Sharing MMI program demo/code

Sharing the code for a program that I’ve recently written. This program uses Python to Med100kx8 random number generator to implement Scott’s latest Random Walk Bias Amplification (RWBA) method with advanced processing methods for increased effect size in anomalous cognition details.

The program output depicts a bar-chart, a spinning cube, and an audio tone that all react to the output of the random number generator. The goal is to use mental influence exerted on the random number generator to cause the program to respond in the desired manner - make the cube spin faster, make the bar chart fill more, make the audio tone increase in frequency.

The program supports various modes. You may select one-tailed, in which a single direction of influence (the RNG produces either more 0s or more 1s) is attempted. More interestingly, two-tailed is also supported, in which the user attempts to switch between attempting to influencing in both directions, thus turning the provided indicator ON and/or OFF based on hitting upper or lower probability targets. More details descriptions are included in the Github repos, links below.

All runs are continuously executed, which means numbers are constantly pulled from the RNG without the user needing to manually trigger. The program can be configured to run for a given number of seconds or to continue indefinitely until the user signals to stop.

The basic version of this program does not store any data to a database. I expect this will be the more popular implementation as it’s simpler to configure. There is a more advanced version that accumulates a large variety of data throughout the trial, including solar data, local environment conditions, subjective user information, EEG, GSR, etc, which would allow for much more thorough result analysis. I’m linking that version below also.

A short demo of a two-tailed (bi-directional influence to turn ON/OFF) program run is included below.

Github repo - no database (simplified install)

Github repo - with database

Both versions are fully compatible with OSX and Windows. I know a couple of you have gotten this running already, which is great. I’m happy to answer questions if there are issues.

1 Like

Nice continued work, I’ll try and spend some time delving into it!

Yeah, it’s great that you are writing an open-source code that is being maintained. This is my goto right now for test driving the Med100kx8.

I have been playing with it and started digging into the code. I know the README covers some of this but it might be helpful to have a description of the outputs e.g.

Last window z-value: -0.15951573914068204

Last window p-value: 0.5633687320184062

Last window surprisal value: 0.8278486011315803

Last window was hit?: No

Running overall window bound tracker: 3

Window group z-value: -0.27002940805937264

Window group p-value: 0.828125

Window group surprisal value: 0.27207954543680074

Thank you! I needed something like this

I spent about 2 hours trying to figure out how to get ftdi to behave, probably some silly mistake i am making, will update when i figure it out, just updating in case anyone else runs into this error

Are you using Windows? Did you follow the instructions here?

If you do that and still no luck please send to me via PM your code and error you get. I’ll try to help. I had a ton of issues with getting the FTDI drivers to work initially too, which is why in the end I opted to interact with the device directly and not use the drivers.

Weird, I had tried that a couple times but kept getting the same error, for whatever reason under your instructions it worked.

Last window z-value: 0.33378919188273815
Last window p-value: 0.7385387665579441
Last window surprisal value: 0.43725444356146975
Last window was hit?: No
Running overall window bound tracker: -4
Window group z-value: -0.3717515420454472
Window group p-value: 0.7100779556172829
Window group surprisal value: 0.49395067589157693

Total time taken: 59.21 seconds
Total trials executed: 294
Average time per trial: 201.40 milliseconds

Thank you so much for this! I have built similar things but am not much of a programmer so its very nice having this tool available.

Awesome, glad to hear! I’m happy to help if there’s anything you need.

I will indeed, I got the Geiger one running on windows, this is very commendable and useful, I wonder how hard it would be to run on scotts PCQNG or similar

Congratulations! You have influenced the RNG.

what a fun feeling. hook this up to a leaderboard and you have a popular game on your hands.

Hey Joshua,

Great to hear you are enjoying this. I also have been using it for quite a few months and am hoping to run some simple analyses once I get enough data (and time to analyze it :slight_smile: ). At the moment, it feels more like a training ground for me to build my PK skills. I play it for a few minutes every day. Are you able to get consistent results when you play?

I’ll let you know, I think it is amazing you added the database, many interesting things could come from that. This is the sort of thing we would chat dreaming about and here you are doing the work! Well done.

