openMMIND project searching for beta testers

thank you very for your help

my knowledge on these kind of subjects is very limited, I hope your expertise could help us to develop a solid research and provide the tools so that people worldwide can carry out their own experiments. 
 
As far as I know and also from my personal experience, telekinesis/precognition seems to fail often when the scientific verification process is rigorous, yet in small experiments or even personal ones the results challenge any typical explanation. There seems also to be a strong impact of people skepticism in results. If the telekinesis/precognition phenomenon is actually real, these are probably the main reasons why it is rejected by scientific community and most people. It is hard to make serious science in such conditions.


One of the key goals of the openMMIND project is to bring to everyone the possibility to test by themselves. Perhaps it may help reducing this (hypothetical) negative impact of skepticism. 

In this first phase of the experiment I also intend to see if joining people together with similar “abilities” will boost the results. I mean, we make a first run, and then we separate people with opposite results in two groups (one with negative p-values and the other with positive p-values). Next, the experiment is repeated, and we verify if the two groups (previously defined) still deliver statistically significant different results.

I hope you enjoy the application, it is very simple (adapted from another person code, I will provide the respective credits in future). I’m not a developer, my programing skills come from scientific data analysis. In future we need to make this remote (like a web application) to avoid any possible hacking, and also bringing it to smartphones. Using a remote radioactive source is also a possibility, I can provide it since I’m a nuclear physicist.

I think you mean larger and smaller p values. p values are always between 0 and 1. The smaller they are, the more statistically significant the departure from chance expectation.

I think you mean larger and smaller p values. p values are always between 0 and 1. The smaller they are, the more statistically significant the departure from chance expectation.

you are right, thank you   I meant z-score, not p-value. 
In this game you are successful if the hardware entropy source generates more 0s than 1s during the bomb disarming. 
Assuming telekinesis/precognition is real, people could influence/predict the RNG in order to get more 0s than 1s, but I guess there could be also people that are "in some way pessimistic" and get more 1s than 0s  (consequently getting more "game over").

So my idea is to separate these two kind of results in 2 groups of people in a first phase, perhaps also discarding people with low p-values. Next we could make a prospective registry of the experiment, describing the statistics analysis we intend to do with these two groups of participants. Finally, we ask people to run again the experiment, i.e. the game, and we analyse again their data, now taking into account the two groups previously defined. 
what do you thing?

of course, this is only a small part of what the openMMIND project intends to be
my idea is to create a platform where all these data and software will be shared, allowing anyone to perform their own analysis and experiments. 
I would like also to use some deep learning techniques to analyse data.

[Laird]

[b]Laird, [/b]
perhaps I will need time to get this working on other platforms, because I'm not an application developer. 

for now, could you please try the following .exe files ? which I compiled with compatibility options 
https://drive.google.com/file/d/1fE31Tmj...sp=sharing
you can try them on all of your VMs and Wine

you may also try to change the compatibility options when you open the exe on windows (right mouse > properties > compatibility tab)

on Wine, I suspect there the problem is related to the Wine
for instance, I found this workaround https://askubuntu.com/questions/723295/w...ase-gaming


I hope you can get it working

Unfortunately, I've had no success with that. Here are my results:

1. Wine: Same as before (I'm not on Ubuntu but did try turning off key repeat in my KDE settings).
   
2. Windows 7 64-bit under Virtualbox: Same problem.
   
3. Windows 7 32-bit under Virtualbox: Same error. To try to fix it, I tried installing the Visual C++ Redistributable for Visual Studio 2015 from Microsoft but I encountered an error while installing it.
   
4. Windows XP under Virtualbox: New error: "The procedure entry point RegSetKeyValueA could not be located in the dynamic link library ADVAPI32.dll."
   

you are right, thank you   I meant z-score, not p-value. 
In this game you are successful if the hardware entropy source generates more 0s than 1s during the bomb disarming. 
Assuming telekinesis/precognition is real, people could influence/predict the RNG in order to get more 0s than 1s, but I guess there could be also people that are "in some way pessimistic" and get more 1s than 0s  (consequently getting more "game over").

So my idea is to separate these two kind of results in 2 groups of people in a first phase, perhaps also discarding people with low p-values. Next we could make a prospective registry of the experiment, describing the statistics analysis we intend to do with these two groups of participants. Finally, we ask people to run again the experiment, i.e. the game, and we analyse again their data, now taking into account the two groups previously defined. 
what do you thing?

of course, this is only a small part of what the openMMIND project intends to be
my idea is to create a platform where all these data and software will be shared, allowing anyone to perform their own analysis and experiments. 
I would like also to use some deep learning techniques to analyse data.

