the record seven years before the first crash: in 2012 the inspector general reported that some employees in the FAA office overseeing Boeing felt they could not raise compliance concerns without fear of retaliation. You did not need to know a single Boeing employee's name. The incentive map was the forecast.
And once you have the habit, you see it everywhere, including in softer places. Why is the news so relentlessly negative? The person frame says biased journalists, and entire media empires run on selling you that answer. The structural answer is shorter. Outrage is the product because attention is the revenue. The journalists could all be replaced by saints tomorrow and the output would barely move, because the saints would face the same dashboard.
It works in the positive direction too, which matters, because this question is not just a way to catch failure. The Federal Reserve was built with long staggered terms and a budget Congress does not control. Not because anyone believed central bankers would be trustworthy people. The design assumes they might not be. The structure was set up to make the easy betrayal, juicing the economy before an election to help whoever is in power, hard to do and easy to see. The founders had a phrase for this move: ambition must be made to counteract ambition. It is the oldest trick in institutional design. Notice what it never depends on. It never depends on anyone being good.
Ask who pays them. The answer is usually public, usually boring, and usually the whole story. Boring is where the answer lives.
Two. Can I see the mistakes and the corrections?
Start with a piece of arithmetic, because it retires a question you have probably been asking your whole life. Take an institution making a million decisions a year. Claims processed, patients seen, flights dispatched, cases decided. Give it an accuracy rate no human endeavor has ever achieved, 99.9 percent. It still produces a thousand failures. A thousand, every year, forever, at a performance level that does not exist. At scale, being wrong is not a possibility to be alarmed about. It is a certainty to be planned for. So the question "have they made mistakes" carries no information at all. Of course they have. If you cannot find any, that is not cleanliness, that is concealment. The only questions left are the ones about what happens next. Did the mistake surface, or did it get buried? Did we find out from them, or from a leak? And when it surfaced, did anything change?
Here the person frame does not just mislead. It inverts. In a person, changing your story is the red flag, and rightly so. Memory plus honesty produces consistency, so revision means something was false. In an institution the sign flips completely. An institution that never revises anything is not honest. It is dead. The correction machinery has stopped, and the errors, which are still arriving on schedule, are going somewhere you cannot see.
Commercial aviation is the proof that this can be done right, and it might be the most underrated institutional achievement of the last century. Every crash, every near miss of consequence, gets an independent investigation by a body whose only job is finding out what happened. The findings are published for anyone to read. The rules change in response. Blame is not the point and is often explicitly set aside, because blame makes people hide things, and hidden things crash the next plane. The result of running this loop for decades is that flying went from routine disaster to the safest way to travel, by a wide margin. Think about what your trust in flying is actually made of. It is not the pilots' character. You have never met them. It is not the airline's sincerity. You have heard their apologies. It is the pipeline that turns each crash into new rules. That pipeline is what trust in an institution should mean, and it is checkable, and it is public.
The Catholic Church abuse scandal is the same test, failed, and it is worth being precise about what the diagnostic was, because it was never what the person frame looks at. Every parish trusted its priest, and the trust was person trust of the highest grade. Warm, familiar, tested across years of weddings and funerals. That trust worked exactly as designed, and it protected the institution instead of the children. The fact that would have caught it was structural and it was boring. The reassignment pattern. Priests moved quietly from parish to parish, the same names resurfacing, the paperwork accumulating in diocesan files for decades. Errors never surfaced from inside. They surfaced through journalists and courts, which is to say through force, applied from outside, against resistance. When an institution's failures are only ever discovered by outsiders, that is the audit result. You do not need to wait for the scandal. You need someone reading the paperwork structurally, and for decades, nobody was.
Volkswagen ran the same pattern in miniature, and the details are almost too on the nose. The diesel emissions cheating was not caught by any internal process, and not by the regulator's lab tests, which the cars passed. It was caught by engineers at West Virginia University, working on a sixty nine thousand dollar contract from a nonprofit, driving rented VWs around California with a portable tester because someone wondered why the road numbers did not match the lab numbers. One of the cars exceeded the standard by a factor of fifteen to thirty five.
