The Machine

Why good people build a bad world.

I.

It’s seven in the morning at TB Simatupang. You’re in traffic that doesn’t make any sense. You’re angry at the driver next to you who just cut in from the shoulder. He’s angry at the TransJakarta bus stopped in the middle of the lane. The TJ driver is blaming the ojol weaving around without a signal. The ojol is blaming the pedestrian who crossed at random.

Everyone has their own antagonist. Nobody has the real antagonist.

A better question. Who, specifically, wants Jakarta to be jammed?

Not you. Not the driver next to you. Not the ojol, the TJ, the pedestrian. Not the provincial government, which loses billions a day to productivity evaporating in stationary vehicles. Not the real-estate developer whose property value falls because access to it is hellish. Not the company whose employees arrive at the office already tired before the workday begins.

Nobody chose this. But it happens. Every single day.

Change the example. Narcotics. BNN records transaction flows in Indonesia in the trillions of rupiah, with the user profile getting younger and poorer. Who wants this? The dealers, obviously. But ask an individual dealer honestly, he might say he wants out too. If only his competitor would leave first. The users don’t want to be addicts. The users’ families don’t want their savings to vanish into something that evaporates in minutes. The couriers don’t want to get shot in the leg. The government doesn’t want the country’s reputation to become a case study in a regulatory-failure paper.

No antagonist. Only victims. And one industry that keeps growing.

The two examples feel different on the surface but are structurally identical, and that isn’t a coincidence. Jakarta gridlock and the narcotics trade are manifestations of one mechanism. The same mechanism runs the deforestation of Kalimantan. The TikTok content that’s frying kids’ brains. The money politics that makes the best-record candidate lose to the candidate with the thickest envelope. The tobacco industry whose executives know the product kills but cannot unilaterally stop without a competitor filling the gap in weeks.

Something is running behind the scenes. It has no head office. It has no CEO. It has no cabinet meeting. It has no intent in any meaningful sense. And still, it is writing most of modern human history.

I’m going to call it The Machine. (Originally “Moloch”, from Slate Star Codex’s “Meditations on Moloch”.)

Now. Before you close the tab.

I know how that word sounds. I know who usually uses it with a capital letter. The thirty-eight-year-old Discord moderator who lives in a basement with Reddit and Wikipedia-on-Bilderberg tabs open next to a Notepad++ window holding a twelve-thousand-word “manifesto”. The high schooler who’s been on 4chan too much and likes to say “they don’t want you to know”. Basically Ted Kaczynski without the bombs but with a podcast.

I don’t believe there are Reptilians in the Federal Reserve. I don’t believe there’s a cabal of lizard people in Davos pulling strings. That’s exactly why I use the term The Machine.

Every conspiracy theorist who’s ever lived holds one intuition that, beneath all the bad analysis, feels true in a way that’s hard to describe. There is a pattern. It isn’t a coincidence that everything ends up in the same place.

They are right about the pattern existing. They are wrong about its contents. Wrong on a spectacular level, because they assume the pattern needs conspirators. Needs someone in the shadows. Needs a secret meeting where The Decision is made.

It doesn’t. The Machine is worse than any conspiracy you can imagine, precisely because it doesn’t need conspirators. It doesn’t need anyone who is consciously aware they’re serving it. It only needs incentives. Incentives are everywhere, all the time, operating on every one of us, since before we were born.

The main question of this essay isn’t “who is evil”. The main question is, what makes individually good people produce a collectively bad world?

If your reflex answer is “greed” or “capitalism” or “the oligarchs”, you’re half right. Half right, on a functional diagnosis, is the same thing as wrong. Those explanations contain real signal. But they point at the symptom, not at the mechanism. Without understanding the mechanism, any intervention you design will lose before it begins. Look at the evidence. Anti-corruption campaigns that flop. Environmental regulations that don’t bite. Moral movements that evaporate in two seasons. Diagnosis that stopped at the surface.

To get to the mechanism, we have to go back. A long way back.

All the way to the first ocean.

II.

To understand The Machine, we need to know where it was born.

