Summary of my efforts, questions on where to go

Hi folks, I’m new here. I’m an amateur researcher and recently stood up a non-profit aimed at scientific exploration of consciousness. One of the areas that’s interesting to me is psi, or MMI. I’m broadly familiar with research in these areas, and like the summary here. I’m particularly motivated by experiments performed by Helmut Schmidt and also Dean Radin.

I’ve done a number of experiments to try to get even moderately consistent results with using focused intention to impact physical reality in measurable ways, primarily with emphasis on influencing random number generators in particular directions. I’ve so far iterated through a few different types of RNGs. First, I tried using once that produces random numbers on-demand in real-time by measuring the quantum fluctuations of the vacuum (more details). Not getting the results I hoped for, and not seeing what appeared to be sufficient randomness, I moved on to a hardware-based TrueRNG3, which uses the Avalanche effect of a diode to generate the raw random bits. At that point still not getting the results I wanted, I opted to make an investment in using the same type of random number generator used by Helmut Schmidt in his many successful experiments over the decades - a Geiger counter connected to a computer that was constantly scrolling through a list of numbers. Whenever the Geiger counter detected particle decay, the computer would stop and select whichever number it happened to be on at that time. A limitation of this approach is that particle decay from background radiation is usually only detected about once a minute, which means one random number per minute, but by placing a few chunks of radioactive uranium ore directly next to the RNG’s Geiger counter I increase this to a number every couple of seconds.

My experiments are performed using Python code I’ve written. It basically uses the Binomial Test to determine the probability (p) and gives real-time feedback of the probability of randomness, both via a constantly-updated line-chart and via an audio tone that increases in frequency as the p value decreases. We aim to get a p-value less than or equal to 0.05 with some consistency. The code also has the ability to generate pre-run sets of random numbers whose outcomes are respectively both observed and unobserved, and then both sets are intermixed with random numbers generated in real-time while being presented to the influencer. Though the influencer cannot distinguish between the various sets during the experiment, their influence on each of the different sets (pre-run observed, pre-run unobserved, real-time generated) are measured distinctly for post-experiment analysis. The hypothesis here builds on existing experiments (example) suggesting that unobserved results should be as influenceable as results generated in real-time.

Mostly I am performing these experiments myself, meaning using my own intention, but do have a woman that regularly participates as well. Frankly probably neither of us have actually invested the amount of time and effort that it may take to build these abilities sufficiently.

Regardless, in the end, I have not been able to achieve consistently positive results. Occasionally I get a very positive outcome, but not with the consistency I’d need to show that it wasn’t due to random chance.

I’ve spent some time reading through Scott’s various papers, patents, and the content on this forum. I’m excited about efforts to dramatically enhance the effect of MMI through new software and/or hardware techniques. As far as I can tell though, and please correct me if I’m wrong, despite a lot of effort and approaches that seem to have some tentative potential, there haven’t been significant and consistent successes. Do I have that right?

Basically I’m really interested in finding ways to get more successful and consistent experiment outcomes. If there are algorithms or different types of RNGs or approaches I can be trying here I would love to do so. (side note, I’ve tried installing the METrainer and QNG360 but am getting errors in Windows - perhaps related to the fact that I am using Parellels to run a Windows 11 VM on a Macbook Pro with an M2 chip - but would love to get some assistance on that. I get errors about DLLs failing to load when installing QNG360, but it installs. Despite that when I try to run either of the .exes, even in various compatibility modes, a window pops up and immediately goes away).

One idea - I have an Emotiv Epoch on hand and am aware of research that has shown a correlation between various brainwave states and success with MMI (psi, pk, etc). (examples: 1, 2, 3, 4). Perhaps constraining analyzed results to those correlated with one of these brainwave states helps reduce noise. But, I have enough experience with EEGs in the past to know how difficult it is to get usable data from them due to so much fluctuation across different frequency bands in different parts of the brain, and many factors influencing those fluctuations.

