next »

[meushiman]

Wait I got 34?!  YES!  I guess I checked the numbers the wrong way.  Wahoo!

[Akaran]

Or he's the first person to get numerically struck by lightning. Him getting most of the right numbers is just as likely as getting another specific result. To better validate the test you would have to repeat it multiple times.

I'm not saying that The Adfeng isn't a psion, I'm just saying that some people DO win the lottery. (That's why the losers keep coming back.)

[meushiman]

Good point, but I have faith in him.  Hey what is the date today?

[linkmaster03]

(I am not a troll. I am interested in psionics and used to train, but recently, I have become very skeptical.)

Seriously? If anything, this test disproves psionics. Your participants performed just as good as one would expect if they chose their numbers randomly. To test this, I quickly whipped up a Python script to grab 50 random integers from random.org, just as you did, and compare them with the numbers you chose. After this, the script reports how many of the new random numbers match up to the ones you used for the experiment. Here are the results of ten runs:

Code:

Correct: 39
Correct: 38
Correct: 39
Correct: 37
Correct: 36
Correct: 35
Correct: 39
Correct: 32
Correct: 38
Correct: 33

It looks like Adfeng's results are no better than random chance. I'm not trying to troll or flame here, just provide facts. Here is the script, if anyone wants to run it for themselves:

Code:

import urllib2

correct = 0
# Define the list used in the experiment, provided by Notagh
RANDOM = ['2', '5', '4', '5', '1', '1', '2', '7', '9', '10', '4', '7', '1',
    '7', '7', '6', '2', '9', '5', '7', '6', '6', '6', '4', '10', '5', '5',
    '2', '6', '9', '4', '4', '9', '1', '10', '10', '7', '7', '4', '10',
    '7', '4', '9', '2', '6', '10', '10', '2', '4', '6']

# Get a new list of numbers from random.org
page = urllib2.urlopen('http://www.random.org/integers/?num=50&min=1&max=10&col=1&base=10&format=plain&rnd=new').read()
new_list = page.split('\n')[:-1]

# Run over each number from the new list, and if it is in the original list,
# remove the first one and increment the amount of numbers correctly matched
for number in new_list:
    if number in RANDOM:
        del(RANDOM[RANDOM.index(number)])
        correct += 1

print('Correct: %i' % correct)

[Caliph]

anyways why would you want to prove it to anyone... besides for egotistical reasons.. better to keep it in the dark..too many people out there who would abuse power.. better that they remain ignorant..

[Notagh]

(I am not a troll. I am interested in psionics and used to train, but recently, I have become very skeptical.)

Seriously? If anything, this test disproves psionics. Your participants performed just as good as one would expect if they chose their numbers randomly. To test this, I quickly whipped up a Python script to grab 50 random integers from random.org, just as you did, and compare them with the numbers you chose. After this, the script reports how many of the new random numbers match up to the ones you used for the experiment. Here are the results of ten runs:

Code:

Correct: 39
Correct: 38
Correct: 39
Correct: 37
Correct: 36
Correct: 35
Correct: 39
Correct: 32
Correct: 38
Correct: 33

It looks like Adfeng's results are no better than random chance. I'm not trying to troll or flame here, just provide facts. Here is the script, if anyone wants to run it for themselves:

Code:

import urllib2

correct = 0
# Define the list used in the experiment, provided by Notagh
RANDOM = ['2', '5', '4', '5', '1', '1', '2', '7', '9', '10', '4', '7', '1',
    '7', '7', '6', '2', '9', '5', '7', '6', '6', '6', '4', '10', '5', '5',
    '2', '6', '9', '4', '4', '9', '1', '10', '10', '7', '7', '4', '10',
    '7', '4', '9', '2', '6', '10', '10', '2', '4', '6']

# Get a new list of numbers from random.org
page = urllib2.urlopen('http://www.random.org/integers/?num=50&min=1&max=10&col=1&base=10&format=plain&rnd=new').read()
new_list = page.split('\n')[:-1]

# Run over each number from the new list, and if it is in the original list,
# remove the first one and increment the amount of numbers correctly matched
for number in new_list:
    if number in RANDOM:
        del(RANDOM[RANDOM.index(number)])
        correct += 1

print('Correct: %i' % correct)

Thus probability chances based on assumption that they have to be whole numbers.

