RNG Training Experiment Idea.

Back before I was legally blind I used to train trying to manipulate and read data from video games. I figured that, since it’s all just data in the computer, and that it was known to be possible to manipulate random number generators, that it should work. I started with Assassins Creed 2 where I tried to feel for secret items and follow my intuition to see if I could find them. It was surprisingly effective. Later on I would try to manipulate arena betting in Elderscrolls IV: Oblivion by trying to control the AI behaviour of one of the fighters and make them win. To which I discovered that, probably, Bethesda programmed the arena to make one combatant invincible to guarantee it wins and thus reduce the arena to a pure RNG event. Because when you have one guy with a claymore beating the shit out of another guy for 5 minutes who’s only got a dagger and hasn’t landed a hit yet, it’s pretty suspicious when dagger guy somehow wins shortly after I get tired of controlling claymore guy.


I also would try to detect and later manipulate critical hits in Pokemon. When Dreamsoap and I moved in with each other she was playing a game called Dungeon Fighter Online which had an event running at the time where special coins would sometimes drop in dungeons and if you collected enough of them you could get unique rewards for as long as the event ran. I don’t know who came up with the idea but we decided we should try manipulationng the drop rate with magic. According to her the rates went from 1-2 coins er dungeon to a few coins per room, a dramatic increase. What was most interesting to me was that during it I was able to accurately tell whether she was focusing on it or not and I would remind her to keep her magic going. So I was sensing her energy. Since then she's had her own successes in other games detecting invisible traps, manipulating RNG boxes and the like.


Over time this got me thinking about how you’d put together a better experiment on manipulating RNG. One of the biggest problems being that it gets really boring really fast. Well, a lot of these games had RNG elements but they were hidden behind layers of other gameplay and visuals. I figure if you wanted to design a good one you’d do that too. I’ve been thinking about how you’d design one for Psiquest that could be run continuously.


A simple game with opt in data collection that could easily do all the math to show the statistics so that no one could complain about accuracy. Game design would have to make RNG a focal point of the game. I figure all calculations would have to be done server side because it’s guaranteed that someone would try to hack the game if that’s handled locally. Thus simple turn based games would be best to not overload a server. I think the games would be best portrayed as being about training magic rather than testing to see if it even exists and the games should reflect that. We’re so far past the point of asking if it’s real or not it’s kinda sad so let’s not make it about that. Such as a Pokemon style battle system where players could only do damage with critical hits.


In order to make it most accessible it would probably have to be browser based which probably means coding it in Javascript. I don’t know Javascript, I do know D, more or less, which has a decent webframework called Vibe and a game engine called Dash but I haven’t dabbled in any of that yet so I don’t know if it could do the same thing. Not that I’m promising to do this or anything although I have been working on my own projects that, if finished, could be used to implement this sort of thing.


In any case what do people think about this overall idea and would anyone want to try working on it?

[Laird]

In any case what do people think about this overall idea and would anyone want to try working on it?

Love the idea and in theory would want to help out, not sure about in practice though. Am passable at Javascript. An interesting tool/framework (I'm not really sure how to describe it) that I came across not long ago is Electron, which (as best I understand) allows you to create desktop apps out of web apps - could be useful for your project so folks could choose between browser or standalone app.

My questions would centre around the RNG. I know Max has discussed this area in the past. At any rate, I think it is necessary to distinguish between pseudo and true generators. These are not equivalent, the behaviour and our possibilities for interacting with them are very different.

My questions would centre around the RNG. I know Max has discussed this area in the past. At any rate, I think it is necessary to distinguish between pseudo and true generators. These are not equivalent, the behaviour and our possibilities for interacting with them are very different.

Realistically this is far more about pk than anything else. My theories always centered on the idea that the data was bits in the machine ad that certain combinations equal certain values. How they got to that state didn't matter. Pretty sure Oblivion and DFO weren't using true random number generators. Which is a misnomer by the way. Mathematically as best as anyone can tell randomness doesn't actually exist and TRNG's use atmospheric noise and other environmental variance to generate their numbers. So it's just more complex, not random.

I don't plan on getting to this project in the near future but I'd want to plan it out in case/for when I do.

Like having one game that is just opening a box that's locked by a boolean with nothing in the code that will change it. All you do is push a button to tell the server to check the value, if it's true then the box opens and you get a bunch of stuff. Which means the only way to open it is telekinetically changing the bits to be true... or hacking. Guaranteed people will try hacking all of this to prove how psi isn't real and stuff. But that's why this needs to be about training not proof. But also effects that are of a practical strength. Not just statistical variance.

People may not like JREF much but they had the right idea with the whole "if it's so real then prove it and we'll pay you" attitude. If you can't do something practical, don't bother. Train up first.

