Dawn(ehh) of the Second Day, ~48 hours remain
Alright fine.
How to Succeed in Life by Playing Chicken with the Universe
Or: A Brief Theoretical & Practical Guide to Outcome Pumps
Part 1: Underlying Theory
Suppose there's a magical object with the following properties:
- A binary utility function U(x) whose argument is the entire state of the world.
- An ability to globally manipulate probability, such that it could cause any sequence of events whose probability is higher than P.
- Omniscience.
This is a so-called "outcome pump". It works as follows:
- If U(world) = 1, it does nothing.
- If U(world) = 0, it:
- Looks at all possible timelines branching off from this temporal point;
- Ranks them based on a) the time T it takes until U = 1 again, and b) the prior probability of this timeline p;
- Removes all timelines where p<P;
- Chooses the timeline with the smallest T from the remaining timelines;
- Makes it actual.
To give an example, suppose you have a Cleaner outcome pump concerned with ensuring there is no trash in a particular radius. The moment someone litters, the outcome pump manipulates their thought process, nudges it onto a track where they'll feel guilty about littering, turn around, and clean up after themselves. If a tree falls down, perhaps Cleaner manipulates some corporation into cleaning it up; or, if there are no humans nearby, perhaps it picks a timeline where a lightning sets the tree on fire and a gale-force wind scatters its ashes.
Outcome pumps are as ridiculously powerful as they are dangerous. The crucial thing to note is that
they are not your friends; they are literal genies, they are paperclip-maximizers. If the quickest way to get rid of that tree from above is to cause a bus full of people to ram into it, throw it off a conveniently-situated cliff, and follow it into the abyss, Cleaner would absolutely make that happen — even if there was an infinitesimally-less-likely timeline where that tree is peacefully cut into pieces by a bunch of wandering lumberjacks.
But if you're sufficiently careful, the possibilities for exploitation are endless, boundless.
Suppose that you have a very strong will. You go to Cleaner, throw some trash on the ground, resolve to
only pick it back up if you find a valid proof/disproof of the Riemann hypothesis, and start furiously thinking about math. What next? That depends. Maybe there's another person nearby, and they walk up to your trash and clean it up. Maybe there's a
moose nearby, Cleaner causes it to see you as a threat, and it throws your trash off a cliff with a careless kick... in the process of throwing
you off that same cliff. Maybe you've overestimated your resolve, Cleaner causes you to think of nightmarish scenarios where it manipulates a moose into throwing you off a cliff, you chicken out and clean up yourself.
But if you've ensured that there would be no interference, that the most probable trashless timeline is the one you chose, then you arrive on the valid proof in a few hours/days, clean up, and go on to claim
all kinds of prizes.
How to exploit an outcome pump, more formally:
- Determine the outcome pump's utility function and "power" (i. e., value of P).
- Choose an Event A that you want to happen. Estimate its probability, P(A). Only proceed if P(A) > P.
- Choose a convenient Intervention X such that it minimizes the outcome pump's utility.
- Minimize chaos within the spatial location relevant to the outcome pump, until the probability P(!X) of a random event undoing Intervention X is less than P(A).
- Conduct Intervention X. Commit to undo it IFF Event A happens.
Why am I talking about it? Well, because we already have several outcome pumps on our hands. Their utility functions are to be in possession of their owners.
Part 2: Relevance
"Wait! Your Grace!" a voice called from behind.
Or not. She recognized the boyish voice as belonging to Cove. She hung her head, turning around. "Yes? What is it?"
"Oh. Sorry to bother you," he said, catching up to her. "Nothing urgent. It's just… well… I think these belong to you?" He produced a small box, opening the lid.
Inside were her lost bracers.
"Cove!" Zelda exclaimed, snatching them from the box. There was no mistaking it; she recognized their draw instantly. These were her bracers. She traced her fingers along the ridges of the golden emblems they bore. "How did… what? I don't… how did you come by these? Has Vesyrn been captured?"
Cove shook his head, putting the box away. "No, Your Grace. I'm sorry, but I don't know anything except for what I was told. They were handed to me by one of the other Sheikah sent after the roving Blin when he checked in. Malg, I think his name was? He said they were found on the body of a moblin that was killed in an attempted raid somewhere just to the north." He scratched his head. "What a crazy coincidence, huh?"
Zelda laughed uneasily. "Yeah. What a crazy coincidence."
But, no. Something felt… wrong.
With a growing panic, Link felt at his chest. The necklace! How had he lost it? Was it—
He dashed backstage and slid under the curtains, searching for the temple prop. It was hard to make out in the dim light, though he found it off to the side and scoured over every inch of it in the hope the necklace had snagged on something during the scuffle.
No such luck.
He darted around hoping to find a stagehand — or anyone else for that matter — but saw no one. Just as he was about to head back to the hall, he heard a door slam shut.