Then comes the part that should bother you. They presented the results publicly in San Diego in the spring of 2014, with EPA staff in the audience, and the sponsor published the whole thing online. It sat there, checkable by anyone, for more than a year. The auto press missed it. The EPA and California regulators did not, and they spent that year doing the follow up testing that produced the violation notice in September 2015, while VW told them the discrepancy was a technical glitch. So the machinery did work. It just worked slowly, and only because a handful of people were reading a published document that everyone else scrolled past. Verification is not automatic. It requires someone to actually verify.
Which brings the CDC masks story home, with the resolution it deserves. The reversal that half the country filed as "they lied" was the correction machinery firing in public, in five weeks, driven by studies from February and March showing that people without symptoms were spreading the virus. That is the good part, and the person frame read it as the crime. But there is an honest critique in the story, and it is different from the popular one. In the early weeks, officials communicated with more confidence than the evidence justified. Partly to manage a real mask shortage, which was never even hidden, it was stated in the same breath as the effectiveness claim, and partly for a worse reason. They knew how they would be graded. They knew the public runs the person program and scores consistency, so they performed consistency, the person frame signal, instead of offering calibrated uncertainty, the honest institutional one. We taught them to do that. And then we punished the correction when it came, which teaches every institution watching that the smart move next time is to hide the error longer. This is the vicious loop at the bottom of the whole problem. The person frame does not just misjudge institutions. It trains them to get worse.
Three. Can someone outside verify the claims?
There is a small piece of theory here worth having, because it explains in one stroke why everything institutions currently do to earn trust is worthless. A signal only carries information if it is expensive to produce when false. That is the whole theory. A peacock's tail means something because a sick peacock cannot afford one. Face to face, human sincerity works on the same principle. Genuine warmth, spontaneous emotion, shame that reaches the eyes, these are hard to fake in a living room, which is why your person trust instincts rely on them and why those instincts served your ancestors well.
At institutional scale, every one of those signals becomes cheap. Sincerity is a deliverable. The values statement cost a consultant two weeks. The apology was drafted by committee, softened by legal, and rehearsed by a media coach, and the CEO who tears up on camera is billing the performance to the marketing budget. None of this is even scandalous. It is just what those signals cost to produce at scale, which is approximately nothing, which means they carry approximately no information. Seeming honest tells you nothing. What tells you something is the thing a fraud cannot afford to build: real verification, run by someone who does not work for them, with the results where you can see them.
In 1982 someone in the Chicago area poisoned Tylenol capsules on store shelves and seven people died. Johnson and Johnson's response is taught in business schools as crisis management, but look at what they actually did, because the lesson is usually mislabeled. They did not lead with how much they cared. They pulled every bottle in the country, thirty one million of them, about a hundred million dollars of product, against short term interest, and they did it over the objection of the FDA and the FBI, who thought a national recall was an overreaction. Then, forty three days after the first death, they did the thing that matters. They shipped the tamper evident cap. The foil seal, the plastic ring, the click. A fix you could verify with your own thumb, in the store, in two seconds, forever. Nobody had to trust their sincerity, because the proof was in your hand. That click when you open a bottle is what an institutional apology sounds like when it is real.
Now run the contrast. BP after the Deepwater Horizon spill. Eleven workers dead, oil flowing into the Gulf on a live camera feed for months. The response led with the person register. Apology advertisements. Commitments to make it right. The chief executive telling cameras he would like his life back, a sentence he presumably did not focus group first. Maximum sincerity, minimum checkability, and the sincerity made it worse, because everyone could see what it was, which was a performance purchased at the exact moment performances are cheapest to buy and worth the least.