Realistically this is far more about pk than anything else.

Sure. But a pseudo-RNG is deterministic. One can only predict, but not change the values generated. If your approach is dependent on being able to predict a future value, then ok.

There are other ways to generate some sort of randomness, but it may not be satisfactory. For example one could perform some manipulation of user input, such as key or mouse input, and get the user to supply the non-deterministic portion.

I don't know whether this is any help, but Richard Broughton did some work using a computer game interface for psi experiments. His profile at the Parapsychological Association includes references to a couple of papers about it:
Broughton, R. S. and Perlstrom, J. R. PK experiments with a competitive computer game. Journal of Parapsychology, 1986, 50, 193-211.
Broughton, R. S. and Perlstrom, J. R. PK in a competitive computer game: A replication. Journal of Parapsychology, 1992, 56, 292-305.

I agree with Typoz that it's important to think about what kind of random numbers to use - pseudo (meaning the numbers are actually deterministic, though possibly depending on user input, such as the time at which a button is pressed) or true (meaning there is genuinely unpredictable physical input).

Sure. But a pseudo-RNG is deterministic. One can only predict, but not change the values generated. If your approach is dependent on being able to predict a future value, then ok.

I've done several poltergeists where I manipulated electronics, starting with a clock radio. I figured out that, since it used a toggle, all I had to do was close the circuit to turn on the radio, it didn't matter how the circuit got closed just that it got closed. I also messed around with a lottery machine at a place my girlfriend worked at the time and bypassed some of it's code to run the print ticket routine manually by stimulating the circuit directly. Under normal circumstances it's not possible to print a ticket unless all the other screens are clicked through first. Doing something that should be otherwise physically, or at least programmically,  impossible.

It's all just a bunch of transistors and capacitors and such, that's what the value is, change the stored charge and you change the value. The job of the trainees (which would include me)  is to figure out how to do that in a reliable, repeatable, psychic way. If they throw up their hands and say it can't be done then good riddance. Again, this is about training, not proof. It is expected that people will suck at it at first.

There are other ways to generate some sort of randomness, but it may not be satisfactory. For example one could perform some manipulation of user input, such as key or mouse input, and get the user to supply the non-deterministic portion.

I would intend to have pure prediction tasks as well and maybe this would be a good way to do them so I'll keep it in mind.

I don't know whether this is any help, but Richard Broughton did some work using a computer game interface for psi experiments. His profile at the Parapsychological Association includes references to a couple of papers about it:
Broughton, R. S. and Perlstrom, J. R. PK experiments with a competitive computer game. Journal of Parapsychology, 1986, 50, 193-211.
Broughton, R. S. and Perlstrom, J. R. PK in a competitive computer game: A replication. Journal of Parapsychology, 1992, 56, 292-305.

I agree with Typoz that it's important to think about what kind of random numbers to use - pseudo (meaning the numbers are actually deterministic, though possibly depending on user input, such as the time at which a button is pressed) or true (meaning there is genuinely unpredictable physical input).

I'm glad someone's already done experiments like this. And for pure predictive tests I agree that the type of randomness would be important. I guess in my mind it was never purely about that but I didn't know it or something.

I'm glad someone's already done experiments like this. And for pure predictive tests I agree that the type of randomness would be important. I guess in my mind it was never purely about that but I didn't know it or something.

I think usually the concern with pseudo-random numbers would be that psychokinesis can't change them because they're deterministic, and if they're used for precognition studies then in principle it could alternatively be clairvoyance, for the same reason.

But if you're thinking of psychokinesis that can actually modify the normal working of a device, even when the numbers should be predetermined, then of course that's different - and would be unprecedented if it could be done repeatably under controlled conditions.

I think usually the concern with pseudo-random numbers would be that psychokinesis can't change them because they're deterministic, and if they're used for precognition studies then in principle it could alternatively be clairvoyance, for the same reason.

But if you're thinking of psychokinesis that can actually modify the normal working of a device, even when the numbers should be predetermined, then of course that's different - and would be unprecedented if it could be done repeatably under controlled conditions.

I think that's where my confusion came from because to me if it isn't directly affecting teh physical world then there's no point to it so that's all I've trained for. I know it's possible to condense energy which makes it gain inertia which makes it possible to act kinetically because I've done it. I know from that, also experiments, that this can be used to affect electronics too. It's the repeatability that I have trouble with.

I really want more people to be training this stuff and pushing it forward, if people like Sheldrake and Radin are right then just doing that on it's own will make it easier to do for everyone because there will be less mental interference. If more minds focused on the same thing seem to get better results, then more people training should make training easier. That's one of the ideas here.