Not parliament. Not a boardroom. Not a cabinet meeting. The Machine was born in a pool of water on Earth roughly four billion years ago, before the first cell existed, in a vast ocean whose layout looked nothing like today’s.

What existed in that pool was molecules. Bumping into each other for millions of years. Most of them broke down quickly. Until at some point one molecule formed that had a new property. Abiogenesis.

It could copy itself.

Whether it was RNA, peptides, or something more exotic, the chemistry isn’t the point. The implication is. The moment one self-replicating molecule appeared in a pool full of other molecules, the math changed permanently. Random competition between mortal molecules became competition between entities that produced more of themselves.

A simple law went into operation. I’m writing it in bold because it will reappear throughout this essay.

Whatever is better at replicating itself will become more numerous.

It requires no consciousness. No intent. No God, designer, or soul. Just time, matter, and minimal initial conditions. Four billion years later, you are a product of that law running without rest from the first second.

This is what Dawkins meant when he called genes “selfish”. Genes don’t have intent to be selfish. Genes don’t have intent at all. Genes are chemical sequences that are more or less effective at producing copies of themselves, via the organisms they build.

“Selfish” here is a technical term. A successful strategy is a strategy that makes more copies. Not necessarily moral. Not necessarily beautiful. Just sufficient.

Hold on. This is usually where people stop, either agreeing too fast or feeling insulted as “just animals”. Both reactions are wrong.

If the law above were the only one running, the biological world would be entirely solitary. Every organism killing its sibling for an extra calorie. Abandoning its offspring the moment they left the womb. But that isn’t the world we see. Bees die defending the hive. Mother birds go hungry for their chicks. Volunteers run into burning buildings for strangers.

How does that cold law produce this?

The answer was summarised by William Hamilton in 1964, in an inequality that looks like high-school algebra but changed biology forever.

rB > C

r is genetic closeness, B is the benefit to the receiver, C is the cost to the giver. A gene that codes for altruistic behaviour will be preserved as long as rB > C. That’s why parents sacrifice for children. It’s why you’re more likely to lend money to a cousin than to a stranger. It’s why a bee colony, sharing 75% of its genes, literally dies for itself.

In 1971, Robert Trivers added reciprocal altruism. Cooperation can be stable even between individuals who aren’t genetically related, as long as the interaction repeats, there’s memory of who cooperated vs defected, and there’s a way to punish defectors. The foundation of gossip, reputation, the social contract we all live inside.

We inherit two subsystems that are equally old, equally deep, equally real. One pulls us toward individual interest. The other pulls us toward cooperation, where conditions allow.

What determines which one is active at any given moment is the configuration of incentives around us.

III.

II gave us an organism with two subsystems, selfish and cooperative. The question is what determines which one runs.

The answer isn’t personality. It isn’t faith. It isn’t your parents.

All three still matter. But they operate on top of a more fundamental substrate. That substrate is the configuration of incentives around us at the moment a choice is being made.

Charlie Munger once said a single sentence which, if you really get it, makes most of this essay unnecessary. “Show me the incentive, and I’ll show you the outcome.”

Not probabilistic. Not a tendency. Outcome. Munger spent sixty years observing humans in high-stakes decision-making positions, from Fortune 500 CEOs to politicians, and he came out with an uncomfortable conclusion. In the long run, in almost every case, incentive structure beats personality.

A few examples.

Pay an annual bonus based on short-term profit to a finance executive. You get the 2008 crisis. None of them designed the crash. They were just optimising what you asked them to optimise.

TikTok has one objective function, watch-time. The result is a content industry currently rewiring the reward systems of primary-school children. The engineers at ByteDance don’t want that. They’re just optimising the metric they’re paid for.

Pilkada in most of Indonesia chooses winners by envelope distribution on election day. The parliament that gets formed is good at distributing envelopes, not at drafting legislation. The good candidate with a thin wallet was filtered out in the first round.

Even altruism operates inside this field. Donations go up when the tax structure protects the giver. Volunteering goes up when there is public recognition. Philanthropy flows most strongly from people whose name can be put on a building. This doesn’t deny sincerity. Even the most sincere motivation operates inside an incentive structure, not separate from it.