Edit: Even ignoring that assumption, if a 78% correct was close to random chance then successfully gambling would be that much easier.  Guessing 50 numbers that are 1->2 and getting 78% right (roulette) would no doubt get you quite rich depending on how much chips you put on each time.

[linkmaster03]

You only generated integers 1-10, so your assumption means nothing. This doesn't relate to gambling at all, because in this test, the order did not matter.

[meushiman]

That code thing says ten are correct out of 50 you said.  That's 1/5 or more simply 20%, no where near 78%

[Nervous]

anyways why would you want to prove it to anyone... besides for egotistical reasons.. better to keep it in the dark..too many people out there who would abuse power.. better that they remain ignorant..

If psionics was to be proven, people probably would'nt abuse it. This is because they can easialy find more effective means than psi. Still there may be some who would, but I doubt they'd get very far with what ever they do.

[Notagh]

You only generated integers 1-10, so your assumption means nothing. This doesn't relate to gambling at all, because in this test, the order did not matter.

Did you bother to read my example?  A success rate of 78% in predicting the roulette numbers right is not the same as 50%, which would be random chance.  Placing an equal amount of chips each time, you could presume that a 78% correct would double your money after x tries...

And as I have mentioned, the experiment is flawed and I will have to redo it to confirm it's validity.
I've talked to leodragon about this and I will do it soon, but the experiment's flaw has nothing to do with what you said, rather, his concern was that random.org was not tested to be random by us during the course of experiment, ect.  He suggested testing for alteration of data from a controlled source via micropk be a better idea, and I'm working on it.

And, as for your script, I didn't try it, but :

That code thing says ten are correct out of 50 you said.  That's 1/5 or more simply 20%, no where near 78%

Something tells me I don't have to

If you don't think this is proof, kudos to you, because it might not be.  It's probably proof, however.  On the other hand, what do you think of the Retropsychokinesis online experiments?  I'll gladly show you my log, for it probably also is proof.

Of course, I don't think you're actually looking for proof when you look at this topic.

[linkmaster03]

That code thing says ten are correct out of 50 you said.  That's 1/5 or more simply 20%, no where near 78%

I highly doubt you ran my script. If you did, run it more than once; 10/50 is incredibly improbable. I'm using Python 2.6.4 on Linux to run it. Here are the results of 60 runs, and the lowest was 27/50 (highest 43/50, average 35.53/50):

Code:

Correct: 39
Correct: 40
Correct: 35
Correct: 36
Correct: 32
Correct: 31
Correct: 32
Correct: 31
Correct: 38
Correct: 37
Correct: 27
Correct: 34
Correct: 35
Correct: 34
Correct: 36
Correct: 32
Correct: 33
Correct: 37
Correct: 36
Correct: 32
Correct: 37
Correct: 36
Correct: 36
Correct: 36
Correct: 34
Correct: 33
Correct: 33
Correct: 36
Correct: 37
Correct: 30
Correct: 34
Correct: 40
Correct: 32
Correct: 36
Correct: 34
Correct: 35
Correct: 38
Correct: 36
Correct: 39
Correct: 35
Correct: 42
Correct: 34
Correct: 33
Correct: 40
Correct: 35
Correct: 38
Correct: 31
Correct: 38
Correct: 36
Correct: 38
Correct: 35
Correct: 38
Correct: 34
Correct: 36
Correct: 38
Correct: 35
Correct: 37
Correct: 36
Correct: 43
Correct: 41

Of course, I don't think you're actually looking for proof when you look at this topic.