FTX is the recent version and maybe the purest. Billions of dollars of customer money, and by the sworn account of the man who cleaned it up, no board of directors and virtually none of the systems or controls you would expect of anyone holding other people's assets. What it had instead was a founder optimized for person trust. The rumpled genius in shorts who does not care about money and sleeps on a beanbag. Journalists loved him. Politicians loved him. Sophisticated investors who would never buy a stock without an audit handed him billions on vibes, because the vibes were excellent, and vibes were the entire security system. When it came apart, the internal messages put the hole at eight point one billion dollars of customer assets that were not there.
I want to be careful with that number, because the essay owes you the same standard it is demanding. The bankruptcy estate later recovered enough to pay customers back in dollar terms, and two law professors have argued the eight billion loss figure the prosecution used was never really substantiated. Fine. The shortfall on the day the withdrawals came was real, the absence of controls was real, and the fact that a rising crypto market later filled the hole is not a defense of the design. That is the point, actually. Person trust selected exactly wrong, and it took a bull market to disguise how wrong, while the boring exchanges with boring audits kept boringly working. You may have noticed that boring keeps showing up in this essay wearing a medal. That is not an accident. Checkable and boring travel together, and so do thrilling and unverifiable.
So the question is never "do they seem honest." The question is what can be checked, and by whom, and whether anyone has actually done it lately. If the answer is "nothing, but they seem very sincere," you do not have a gap in your information. You have your answer.
Four. Can I leave?
Trust between people is secured by something neither side ever says out loud. You have both invested in the relationship. Years, favors, secrets, standing in a shared community. Betrayal costs the betrayer the relationship itself, and everything stored in it. That is the collateral, and it is why person trust works without contracts. Your friend is not good to you because a rule requires it. He is good to you because losing you would cost him something real.
An institution holds no such collateral with you. You are one of millions. Your history with it is a row in a database, and your individual exit is noise in its quarterly numbers. It cannot fear losing you, because it will not notice losing you. So the substitute has to be structural, and there is only one: exit at scale. Competition. Alternatives. Appeals. Somewhere else to go, and an institution that knows you know it.
Where exit is impossible, trustworthiness becomes optional, and eventually optional becomes absent. Equifax is the cleanest specimen I know. In 2017 it exposed sensitive data on about 147 million people. Names, birth dates, 145 million Social Security numbers, the skeleton keys to a financial life, after failing for four months to install a patch its own security team had said to install within forty eight hours.
It did pay. The settlement with the FTC, the CFPB and fifty states and territories was at least five hundred seventy five million dollars and possibly seven hundred, the largest data breach settlement in American history. Notice where every dollar of that came from. Regulators, attorneys general, courts. Not from customers leaving, because here is the structural joke: none of those 147 million people are its customers. You cannot fire your credit bureau. You never hired it. The lenders are the customers. You are the product, and the product cannot leave. Equifax is still there, still holding your file, and your file did not move, because there is nowhere to move it to. Where exit does not exist, the entire burden of discipline falls on regulators, which is a load bearing wall made of one material. An institution whose users cannot exit and do not pay has no structural reason anywhere in it to be trustworthy, and its behavior will find that floor eventually, no matter who works there, no matter what the values page says.
The platforms run a softer version of the same trap. Every scandal of the 2010s produced its Delete Facebook moment, and the graph barely moved, because leaving costs you your network, your history, your groups, the birthday reminders for people you would otherwise forget. Exit that expensive is not exit. And an institution structurally insulated from the outrage can host the outrage economy all day long, which, if you think about it, is exactly what happened.
Which brings me back to 1988.
I could not leave either. Not for a day. The nine year old with the characters in his head could not switch products, could not boycott, could not take his business elsewhere, because his business was staying alive. I was a captive market before I could spell either word. And here is the thing I want to say carefully, because it is the hinge of this whole essay. The kid's grievance was not wrong. Something really was off, and the fourth question names it. Treatments kept improving and the cure kept receding, and a market where the customer can never leave, where the recurring revenue is guaranteed by biology, has weak structural pressure to deliver the one thing the customer actually wants. That is not a conspiracy. It does not need to be. Nobody has to sit in a room and suppress the cure. The incentive just has to point where it points, year after year, in ten thousand funding decisions and product roadmaps, none of which feel evil from inside.