From all of this, one claim.

You don’t fix outcomes by fixing people. You fix outcomes by fixing incentives.

This will feel cynical to some readers. Cynical isn’t automatically wrong, though. And the alternative, trying to fix outcomes by fixing people, has been tried for a few thousand years, and the track record is pretty clear.

IV.

II gave us an organism that, by default, operates in its own interest. A mechanical description, not a moral one. It is the product of a selection process that has already killed its ancestors who didn’t. III gave us incentives as a field surrounding every choice. The substrate where every choice actually happens.

What’s left is to put the two together. One term from economics and game theory, and one definition sharp enough to leave no escape.

The term is multipolar trap.

The definition. A situation in which every actor, acting rationally as an individual, chooses the option that produces an aggregate outcome which is bad for every actor, including himself.

Let me say it again with more sentences, because this claim has to land. From every player’s point of view, defection is the dominant strategy, the choice that benefits you at the individual level regardless of what other players do. Because all players are rational, all of them defect. Because all of them defect, the collective outcome is worse than if all of them had cooperated. But no individual player can change the outcome unilaterally. If he chooses cooperation alone, all that does is make him lose faster.

Now look at Jakarta traffic through these glasses.

Every private-vehicle user makes a choice that is rational from his point of view. More comfortable than a packed TransJakarta. Safer than a motorbike in the rain. If the KRL schedule doesn’t fit, the private car goes anyway. Rational individual choice, multiplied by millions of vehicles, produces a city that cannot move.

The fix is structural, not moralising. “Reduce private vehicles”, stuck on pedestrian bridges, doesn’t work. I’ll demolish the moralising approach in V. What works is congestion pricing, the way Singapore and London do it. Transit that is genuinely faster than driving. Disincentives that show up directly in the wallet.

Now narcotics.

Every dealer still operating is making a rational choice. Massive market. Fat margins. Legal risk that can be arbitraged through offshore jurisdictions. Any dealer who exits unilaterally just hands his market share to a competitor in weeks. Every middleman accepting his transactions is also rational. Insane volume, attractive profit splits.

Are there dealers who feel guilty about the divorces they cause? Possibly. Are there middlemen and couriers who lose sleep over the savings of poor families moving into the shell companies of their bosses? Maybe. But the multipolar trap doesn’t care about feelings. The multipolar trap only cares about one question. If actor X exits, will actor Y enter and fill the slot?

For narcotics, the answer is always yes. As long as the answer is yes, individual exit is not a solution. Individual exit is a sacrifice with no effect.

The same pattern runs the cases I mentioned in I. The candidate who refuses envelope politics loses to the one who doesn’t, because the electorate in aggregate still responds to envelopes. The tobacco executive who wants to close his factory knows it will be bought by a competitor within three months. The content creator who refuses engagement bait gets out-competed until he has to choose between compromising and quitting.

This is what I’m calling The Machine.

The Machine is not a thing. Not an organisation. Not a social class. The Machine is the name I’m giving to the attractor state that emerges from interactions of many rational agents, none of whom can change the outcome unilaterally.

The Machine emerges from below, from micro interactions, with no one above designing it. This is what makes it more dangerous than any conspiracy. If your enemy is Bilderberg, you can expose Bilderberg. If your enemy is Davos, you can protest Davos. If your enemy is an incentive configuration with no face, you have no target.

And this is what makes individually good people produce a collectively bad world. They are not secretly evil. Inside a multipolar trap, “good” at the individual level does not guarantee “good” at the collective level. Two different variables. But we, from primary school through Friday sermons through motivational posts on LinkedIn, are taught over and over that they are the same thing.

That is the most dangerous lie we still maintain as a civilisation.

V.

Every culture that has ever faced a multipolar trap has had the same default response. Not structural reform. Not redesigned incentives. The Sermon.

“Don’t take bribes.” “Don’t be greedy.” “Think of your grandchildren.” Every billboard. Every pulpit. Every presidential address. Every corporate CSR campaign. Every Instagram caption from a public figure. Every parent giving advice to their kid.