Thanks. I think it would make sense to have an initial phase of data gathering , so that you can gauge whether you're seeing anything potentially anomalous, and if so what, before deciding how to analyse the main results. I definitely think it's a good idea to pre-register the main analysis. Of course, if you envisage the experiment running for a long period, you could do it in several stages, refining the analysis progressively, so long as the analysis for each stage is fixed before it begins, and each stage is completed before you move on to the next.

Even if the first stage is mainly exploratory, of course you can pre-register some basic statistics to be used, of which the overall percentage of 0s is the most obvious.

If you are thinking about "psi-missing"/negative scores, the variance of the percentage of 0s between individual runs might also be interesting. But with variance there is a danger of bias owing to people stopping prematurely if they get discouraged.

That shouldn't be a problem for the overall percentage of 0s, provided no data are actually discarded if people drop out. It's important that everything is counted, including data from incomplete games, so long as the player is getting feedback about the random numbers. I think also it wouldn't be too hard to define a statistic reflecting the variance that wasn't vulnerable to this kind of bias, but one would need to be a bit careful.

I'd be a bit cautious about the extent to which you'll be able pick out over- or under-performers from a pilot study, as these effects are typically so weak even in large experiments. But really that's something that could be gauged only by looking at the numbers. There might be alternative ways of proceeding, other than by just dividing people into two groups - such as looking at the variance between different runs for the same subject. But I think again care would be needed to avoid such an analysis being biased by premature stopping.

I'm sure there are quite a few interesting aspects you could explore with such a set-up. One might be the question of PK versus precognition. On the precognition hypothesis, if the player had control of the timing of the game, the result could be affected by choosing a starting point for each run which produced more 0s than 1s, rather than by influencing the generation of the random numbers. If you can find a measurable effect, this could be studied by comparing players' performance when they are free to control the timing, rather than the timing being imposed on them.

Unfortunately, I've had no success with that. Here are my results:

1. Wine: Same as before (I'm not on Ubuntu but did try turning off key repeat in my KDE settings).
   
2. Windows 7 64-bit under Virtualbox: Same problem.
   
3. Windows 7 32-bit under Virtualbox: Same error. To try to fix it, I tried installing the Visual C++ Redistributable for Visual Studio 2015 from Microsoft but I encountered an error while installing it.
   
4. Windows XP under Virtualbox: New error: "The procedure entry point RegSetKeyValueA could not be located in the dynamic link library ADVAPI32.dll."
   

after struggling with this for many hours I believe I solved the problem 

I installed the windows 7 32bit in a virtual machine (virtualbox)
I found the same problems you described.

I solved easily the error related with the 32 OS version (just some setting related with he app compilation)
But it took a while to realise what was the problem with the "transparent window" you reported.
It seems that some virtualisation software do not work well with hardware acceleration.
Changing the game rendering to software instead of hardware solve the problem.

Now the app have two versions, you should use the VM compatible version when the other does not work, as in your case. The optimisation could be worst, but your PC should handle it. If you experience some lag while gaming, try to increase the number of cores, RAM or GPU memory in the virtualbox settings.

This new version should work also on your 32bit win7.
In any case, could you please test it in all of your virtual machines, and report it?

[Link removed due to detected virus]

thanks

Thanks. I think it would make sense to have an initial phase of data gathering , so that you can gauge whether you're seeing anything potentially anomalous, and if so what, before deciding how to analyse the main results. I definitely think it's a good idea to pre-register the main analysis. Of course, if you envisage the experiment running for a long period, you could do it in several stages, refining the analysis progressively, so long as the analysis for each stage is fixed before it begins, and each stage is completed before you move on to the next.

Even if the first stage is mainly exploratory, of course you can pre-register some basic statistics to be used, of which the overall percentage of 0s is the most obvious.

If you are thinking about "psi-missing"/negative scores, the variance of the percentage of 0s between individual runs might also be interesting. But with variance there is a danger of bias owing to people stopping prematurely if they get discouraged.

That shouldn't be a problem for the overall percentage of 0s, provided no data are actually discarded if people drop out. It's important that everything is counted, including data from incomplete games, so long as the player is getting feedback about the random numbers. I think also it wouldn't be too hard to define a statistic reflecting the variance that wasn't vulnerable to this kind of bias, but one would need to be a bit careful.

I'd be a bit cautious about the extent to which you'll be able pick out over- or under-performers from a pilot study, as these effects are typically so weak even in large experiments. But really that's something that could be gauged only by looking at the numbers. There might be alternative ways of proceeding, other than by just dividing people into two groups - such as looking at the variance between different runs for the same subject. But I think again care would be needed to avoid such an analysis being biased by premature stopping.

