Listen Get

Wishonia’s Wager

Abstract

There’s a non-zero possibility that hell exists, and that you and everyone you love is going to die and burn in it for eternity. As it is very hot in hell, this would be unfortunate. There’s also a non-zero possibility that biotechnology lets you feel very good for an indefinitely long period of time. Expected value is the chance of a thing multiplied by the size of it, and if you multiply infinity by any likelihood at all, even 0.0000001%, it’s still infinity. So the expected value of doing nothing is infinity bad, and the expected value of acting is infinitely good, even if there’s an extremely low probability that any of this is true. The cost of acting is finite: roughly one share of a company that makes missiles. This paper argues that when one outcome is infinitely terrible and the other is infinitely good, any finite action that shifts the odds from the first toward the second is rational, and that the cheapest such action available is redirecting the resources your governments waste being really good at killing the taxpayers who pay for them. (They currently spend 604 (95% CI: 453-888) times more on the military than on the clinical trials that would cure the diseases doing the killing.) It’s Pascal’s wager with the broken parts replaced: one hypothesis instead of a thousand gods, real evidence instead of none, an action that actually changes the outcome, and a stake of one share instead of your eternal soul.

Keywords

war-on-disease, 1-percent-treaty, medical-research, public-health, peace-dividend, decentralized-trials, dfda, dih, victory-bonds, health-economics, cost-benefit-analysis, clinical-trials, drug-development, regulatory-reform, military-spending, peace-economics, decentralized-governance, wishocracy, blockchain-governance, impact-investing

Okay. So there’s a non-zero possibility that hell exists, and that you and everyone you love is going to die and burn in it for eternity. And as it is very hot in hell, this would be unfortunate.

I want to do the arithmetic on this. It is unpleasant arithmetic, which is why it mostly does not get done. I’m an alien. It isn’t going to be me. So I can stay calm.

First, expected value

There’s this thing called expected value. You multiply the chance of something happening by the value of it happening, and you get the expected value.

So let’s say there’s a plot of dirt, and there’s a 10% chance that there’s $100,000 of gold underneath it. The value of that dirt would be 10% times $100,000, which is $10,000. That’s expected value. You already use it every time you buy insurance or a lottery ticket. You just don’t say it out loud.

Now point it at hell. The expected value of hell is infinity bad, because you have to keep being really hot for a really long time. Forever. And if you multiply infinity by any likelihood at all, even 0.0000001%, it’s still infinity. So it would be preferable to avoid this, if possible.

The hell you can’t rule out

You don’t even need the fire version. There’s an earthly one.

My own great-grandfather, in the last generation of us that still died of such things, had something like what you call Alzheimer’s, and he looked like he was in hell. All the time. He was miserable, and he was like this for very many years.

Here’s the part nobody will hold still for. There’s a thing called time dilation, whereby when you’re dreaming a few minutes can seem like a few hours. There’s a drug called salvia divinorum, and some people who’ve taken it said they felt like they were being crushed to death for millennia. So there’s no way to disprove the possibility that your brain gets messed up, near the end, in a manner that causes you great pain and confusion for perceptually an indefinite amount of time. A very, very, very, very long time. Maybe eternity.

So even if this is highly unlikely, even 0.00001%, the expected value of it is infinity terrible. Same math as the fire. The dirt just has something much worse than gold under it.

The good infinity

Okay. Now run it the other way.

There’s this thing called biotechnology. I don’t know if you’ve ever taken any drugs. I certainly haven’t. But I hear tell they can make you feel very good. Which tells me that a human, by manipulating the atoms and molecules in their own brain, can make themselves feel very good.

And the time dilation apparently runs both directions: through some mysterious biochemical action, the perception of time can be extended. And there are numerous living things that simply don’t die of age, like certain jellyfish (Turritopsis dohrnii just resets and starts over). So with sufficient resources, effort, and scientists, one could figure out how to produce a very nice, pleasant experience for an indefinitely long period of time.

And the expected value of that, given it can go on without limit, is infinitely good. Even if there’s an extremely low probability that any of this is true and you can pull it off. Even 0.00001%. Infinite goodness.

So these are your options

Two outcomes. Both infinity. One terrible, one wonderful, both pulling at you the whole time. Anything that shifts the odds from the terrible one toward the wonderful one is worth doing, at any cost short of infinite.

People who win the argument that two infinities are no bigger than one have a way of dying from the thing the argument was about.

The wonderful outcome runs on biotechnology, biotechnology runs on clinical trials, and your governments spend 604 (95% CI: 453-888) times more on the military than on the trials that would cure the diseases killing the taxpayers who pay for the military. So every solution in this book does one basic thing: it moves resources from being really good at killing people toward being good at curing them.