My complaint was correct. My frame was not. I filed a structural problem under character. I made villains out of org charts. And then, because person trust is holistic, because it transfers, I let the villains stand in for everything wearing the same coat, and I walked away from a lot of science that had done nothing but keep me alive. That is what the person frame does with a real grievance. It takes a defensible structural observation, converts it into a character judgment, and then spreads the judgment across an entire category. One betrayal by "them" becomes all of them. It is the exact move a country made with "corporations are people," and I had made it alone, at nine, about my own body.
And I should finish the story honestly, because it did not end where the nine year old left it. In 2000 a team in Edmonton showed that transplanting islet cells could actually free people from insulin, limited by donor supply and a lifetime of immunosuppression. In 2025 a trial of lab grown islet cells reported that ten of the people who received them no longer needed daily insulin after a year. Not five years. Thirty seven, and counting, and still with caveats. So the thing arrived, late, and it arrived from the part of the institution that was never selling me supplies, which is exactly what the incentive map would have predicted if I had ever bothered to draw one instead of casting a villain. The kid was right that something was wrong and wrong about what it was, and it cost him a couple of decades of trusting the wrong things for the wrong reasons. That is the whole price of the bad frame, paid in one life, and I got off cheap.
One more thing lives under this question, and it is the darkest thing in the essay, so I will keep it short. Pay attention to who performs loyalty at you. When someone at the scale of a company or a country says only I understand you, only I am on your side, only I can fix it, your person trust wiring hears a friend. Read it structurally instead. It is a lock in pitch. Loyalty is the costume, capture is the product, and the people most invested in you running the person program are the ones building the exits shut behind you while you gaze at their sincere face.
The card
So that is the card. Four questions. Who pays them. What happened last time they were wrong. Who can check from outside. Can I leave.
Read the list once more and notice what is missing. There is not one question about character. Not one about sincerity, intentions, or anyone's feelings. That absence is not an oversight. That absence is the method. Character can be performed. These four things can only be built, and you can check whether they were.
How does anything change? From two directions at once, and neither one needs permission, legislation, or a movement with a logo.
From below, it is literacy. Nobody legislated media literacy either. It got taught, it got named, and it slowly became embarrassing to lack. This one spreads the same way, in small moves. A reader asking "who pays them" under a viral story instead of sharing it. A voter grading a candidate on whether he concedes losses rather than whether he would be good company at a barbecue, which is a real thing we have polled for, in a country choosing who commands its military by the standards of choosing a drinking buddy. A journalist ending the interview not with "do you take responsibility," a sincerity question that comes with a scripted answer, but with "what is now impossible to do twice," a structural question whose answer can be checked in a year. The outrage economy exists because we pay in clicks for villain stories. Attention is the revenue. Change what you pay for and you change what gets supplied. That is not idealism. That is the one law of media economics that has never once failed to operate.
From above, institutions need this more than we do, and most have not noticed yet. They are losing the trust game while spending record amounts on it, because every dollar goes to person frame signals. Sincerity campaigns. Values statements. Apology tours. Performing character is a game an honest institution loses to a demagogue every single time, because the demagogue produces the signal cheaper and better. It is his native language. The four questions are the only field where a sound institution holds an advantage that cannot be faked. A real corrections log. A real outside audit. A real exit option, offered freely. A fraud cannot counterfeit those, because building them is the one thing a fraud cannot afford to do. They would catch him. Which means legibility is a first mover opportunity, the way organic labels and crash test ratings and security audits each started as one company's differentiation and hardened into everyone's table stakes.
And the two directions feed each other. Demand makes legibility valuable. Early legible institutions advertise it. Their labels make the four questions easier for everyone to ask. The questions get asked of the holdouts, louder. Neither side has to move first, because each side moving makes the other side's move cheaper. That is the kind of loop that starts slow and then does not stop.