There seem to be a variety of factors impacting PK (MMI) ability. It sounds like it may even be useful to build a Faraday cage to shield factors such as geomagnetic activity that seem to mute any psi/pk/telekinetic/MMI ability.

I’ve even contemplated using Kozyrev Mirrors in conjunction with these experiments, or Holotropic Breathwork.

There also may be some potential in trying to influence biological systems rather than non-biological (1), as some evidence suggests greater ease of influence in that domain.

I guess I’m just looking for some guidance and/or inspiration on how to proceed with this stuff. I’d love to start seeing some more promising results. Lastly, as an aside, I do have another area of active research, which relates to quantum physics and designing and executing an experiment meant to test and ideally validate the Von Neumann–Wigner interpretation that consciousness causes collapse of the quantum wave function. I’m excited and tentatively optimistic there, but recognize it’s out of scope for this forum.

Hi James, welcome. Always glad to see people actively working in the field of anomalous cognition and effects (ACE), also called MMI.

I, and a number of associates have gotten very significant results for years using specially designed high-speed generators and bias amplification algorithms (p down to 0.00001). I used to have a full setup available online with multiple testing and training applications connected to these generators, but as the Internet changed over the years, this very complex system became non-operational and I just couldn’t maintain it.

I can lend you a MED100Kx8 generator for active development use, easily done if you are in the US, but international mail still works very slowly. Using this and the METrainer, most people with a little practice can usually get results equivalent to what Schmidt got – effect size of 2-4%. You can message me directly through this forum or the contact page on my website,, for what support is available. You can also download the software-enabled PCQNG TRNG at This program uses hardware present in Intel-type processors. Anyway, if you can get it to run on your Mac, that would be a start. If it works, I can send you a license.

You would find the METraniner very useful. Did you follow the installation instructions exactly? Be sure to manually install the two VC redistributable files, or the program will not work. You don’t need to separately install the QNG interface software for the Trainer, but you will if you use a hardware generator to provide raw random numbers. You can’t test the hardware interface software unless you have hardware attached. I have not personally tested the trainer on Windows 11. I’m not sure about running it on a virtual machine on Macbook, especially Win 11. The METrainer was designed to run on Windows. It includes a special version of the PCQNG, as well as its own pseudorandom generator and an interface for using MED100Kxx generators. These sources are user-selectable within the program. All hardware generators will run on Windows7-10, Linux and Mac (the Linux source code would have to be compiled for your version of Mac OS – not tested beyond OS X)

The other generators (or entropy sources) you mention will not give notable results without bias amplification algorithms.

By the way, you may assume nothing is beyond the scope of this forum if it relates to mind, consciousness, anomalous cognition and effects or MMI. Schmidt assumed that consciousness collapsed the wave function and he derived a set of equations to illustrate the effect. Feel free to post any comment or question.

Hey Scott, thanks so much for your reply and support. That p value sounds incredibly promising, I’d be thrilled to be lent an MED100Kx8 to experiment with - thanks for that offer. I’m in the states and will send you my address via PM.

I’m sure I installed the things correctly in my Windows 11 environment. I did notice a note on the Comspire site that PCQNG will not function correctly in a Virtual Machine environment. QNGmeter is the only thing I could get to run. I’m sure I could get this stuff working somehow. I’d like to avoid having to compile code for different OSX versions but running the Linux drivers in an Ubuntu VM while using the hardware RNG could be a good path.

I’m curious, when you reference a little practice, how long does that typically take and using what techniques to practice?

The installation for the RNG and the METrainer are separate. If you can run the QNGmeter, that indicates the RNG interface may have installed correctly. But, as I said, you cannot test that installation without hardware connected to your computer. The METrainer is self-contained unless you also have and want to use a separate hardware generator connected to the computer. Remember, the advice about Linux or OSX only applies to the RNG interface, not the Trainer. If you had RNG hardware in hand (or the MED100Kx8 generator), your RNG interface might already be working. I suggest you may try the PCQNG just to see what happens.