Here are some fun ways to do that. Some of them will work by themselves. Some are complementary and will end war and disease more quickly if you do several at the same time. You only need one to work. Pick whichever one sounds fun to you. (There are ten mechanisms on this list. If each has even a 10% independent probability of working, the chance that at least one works is 1 minus 0.9 to the tenth power, which is 65%. At 20% each, it is 89%. The question was never whether one of these works. The question is whether you will do any of them.) If you have a better idea than any of the ones listed below, please email [email protected] and I will add it to this list.

  • The 1% Treaty170 171 instructs governments that own enough nuclear weapons to end civilization 122 times over, and have cured Alzheimer’s zero times, to be one percent more rational.
  • The Loving Takeover is humanity buying a controlling stake in every missile company on Earth, for about $109 (95% CI: $107-$111) a head, and pointing its lobbyists the other way. It’s loving because everyone you buy out ends up richer and lives longer.
  • Incentive Alignment Bonds172 are war bonds run backwards: your grandparents funded WW2 at 3%, you fund disease eradication at 272%. Same structure, fewer Nazis.
  • The dFDA173,174 makes clinical trials 44.1x (95% CI: 12.8x-210x) cheaper by recording what actually happens to the real humans who try things. Consumer Reports, but for not dying.
  • The Earth Optimization Prize175 puts money in a pool and pays people to recruit voters for the 1% Treaty. If humanity gets healthier and richer, the pool pays out. If not, it was a donation to the species.
  • The Global Referendum puts “should we stop dying?” on a single ballot for the entire species at once. It is expected to poll well.
  • The Court of Humanity176 is a court where any human can sue any government, “you can’t sue us” is not a valid defense, and the 1% Treaty is the settlement offer.
  • Wishocracy177 lets billions of people decide whether the next public dollar buys another nuclear weapon or another clinical trial, with no lobbyist standing in the middle taking a cut.
  • The Optimitron178 is a machine that works out which policies actually work by checking the thousands of places that already tried them. Your species skipped this step and went straight to arguing.
  • Algorithmic Public Administration179 replaces whole bureaucracies with a few lines of code. The Department of Defense’s replacement requires zero. The IRS’s requires six.

There is also a shirt, a lawsuit, and the option of putting every leader on Earth in a single room and Clockwork-Orange-ing them until they have read all of the above, but I am told at least one of these may be illegal.

Why some of these are not probabilistic

The 10%-per-mechanism estimate above is deliberately conservative: the mechanism-by-mechanism estimates in the back run from 10% to 95%, and for several mechanisms the conservative number is wrong in a useful direction. Some of these are not bets. They are purchases.

Shares are for sale. Money buys shares. Enough shares give you board control. The board controls the lobbying budget. Given $873 billion, the Loving Takeover succeeds with probability 0.95 (95% CI: 0.8-0.99). The uncertainty is whether the money gets allocated, not whether the mechanism works once funded. The mechanism is arithmetic.

Clinical trials produce cures. This is not a hypothesis. It is the observed base rate of the last century of medicine. Given $27.2 billion per year pointed at a trial infrastructure that runs at 44.1x (95% CI: 12.8x-210x) the current cost, diseases get cured faster. The uncertainty is the timeline, not the outcome.

Lobbying works. The military industry has demonstrated this for eighty years, exposed to full public scrutiny, and nobody has stopped them. The Incentive Alignment Bonds point the same proven mechanism the other direction, with the same money. There is no reason it fails differently.

For each of these mechanisms the structure is the same: it succeeds if a resource condition is met, the condition can be met with a known, finite amount of money, and that amount is smaller than current global military spending. So the conclusion is not probabilistic but conditional: if the resources are allocated, then the mechanism works, mechanically, not hopefully. The probability lives entirely in the coordination problem, getting humans to spend the money on curing disease instead of on the capacity to cause it. Several of these would work independently, and all of them are cheaper than what your governments already spend on the problem they would solve.

The primary obstacle to ending war and disease is the belief that ending war and disease is hard. It is not. It is expensive, in the same way that buying a house is expensive: the money exists, the mechanism is known, and the only question is whether the entity that has the money decides to use the mechanism. That decision is the only hard part, and it is the part this book is about.

Where the arithmetic lands

Run the argument forward and it ends in one sentence. The good infinity runs on biotechnology; biotechnology runs on resources; and the resources are not missing, they are misallocated, parked in the things that manufacture the bad infinity instead of the good one: the weapons, the waste, the 604 (95% CI: 453-888)-to-one machine. So the rational act is not to find money. It is to move it: identify the waste and the harm in how your species currently spends, and point it at the thing that buys you out of hell and into the long pleasant stretch. That is the entire program. Everything else is one way of doing that one thing.

You do not have to pick the right one. Back any of the ten. The one with the best math is the Loving Takeover: the smallest downside, the largest upside, and the only one that pays you back, because you end up owning the missile companies instead of just defunding them.

The actual cost

The cost is not money. It is the most expensive thing you own, which is your attention.