All. Of. It. Does. Not. Work.

The audience agrees, mostly. The majority, if asked, will say they intend well. The majority, if observed, defect anyway when the moment of choice arrives.

The cause is structural. Inside a multipolar trap, defection is the dominant strategy, regardless of your intent. Good intent from 95% of players doesn’t save anything as long as the 5% who defect still win. The tobacco industry doesn’t need 100% of people to smoke. It needs enough. Narcotics doesn’t need 100% of players. It needs enough. Money politics doesn’t need 100% of candidates to play. It needs enough to filter out the ones who don’t.

Worse. Sermons that don’t touch incentives often become the mechanism that props up The Machine. Angry energy gets channelled into cheap moral gestures. Posting frustration about corruption. Cursing big pharma. WhatsApp statuses about the broken youth of today. The energy that ends up there is energy that doesn’t end up driving structural change. From The Machine’s point of view, this is the ideal outcome. A population that feels like it has already done something while the world keeps moving in the same direction.

If you’ve ever read Taleb on virtue signalling and nodded, but your feed is still full of moral posts not followed by any structural action, you are the subject of this paragraph. So am I.

VI.

Stop sermonising. Start engineering.

That’s one sentence. I’m not going to explain it. I’m going to translate it.

At the policy level, engineering means Pigouvian taxes on known externalities. Indonesia’s cigarette excise, one of the most successful public-health interventions of the last two decades, has measurably reduced teenage smoking. High schoolers don’t need divine guidance. The cigarettes just need to be more expensive. Carbon tax follows the same logic for emissions. Sugar taxes follow the same logic in Indonesia. Not elegant. Not inspiring. But it works.

At the institutional level, engineering means coordination mechanisms that bite. Elinor Ostrom won the 2009 Nobel in Economics for documenting that commons, under certain conditions, can be managed without privatisation and without state control. The conditions are specific. Clear boundaries on who is in and who is out. Rules made by the participants themselves. Monitoring by or for the participants. Graduated sanctions for defectors. Cheap and fast conflict-resolution mechanisms. The lesson scales from fishing cooperatives to platform regulation.

At the individual level, engineering means exposing defection and shrinking its cover. The Machine is strongest when defection is invisible. The tobacco industry doesn’t collapse as long as smoking looks normal. Money politics doesn’t collapse as long as envelopes are handed out in private. Every time you, as an individual, make defection visible, make the normalisation uncomfortable, make cooperation socially defensible, you are pulling one vector against The Machine. A small one. But not zero.

None of this sounds epic. That’s a feature, not a bug. Epic solutions to multipolar traps, throughout history, almost always become tyrannies. Epic solutions require one actor strong enough to coerce every other player into cooperation at once. That actor, once given the power, rarely gives it back.

What works isn’t heroic. What works is boring. Tax goes up 5 percent. New disclosure regulation. Routine audits. Certification that makes defection expensive. Reward schemes that make cooperation more profitable.

Boring beats heroic. Repeatedly. Look at the data.

VII.

We talk about The Machine as if it’s external. It isn’t.

The Machine is the name we give to the configuration produced by the sum of our individual decisions. You, me, everyone reading this. Including the decision to stay quiet when we know something is wrong. Stay defecting when we know there’s another way. Keep picking the path of least friction because it’s the cheapest.

The Machine will not be defeated. Maybe it won’t be. Maybe it can’t be. But it can be re-engineered. Every incentive we change, however small, shifts an equilibrium that every other player now has to re-optimise around. That’s how multi-agent systems work. That’s also how we get out.

This isn’t optimism. It isn’t pessimism. It’s description.

For the first time, we have a technical language for describing this problem. Game theory. Mechanism design. Incentive theory. The generations before us fought The Machine with their eyes closed. We have no excuse to continue the tradition.

TB Simatupang will still be jammed tomorrow morning. That is a technical question with a technical answer. Whether we answer it, or post a status update about traffic, is our choice.

Most of us will choose the second. That is also The Machine.