General instructions for the METrainer are at: Some people get impressive results from the first attempt, but like any sport or “game” of skill, most people take many sessions of practice to improve – taking up to weeks. There is no point in spending more than about 10-15 minutes at a sitting, because it may get boring or too tiring. It actually takes little physical energy to maintain the required focus, but many people try to force the thing to work. This doesn’t actually work, but it results in becoming tired more quickly.

@fluidfcs1 Welcome to the forum! I’m Simon and together with Scott and the other members host this forum.

May I ask how you found out about this forum? As the forum admin I’m always keen to know how new members came across us.

I love the fact you’re trying to use Emotiv hardware in MMI experiments. I got myself an early Emotiv device back in 2010~2011, many years before I even discovered MMI. My idea back than was to try and use brainwaves to control an AR environment (alas, nothing came of that short-lived endeavour).

If you want to plugin to some of the MED devices provided by Scott, I run what we call the “MED farm”. It’s basically a few of his devices plugged into a server I run from my house. Details are in this thread: MED Farm (API server to provide generator data)
But I have to say due to the under-the-tv-old-pc-hardware and kids-always-unplugging-things, it’s not online all the time. Although I do try to make sure it’s back up as soon as possible.
They’re physically located in Singapore, which is where I live.

Looking forward to discussing more with you on this forum, and once again welcome.

hey Simon, pleasure meeting you and thanks for the reply. Regarding how I found the forum, I subscribe to the Paranormal subreddit and saw this post. It mentioned Randonautica, which I was not previously familiar with. It seemed interesting and I got to reading. On the page describing The Fatum Project a few links are provided that relate to Scott’s work. I clicked and read through them. After reading through them I google Psigenics, which brought me to,, and I used the Wayback Machine to view old content from the last two of those sites before going to explore the coreinvention site. It was there that under the Mind-Enabled menu I found the link to the forum. Since I had to create a login even to view the forum I initially signed up with a fake name just to be able to check it out. But now my name is updated. :slight_smile:

I have had that Emotiv for probably six months now but haven’t gotten much use from it. 15 of so years ago I built an EEG using the OpenEEG project, and over the years have tried out most of the sub-$500 consumer models on the market such as the Neurosky Mindwave and the Muse 2. I’ve generally found them very difficult to work with, lots of data, lots of noise, lots of variables, and typically not too user-friendly from a developer perspective, usually requiring the use or creation of complex algorithms to make sense of signal. The Emotiv Epoch should be better, lots of electrodes, what seem to be some moderately decent dev tools… but I haven’t experimented too much probably in large part do to lingering biases from past EEG exploration. Research from others seems to suggest some strong correlation between certain brainwave states, particularly alpha, and successful psi experiments, which could mean that limiting evaluated results during psi experiments to times when the requisite brainwave state is dominant could help to increase significance of experimental outcomes. I aspire to play more with that someday. The Emotiv device does have a monthly subscription fee, which I pay every month, so that’s a motivating factor if nothing else.

The MED Farm link you provided seems very useful in making these devices more widely available to others. I was not aware of this previously. It sounds like Scott is going to lend me a physical device, which I’m very eager to explore, so for now I think I’m probably personally best served using that route as it’s probably simpler to get working and to modify. I’ll spend a bit more time checking out the MED Farm info though. I was having some trouble with the drivers required for the physical device that I hope I can solve. If not then the API seems like a great alternative.

Thanks for the welcome. I’m based in Miami at the moment. I’ve been to Singapore many times, and generally like it quite a lot, for visiting, at least. :slight_smile:

Hello. Have you tried to create a training set for machine-learning-based MMI detection. For example you can display a deviation graph as PEAR did, and make operator influence it to go up. The exact timing of operators intent can be detected using emotiv EEG, since its capable of detecting when someone thinks about certain direction in space. In this exact time, bits from QRNG can be saved and used for ML-training.