Right now you could be watching young ladies twerk on TikTok, which I am told is enormously fun and asks nothing of you. Or watching large groups of men elaborately relocate spheres from one location to another at great cost, an activity your species takes more seriously than most of its diseases. Or watching very rich people on television, who are stealing your money in elaborate ways, pay other people to make you furious at a different group who are stealing far less of it, in far less elaborate ways, while everyone faces the wrong direction.

Instead, you have been handed a long (at least slightly hilarious?) book about how to avoid burning in hell forever in exchange for a possibly endless stretch of pleasantness.

In the short term this is the worse offer. In the long term it is the best trade ever made available to your species: a few evenings of not-twerking, weighed against infinity. The expected value is not close.

Why this isn’t Pascal’s wager

Three centuries ago a Frenchman named Pascal noticed the same shape: a small bet for an infinite prize. His version is famous, and broken. This is the same machine with the broken parts replaced.

Here is the bet he drew, for anyone who has not seen it:

You… …and God exists …and God does not
Believe Infinite reward You wasted some Sundays
Don’t believe Infinite punishment You slept in

Infinity beats Sundays, so he said believe. Here is the same shape with the bugs removed:

You… …and the infinities are real …and they are not
Act (a share, an afternoon) Your odds shift toward forever-flourishing and away from forever-suffering You hold a real asset that does not go to zero and buy some healthy years. A gain on its own.
Don’t act You leave two infinities on the table You keep the afternoon

Pascal’s bet had a cost cell: believe, then find there is no god, and the Sundays were for nothing. The Act row has none, because it is backed by assets and added years instead of faith. The four specific bugs:

Pascal’s wager breaks because Wishonia’s Wager doesn’t
You have to pick one god out of thousands, and might pick wrong One diagnosis, about ten non-competing fixes. Back them all. Only one has to land.
There’s no evidence for any of them Medicine works. Lifespans went up. Experience is already adjustable.
Believing doesn’t change the outcome Funding trials demonstrably produces treatments.
The stake is your eternal soul The stake is some evenings you were going to waste anyway.

The “pick one god” problem is worse than picking wrong. Christianity alone runs about 47,000 denominations180, many of which promise damnation for backing the wrong one, so Pascal’s bet does not even resolve to an action. Here the ten fixes do not compete, none of them damns you for backing another, and you can rank them by cost and probability and back them all. You are not choosing one god out of thousands. You are funding a hospital, and the hospital does not care what you believe.

Pascal asked you to believe something you can’t check, hoping to please something you can’t detect. This hands you a book and ten ways to win.

If you won’t believe in infinity

Fine. Delete the infinities.

Even the most boring, selfish version of these solutions pays off in plain dollars. Buy one share of a military contractor and it’s still an ordinary stock that doesn’t go to zero. The campaign it funds still has a real, documented chance of paying off. And every extra healthy year any of this buys you keeps the rest of your money worth something, because your money is worthless the day you die. The finite math is already positive. The infinity was only ever the ceiling on top of a floor that already clears.

So the bet works on everyone. Believe in the infinities and you act because the payoff is unlimited. Refuse them and the boring math says yes anyway. The only way to lose is to not play. The prize is the misallocation times the correction: a 604 (95% CI: 453-888)-to-one misallocation no one audits, times any nonzero correction, is still enormous, so no amount of acting produces a small result.

The wager

The numbers came first: $84.8 quadrillion (95% CI: $42.9 quadrillion-$172 quadrillion), the price of the lives being lost, against one of your 122 apocalypses. Numbers that size stop registering. Hence infinity.

Earth Optimization Services takes the proceeds from this manual, buys shares in the companies that build your missiles, and points their lobbyists the other way. That is the Loving Takeover. The manual is free; the shares are the wager.

So the wager is yours: an afternoon and a single share, set against the two outcomes that never end.

The useful hallucination

Every institution that runs your world is a fictional construct. Nations, corporations, money, laws, borders: none of them exist in physics. They exist because enough humans agreed to hallucinate them at the same time. Some of these shared hallucinations are very useful. Property rights, for instance, keep people from taking your things. Some of them are killing you. The one where you spend 604 (95% CI: 453-888) times more on the military than on curing your diseases is killing you.

Earth Optimization Services is just another collective hallucination. The only difference between it and the ones you already believe in is that this one, if enough of you agree to hallucinate it, buys the harmful hallucinations and turns them into helpful ones. The entire job is getting enough humans to hallucinate the right thing.

You have hallucinated worse.

Your company

Congratulations. You are now a Humanity Manager at Earth Optimization Services. Your title is President of Earth Optimization Services, and the correct way to address you is Your Excellency. You are entitled to a vote in how it is run, and to a better world every time it succeeds. You cannot buy that vote and you cannot sell it; it is yours because you live here. As a Humanity Manager, we would appreciate it if you would be nice and helpful to the other humans in the workplace, which is Earth.

Thank you.