Once the Predictor has made his Prediction, and Box B either contains $0 or $1mm, my choice looks like:
Two Boxes gets me the max, and One Box leaves $1k extra on the table. As a two-boxer, my choice is independent of the Predictor’s actions, and I’ve made $1k extra thank you very much.
And how much have I made in total?
As a one-boxer, I forego the $1k extra from the second box by trying to raise p 0.1%. I believe that I can raise p > 0.1% by one-boxing.
“You know that this being has often correctly predicted your choices in the past (and has never, so far as you know, made an incorrect prediction about your choices)”
I’ve yet to meet a Predictor like this. I’ve yet to meet someone who has met a Predictor like this. Where does the Predictor live? “Suppose a being in whose power to predict your choices you have enormous confidence.” To suppose the being as stated, the p increase from one-boxing should be ~100% > 0.1% (to one-boxers).
To suppose any being that you or any other human has ever encountered in real life, the p increase from one-boxing should certainly be < 100%, maybe ~0%? which is < 0.1% (to two-boxers).
Let me modernize Newcomb’s (rather metaphysical) Paradox (1969) as Quinn’s Question (2022).
One month ago, I began inviting people to my eccentric compound, selected at random from around the globe. Most declined the bizarre invitation (sent in the invitee’s postal code’s dominant language), but after sending out a few thousand, eventually I had met 1,000 unique individuals who had agreed to fly out to our unspecified location. I spoke with each of them for five minutes, recorded the conversation, and then had each of them fill out a questionnaire asking eccentric questions like: Given the mutually exclusive choice between saving 100 newborn babies, or ten philosophy professors that are all trapped in a burning building, who would you choose? What do you expect your income to be next year? At what age do you expect to die? And so on. After the five minute interview and self-questionnaire, my team of researchers took blood, hair, and urine samples, to conduct various biologic tests. The guest was then allowed to do as they please on the grounds the rest of the day whether it be horseback riding, skeet shooting, whiskey tasting, watching Stalker, anything they like. After a nice sleep, the guest woke up to an appointment time slipped under their door on a sheet of paper. When the time came, each guest came to the Decision Room and was presented with the Game: Box A contains their net worth (verified by my accounting team), Box B contains 20x their net worth, or 0.
The choices were made, the boxes were opened, the money was wired to the proper account and routing number (in a few instances BTC was sent), and the guests were sent home with a small gift package.
My team and I have spent the last month re-devising our strategy for filling Box B. Our only goal is increasing predictive accuracy as, luckily, the money is effectively unlimited (and we don’t invite those with net worth’s > $10mm). I’d now like to invite you to play the game.
Are you the kind of person that my machine can tell will defect? Exactly how confident are you that you are one of the people who can trick my algorithm? Which box(es) are you taking on Decision Day?