The other approach, we tried recently, is replacing unstable human element with plant-based intention source. You can see it in this topic Psionic Generator Experiment

Also if you want to know more detailed information about Randonautica mechanics, it can be found here Mind Matter Interaction | Randonauts Wiki | Fandom

Hi there - I have not tried machine-learning-based MMI detection. I mentioned I am using in my experiments real-time feedback, both via graph and audio tone, that portray the probability of the results being random. This gives the operator feedback regarding how they are doing. It sounds similar to what you’re describing but it doesn’t use any machine learning, and I’m also not specifically familiar with the aspects of PEAR that you are describing (though of course familiar with PEAR itself and broadly of various of their efforts). I’d be interested in learning more about what you’re describing / suggesting here. Any references you can point me to? In what way do you envision machine-learning coming into play?

I spent some time already looking through the setup and results of the plant-based experiment you referenced. It’s very impressive if only for the creativity and dedication that it portrays. The fact that you were able to get a p-value of around 0.05 is quite promising. It seems like it warrants further exploration to try to both get repeated / consistent results as well as to perhaps try to optimize variables for more significance. Looks like that project is still actively underway so I can only assume you are thinking along these same lines. It’s a very interesting approach.

In the PEAR experiments, there was a graph of cumulative deviation that went up when the probability of rolling 1s outnumbered the probability of rolling 0s. The observer, by force of intention, tried to make the graph go up.

To increase the sensitivity of the experiment to user intent, Scott proposed a method called Bias Amplification, which compresses the data without losing statistical information. We thought that if the amplifier was equipped with an ML system for detecting the presence of a signal, this could significantly increase the sensitivity of the experiment by removing noise.

Since emotiv’s EEG interface is designed to control the mouse using brainwaves, it should be great at detecting when the user is thinking of a certain direction. If we mark the data received from the QRNG on the basis of the presence of an upward movement signal from the EEG and the presence of an actual upward movement on the accumulated deviation plot, then we can create a training set for machine learning and learn to recognize the psi-signal presence pattern in random data.

What you describe from PEAR sounds like exactly what I also am doing, which is real-time calculation of probability of 1s versus 0s, well technically in my case technically 1s vs 2s since that’s what the RNG I’ve been using most is set to produce (as mentioned, I use uranium to increase frequency of number generation, but I understand from Scott that frequency is still much too low). I’m recalculating probability after every new number is generated using a binomial test. That is used to drive real-time audio and visual feedback to the observer. Here’s a short demo or it in action, and link to code. [Note that in my approach I’ve presented to the observer (1-p) rather than p directly, strictly because I want to see the graph go UP not DOWN when things are going well.]

The ML approach could be interesting, I think. ML works best for things that have a large number of different kinds of complex but quantifiable inputs and then a clearly-defined outcome. I think our use case could fit, with caveats. In addition to simply number-generated we could quantify things like geomagnetic activity (daily average antipodal index), temperature, heart rate, brain wave state, years meditating (and in what way), level of relaxation, emotional response, belief system, personality type, and assign values from them. Then on the outputs side we train the model based on optimizing for lower p-values. There’s clearly a lot of effort required for all of this, primarily I think that could come in terms of input accumulation - being able to reliably measure/quantify all of those inputs. Probably these tests should ideally be done in a light-tight electromagnetically shielded chamber - I think regarding EMF it’s better to try to exclude the variable rather than quantify because it’s difficult/costly to measure with precision. In any case, a complex approach to execute, but theoretically feasible.

I like your idea regarding use of the Emotiv Epoch. Though I’m aware of the BCI (brain-computer interface) function offered I hadn’t been focusing my thinking in that direction. I was more simply thinking in terms of dominant brainwave state monitoring. This of course means I can’t currently speak to the usability of the data we may obtain by measuring 1/0 or up/down intent, but it does feel like an area worthy of some exploration.