After trying to learn a bunch of Bayesian statistics etc that Scott talks about, I realized it didn’t really matter if I was able to get the math right if I didn’t have the skill to influence the RNG. So, I moved my efforts to creating simple games (I like to call them games because if you tell someone its a PK skill trainer sometimes they will look at you crazy and you can get away with a whole lot more calling something a game. Plus there is some proof that games were created because of divination) like this one, except my coding skills are not as good as yours so I can’t guarantee any of it was working and my scientific method was not good because I was just kind of doing it for the hell of it) which is why I am grateful I am able to look at your code to kind of compare notes.

I tried for a bit earlier and seemed to come up with mostly hits, but I am very interested in what the historic data will look like. So far results with your software are modestly successful, with a decent hit rate, however the z score and p value may suggest that the results are in the realm of chancel I’ll update after I play consistently for a while.

I will say I have seen with my own eyes a few people who played games like this with consistency, and ended up being able to influence it in a very obvious way.

This is very interesting, is there any reason you want to train your MMI skills? And do you have any other related projects you are working on? I am impressed by how well organized and how much thought you put into this.

I appreciate you saying so. I would say I want to train MMI skills because I am personally intrigued by the experience, and it fits with my experiential approach to “spirituality”, or getting to know what is beyond our 5 senses.

In line with this, I work as the Principal Data Scientist at HeartMath, with a main focus on the Global Consciousness Project 2.0, a large global network of RNG’s in which we are essentially looking for MMI signals based on the evidence of the original Global Consciousness Project from Dr. Roger Nelson at PEAR Labs in Princeton.

ah awesome, i am familiar, nice to meet you and welcome and thanks again i have been training with your software

Small clarification @JoshuaLengfelder - code is mine. Glad you’re getting use from it.

1 Like

Woops, well thank YOU it has been very helpful

Thanks for this - I have been playing around with it, however, I am wondering how to tell if I’m successfully influencing the results.

From Scott’s paper, it seems that the hit rate / surprisal value is the number to pay attention to?

I wonder if you could break down these results for someone who isn’t too versed in probability haha.

Last window z-value: 1.4847085110790545
Last window p-value: 0.9311894402287253
Last window surprisal value: 0.10285339683879327
Last window was hit?: No
Running overall window bound tracker: 5
Window group z-value: 1.1535206473472759
Window group p-value: 0.9453125
Window group surprisal value: 0.08113676272540553

I would also be confused by this if I had not attempted to make one myself before, and I have trouble getting results out of the database because I’ve never used it. when i asked @fluidfcs1 had no plans on changing the code. I dont remember seeing a license, would you be willing to liberally license it so that it can be expanded on? Reading your code really helped me, thank you for making it open source.

Hi Joshua,
To be clear, @fluidfcs1 did a good job of implementing one of my advanced processing methods, which involved a lot of effort, but he does not own either the method or the code. One of the conditions for supplying an MMI generator and carefully explaining most steps in its implementation is that the results will be shared on the forum. The only exception is if someone wants to use it, directly or indirectly, to make money. Then we could discuss a license if necessary.

Any specific proposal for a more expanded application, please private mail me. Otherwise, ask specific, detailed questions on this forum. I am super busy with the new tech, but I will find some time to respond.

Hi Jon,

Some details with the user interface were decided by Dan (@fluidfcs1) so I am not totally familiar with them. The meaning of the statistics you provided depends mostly on whether you were intending to produce more ones or more zeros.

If you were intending higher numbers, the results suggest a negative outcome. z-score and p-value are transformable either way, so one is as good as the other. p-vlues should be less than 0.5 to suggest successful results, but should be 0.05 to indicate statistically significant results. Surprisal value is an indicator of success, the higher the better, but if it is less than 1.0, there was no noticeable effect. If these results were for producing more zeros when the tester was set to measure increased ones, the statistics would be all wrong. The probability of the test results, in surprisal value, is 2^-SV: SV of 5 means probability of 2^-5 = 1/32 = 0.03125 (which would be statistically significant).

1 Like

Ahh that makes a lot of sense. So contributions can be made as long as the results are shared back to the forum. The only thing I would personally change is how the results are reported, if I get around to it then I will be sure to loop back to the forum.