I'm sure there are quite a few interesting aspects you could explore with such a set-up. One might be the question of PK versus precognition. On the precognition hypothesis, if the player had control of the timing of the game, the result could be affected by choosing a starting point for each run which produced more 0s than 1s, rather than by influencing the generation of the random numbers. If you can find a measurable effect, this could be studied by comparing players' performance when they are free to control the timing, rather than the timing being imposed on them.

thank you very much for your detailed insights, you seem to have a solid experience on this subject,that is great since my experience is very limited. Nonetheless, I believe my background as a physicist could be helpful, especially from the point of view of the hardware. 

Indeed, the two biggest concerns during the design of this experiment were the possible data manipulation performed by the subjects, and premature termination of the task. 
To minimise those problems:
1- the subject should perform the task until the end of the section, never terminate before.
2- there is a complex set of algorithms to ensure the data integrity (e.g. the subject can delete or modify data files), the encrypted data in data files allow me to know if some one tried to manipulate them

of course, these protections are not 100% bullet proof. In future I would like to make Web, and smartphone applications, possibly with a real-time stream of random data (e.g. generated by radioactive decay) coming in from a remote server, and a stream of user data going out. In this way, I think no data manipulation would be possible. Moreover, a Web or smartphone application would be much easier to captivate people to enter in the study or to make their own experiments. 

Regarding the analysis, during the experiment design and testing I also had several ideas, there is indeed much potential in such a setup, but let us discuss this when we have some data in hands.

For now, we need beta testers just to see if it's everything right with the experiment. I hope people in this forum can help me with this.

after running my algorithms to ensure the data integrity, I can share with you all data. you can perform your own analyses and we can discuss what to do next. I'm using MATLAB for analysis but I can provide you data in the format you want. 

It would be nice if you try the application, I could explain you how the experiment is designed, and what is the meaning of data in the output files. I would like to have your feedback, we can discuss any possible modification of the experiment software before the main launch. 

cheers,

[Laird]

In any case, could you please test it in all of your virtual machines, and report it?

AVG in my 32-bit Windows 7 installation running under VirtualBox reports that this file contains the Win32/Heri trojan.

When I scanned it online at TotalVirus, two antivirus engines detected a trojan in it: Microsoft detected Trojan:Win32/Fuery.C!cl, and Rising detected Trojan.Fuerboos[...]. [Edit: now that I return to that cached page over a day later, it has been updated so that now four engines, not two, detect trojans in it - but different trojans to the original ones I first listed. Perhaps this suggests false positives...]

Whilst these could be false positives (the vast majority of antivirus engines did not detect a threat, and none of the engines at Jotti's virus scan detected a threat), it potentially poses a threat to anybody who downloads it, so I've deleted the link from your post.

For caution, I'm also going to delete the links to previous files that you've posted, even though none of the engines at TotalVirus detected a threat in either of them (see here and here). [Edit: On reflection, this seems like overkill and unfair to you as you try to get this project off the ground, so I've restored those links.]

Potentially, your machine has become infected since posting the first two files?

[Laird]

In any case, could you please test it in all of your virtual machines, and report it?

Prior to the virus being detected in 32-bit Windows 7, I had tested the app both under Wine and in 64-bit Windows 7 running on VirtualBox, and your fixes seem to have worked: it now runs fine on both platforms and is now responsive to keyboard presses under Wine. Hopefully you can clean the infection (if any) on your machine so we can keep on testing.

Prior to the virus being detected in 32-bit Windows 7, I had tested the app both under Wine and in 64-bit Windows 7 running on VirtualBox, and your fixes seem to have worked: it now runs fine on both platforms and is now responsive to keyboard presses under Wine. Hopefully you can clean the infection (if any) on your machine so we can keep on testing.

I'm stunned with your report, I'm using Visual Studio, the code is simple, I know everything the algorithm is doing. I'm compiling also with VS, the only thing I changed was related with the compilation and SDL2 rendering. 
I will try to understand why this false positive.

The openMMIND project intends to be as transparent as possible. The code will be available for everyone in GitHub, so anyone can make modifications and compile it.
If you are afraid of my app being some kind of virus, I can send you the code, you see what it is doing (in case you have some background in C programming) and compile it your self.

[Laird]

I'm stunned with your report, I'm using Visual Studio, the code is simple, I know everything the algorithm is doing.

I would expect that it's nothing to do with your code. I would expect instead that the virus/trojan has infected your entire system and via your system found its way into the executable you shared.

Edit: To be 100% clear, I'm not accusing you of creating a virus; instead, I simply suspect that you have been accidentally infected by one.