I would love to have proof of psionics, and I would happily change my position if you conducted a more convincing experiment.

[meushiman]

First off I don't understand why you are trying to disprove something you are a part of.  Secondly, what are you talking about.  You ran your script sixty times?  Where do the guesses come in?  Did you guess?  Did you keep the same guess throughout?  Are you sure you did not use psionics?  There are a ton of questions that could go along with this.  This is why I trust Notagh's experiment better.  Yes, it could use some tweaking, but yours isn't explained very well

[linkmaster03]

First off I don't understand why you are trying to disprove something you are a part of.  Secondly, what are you talking about.  You ran your script sixty times?  Where do the guesses come in?  Did you guess?  Did you keep the same guess throughout?  Are you sure you did not use psionics?  There are a ton of questions that could go along with this.  This is why I trust Notagh's experiment better.  Yes, it could use some tweaking, but yours isn't explained very well

Here's what the script does:

I quickly whipped up a Python script to grab 50 random integers from random.org, just as you did, and compare them with the numbers you chose. After this, the script reports how many of the new random numbers match up to the ones you used for the experiment.

[The Adfeng]

I don't see how that proves/disproves anything.  Pick your own numbers and run the test, pick different numbers EACH TIME, and then average the results.

[linkmaster03]

Okay, I've modified the script so that two random lists of 50 integers from 1-10 are chosen instead of using the original 50 from Notagh to match against. I hope I undestood your request correctly. Here is the modified code:

Code:

import urllib2

correct = 0

# Get two lists of numbers from random.org
page = urllib2.urlopen('http://www.random.org/integers/?num=50&min=1&max=10&col=1&base=10&format=plain&rnd=new').read()
page2 = urllib2.urlopen('http://www.random.org/integers/?num=50&min=1&max=10&col=1&base=10&format=plain&rnd=new').read()
original_list = page.split('\n')[:-1]
secondary_list = page2.split('\n')[:-1]

# Run over each number from the new list, and if it is in the original list,
# remove the first one and increment the amount of numbers correctly matched
for number in secondary_list:
    if number in original_list:
        del(original_list[original_list.index(number)])
        correct += 1

print('Correct: %i' % correct)

Here are 60 more results from the new script (each result is a separate run, so new numbers are used every time):

Code:

Correct: 35
Correct: 42
Correct: 43
Correct: 44
Correct: 38
Correct: 42
Correct: 41
Correct: 33
Correct: 42
Correct: 36
Correct: 38
Correct: 41
Correct: 35
Correct: 32
Correct: 36
Correct: 28
Correct: 38
Correct: 37
Correct: 33
Correct: 36
Correct: 41
Correct: 34
Correct: 40
Correct: 36
Correct: 37
Correct: 44
Correct: 35
Correct: 38
Correct: 35
Correct: 41
Correct: 36
Correct: 35
Correct: 32
Correct: 41
Correct: 41
Correct: 42
Correct: 39
Correct: 39
Correct: 41
Correct: 37
Correct: 36
Correct: 34
Correct: 38
Correct: 39
Correct: 37
Correct: 35
Correct: 35
Correct: 41
Correct: 33
Correct: 36
Correct: 40
Correct: 41
Correct: 40
Correct: 38
Correct: 35
Correct: 38
Correct: 36
Correct: 43
Correct: 33
Correct: 44

Lowest: 28/50
Average: 37.77/50
Highest: 44/50

From the data, we can conclude that, by entering 50 random integers from 1 to 10 into Notagh's experiment (matching against another list of 50 random integers from 1 to 10), it is very probable that 30-45 numbers will match correctly (when disregarding order). This means that your score of 39/50 is no better than random chance, and psionics have no effect.

Please feel free to correct me if I have made any incorrect assumptions.

[Notagh]

As I have previously mentioned I am working on creating an experiment that should prove this beyond your reasonable doubts.