What would the tools look like? Less exotic than you would think, because every one of them already exists somewhere, and the work is only in making them normal. A "who pays us" page as standard equipment on every institutional website, one screen, standardized, so the incentive map is never more than a click deep. Corrections logs as a cultural norm, the way software teams already publish postmortems and status pages without shame, exported to hospitals, agencies, school districts, newsrooms. Independent investigation boards on the aviation model for the domains that lack them, medical error, algorithmic failure, so that correction stops depending on leaks. Public registries where institutions state in advance what would change their mind, so the claims are checkable later instead of deniable later. And portability rules with teeth, so your records, your data, your money, and your history leave when you do, because exit rights are trust infrastructure. They make institutions behave as if they cared, without requiring anyone inside to actually care, which is the only kind of caring an institution can do.
Fair is fair, so run the card on this essay. Who pays for it? Nobody yet, which you should weigh accordingly. What happens when it is wrong? There is a corrections list at the bottom of the page, it already has entries, and it will grow. Who can check from outside? Every factual claim here, the ratings agencies, the crashes, the recall, the breach, is public record, and the essay's best number gets a footnote precisely because it is the best number. Can you leave? You just did, or you are about to. No account, no subscription. An argument about checkability that cannot be checked refutes itself, and I would rather you catch me than believe me.
Three billion heartbeats
I want to end on a number, because the number is the good news, and there has not been much of it so far.
Nearly every mammal gets about a billion heartbeats. A pygmy shrew spends them in a year and a half at thirteen hundred beats a minute. An elephant takes seventy years at twenty eight. The totals land within spitting distance of each other, one of biology's stranger bits of bookkeeping, and it falls out of the scaling math: heart rate drops with body mass to the minus one quarter, lifespan rises with it to the plus one quarter, and the mass cancels. Humans used to be on that schedule too. For most of our history, a billion heartbeats, thirty some years, was the standard allotment.
We are not on that schedule anymore. We get somewhere between two and a quarter and three billion, depending on whose numbers you take, and the disagreement between those sources is smaller than the gap between any of them and the rest of the mammals. And nothing changed in the organ. The heart in your chest is the same pump our ancestors carried. What changed was everything around it. Sanitation. Clean water that does not carry cholera. A food supply that does not fail with one bad harvest. Vaccines. Antiseptic surgery. Courts that resolve disputes without feuds. Schools. Records. All of it machinery, none of it glamorous, built across centuries by people who mostly never met each other, most of whom got no credit and are not remembered by anyone alive. There is no face to put on it, which is exactly why we never tell it as a story. And before anyone says it, no, this is not just fewer babies dying. That is the standard deflection and the data does not support it. A fifteen year old in France in 1816 could expect to reach fifty eight. Today, eighty three. A sixty five year old then could expect seventy six, and now eighty six. The gains show up at every age, which means the machinery did not merely stop killing children, it kept extending adults who had already made it through.
Two billion extra heartbeats per person is what functioning institutions bought us, and it might be the largest thing our species has ever done, and we are nearly blind to it, because it does not look like anything. It looks like the water being fine.
So this is not maintenance work on a tired inheritance. Building institutions is the thing we turn out to be good at, maybe the best at, better than we are at almost anything we actually celebrate. And we are about to need that skill at full strength, because the problems arriving now are exactly the kind that no hero fixes. AI. Climate. The poverty we have never finished solving. Every one of them is a coordination problem at a scale no person frame can even hold in view. None of them get handled by finding the right person to believe in. That approach is currently on offer, loudly, and it is the oldest product in the catalog. They get handled the way the last two billion heartbeats got handled. By building machinery that works even when the people inside it are ordinary, tired, and occasionally wrong. And by enough of us knowing how to check whether it does.
Corporations are not people. Neither is anything else with a logo. I understand now that this is not the bad news. It is the reason everything can be fixed. You cannot rebuild a character. You cannot audit a soul. But incentives, corrections, verification, and exit are all just machinery, and machinery can be built, and inspected, and built better. We are the species that builds it. We have the heartbeats to prove it.