He took off in the direction of the sound. Near the back of the staging area were two service entrances on either side. He opened the door on the left and dashed through. Just ahead was a dusty storage area, and to the right a stone wall. To his left—
Midway down the hall was a spiral staircase. And snagged on the railing? His necklace.
"Hello, Mr. Hero," Seff said.
Link backed away, regarding the man with suspicion.
"Well?" Seff asked. "Oh, come now. I thought we were friends."
"Did.. did you—"
"Yes, I stole your necklace. Well. To be more precise, I had one of the actors do it," he shrugged, smiling. "Same thing, really. But don't worry. It seems I couldn't take it from you if I wanted to. It would find its way back to you one way or another."
I'm not familiar with Cosmere, but it's been suggested on Discord that "Seff" is one
Hoid, and he would presumably know what he is talking about. The return of Zelda's bracers support his account — they've caused that long string of coincidences with the explicit purpose of returning to Zelda. (The alternative explanation here is that Malg lied to Cove, and they were actually returned on Vesnyr's order, but I think it's not overwhelmingly more probable.)
That leaves some questions, of course.
Part 3: Testing Procedures
- How do they determine ownership? Raolin was able to transfer his fragment of Courage to Link, so literally transferring ownership presumably works. Stealing doesn't work. Would killing the owner work? Surely they're not powerful enough to spontaneously resurrect the owner...
- How do they determine possession?
- Physical proximity? Test it by having Zelda give the bracers to a trusted group of the Sheikah, who will then move to a different city for a week. On day 2, they're to roll 4d6; if all four die roll 6, they're to return the bracers to Zelda immediately; otherwise, stay away for a week even in the case of a dire emergency.
- The owner must be able to access the fragment any time they want? Do the above, but have the Sheikah move somewhere where Zelda (and her people) wouldn't be able to find them.
- Something more complicated? Arrange for them to be stolen by some thief, ideally without Zelda's (or Link's) knowledge. Have Zelda steal Link's fragment, for example, or vice versa, and roll 4d6 on whether to return it within a week or immediately.
- (The worst-case scenario is if there's some intelligent entity watching events and determining whether the fragments are still in their owners' possession on a case-by-case basis, instead of based on some pre-determined rule. If so, it'll just see that they aren't "really" lost if we plan to return them in a week, and it would do nothing.)
- How powerful are they? Hard to say without additional testing. Apparently Link's fragment managed to befuddle an actor, but that doesn't necessarily make it more powerful than P=1/20. Zelda's bracers, however, caused a Sheikah to lose them, and then to be found again by someone who knew whom they belonged to. 1/500? 1/1000?
- Testing could be done the same way: with die. Determine how they determine possession, then just keep repeating the experiments from the section above, but with progressively more die, until the outcome pump stops bothering manipulating their rolls and has the fragments returned some other way/gives up and waits the week. 4d6 is one-in-a-thousand, 5d6 is one-in-eight-thousand, 6d6 is one-in-fifty-thousand, and so on.
These testing procedures
should be safe, since they offer harmless exit conditions (outcomes of die rolls)
and short time limits.
So, we've conducted our experiments, determined whether fragments of Triforce are outcome pumps, how they define possession, and how powerful they are. What next?
Part 4: Practical Applications
Enjoy our limited PtV.
At the foundation of everything here lies the same trick: have the fragment "taken" by some group with a strong will, who precommit to return it if and only if a certain condition is met.
Note that conditions should be verified prior to the fragment's return (otherwise the outcome pump will just fool us into believing they're achieved). Additionally, it is probably not viable to use them for long-term plans (it's unlikely we'll be able to keep
P(!X) low for months or years on end) or open-ended wishes ("Zelda thinks the day was maximally productive" just puts Zelda in a good mood, but likely doesn't make the day
actually productive).
Possible conditions:
- Our causalities in a given battle are low.
- The Sheikah successfully assassinate a Warlord.
- Wymar makes a correct guess about the Blin's troop movements.
- Fi successfully researches an Old Magic.
- Zelda's latest invention is a success.
- Zelda makes a good impression on the Council; they respect her more and are more cooperative.
- Taetus makes a blunder that gives Zelda a leverage on him.
- We find Seff.
- (Tentative) Fi thinks that today was an unusually successful day.
- It's open-ended, but Fi may be an exception. If she's inhumanly rational and possesses quantitative superintelligence, she may be "strong-willed"/"infallible" enough that it's actually more likely for a day to be unusually successful than for her judgement to be compromised.
I'm sure there's a lot of more interesting conditions I could've included here, but this post is already too long. Post your suggestions!
tl;dr: We could use the Triforce fragments' return-to-owner probability-manipulating power as a more flexible, precise, and generally better version of the
Luck spell.