On the other hand, I'd like for you to consider not using random.org as the RNG in both cases, for obvious reasons.

EDIT: Running script now.
Can't run it...  You're using 2.6.4 terminology.  Please upgrade to 3. python ><

~~~
How much math did you learn?

I want you to
   1) Go to this site: http://www.fourmilab.ch/rpkp/experiments/
   2) Try out the experiments, with setting practice or setting record, and put your nickname for the email.
   3) Tell me when you've finished.  I'll go check your log.
   4) Then, I'll show you my log.
   5) Viola! Proof/disproof* of psionics!

*Well, technically a lack of proof does not disprove something.

[linkmaster03]

As I have previously mentioned I am working on creating an experiment that should prove this beyond your reasonable doubts.

I am eager to see it. Good luck.

On the other hand, I'd like for you to consider not using random.org as the RNG in both cases, for obvious reasons.

Such as?

EDIT: Running script now.
Can't run it...  You're using 2.6.4 terminology.  Please upgrade to 3. python ><

It should work in Python 3. You can post the error you get if you really want to run the script.

[meushiman]

First off I don't understand why you are trying to disprove something you are a part of.  Secondly, what are you talking about.  You ran your script sixty times?  Where do the guesses come in?  Did you guess?  Did you keep the same guess throughout?  Are you sure you did not use psionics?  There are a ton of questions that could go along with this.  This is why I trust Notagh's experiment better.  Yes, it could use some tweaking, but yours isn't explained very well

Here's what the script does:

I quickly whipped up a Python script to grab 50 random integers from random.org, just as you did, and compare them with the numbers you chose. After this, the script reports how many of the new random numbers match up to the ones you used for the experiment.

Oh, sorry

[goodolemarvin]

Hi All,

I know this is an older thread, but I wanted to throw a few things into the mix. If you want to test the difference between an expected and observed response you should use a chi-square. The formula is the Summation of [(Observed-Expected)^2/Expected] for each element. We can then determine whether or not the set of observed responses is statistically different from the expected responses. In the case where we know the expected values (Notagh's numbers) and the observed (Adfeng's numbers) we want this value to be non-significant. That is, we want there to be no significant difference between Notagh's and Adfeng's numbers.

In this case, unfortunately the results are significantly different. Chi-squared value (even with Yate's Correction) is 17.97 with 9 degrees of freedom. From the t-table, we can see that this means we can be 99.999% confident that Adfeng did not accurate match Notagh's numbers to any statistically significant degree. I would recommend two possible modifications to the test that might make it a bit more rigorous:

1.) Increase the range of possible numbers (i.e. 1-40 instead of 1-10) as well as the number of items (i.e. 200 instead of 50) and/or

2.) Make order matter (i.e. 8-5-5 is different than 5-8-5)

Either one would give us a more accurate and rigorous measure from a statistical point of view. Hopefully you find this helpful! 

[Notagh]

Hi All,

I know this is an older thread, but I wanted to throw a few things into the mix. If you want to test the difference between an expected and observed response you should use a chi-square. The formula is the Summation of [(Observed-Expected)^2/Expected] for each element. We can then determine whether or not the set of observed responses is statistically different from the expected responses. In the case where we know the expected values (Notagh's numbers) and the observed (Adfeng's numbers) we want this value to be non-significant. That is, we want there to be no significant difference between Notagh's and Adfeng's numbers.

In this case, unfortunately the results are significantly different. Chi-squared value (even with Yate's Correction) is 17.97 with 9 degrees of freedom. From the t-table, we can see that this means we can be 99.999% confident that Adfeng did not accurate match Notagh's numbers to any statistically significant degree. I would recommend two possible modifications to the test that might make it a bit more rigorous:

1.) Increase the range of possible numbers (i.e. 1-40 instead of 1-10) as well as the number of items (i.e. 200 instead of 50) and/or

2.) Make order matter (i.e. 8-5-5 is different than 5-8-5)

Either one would give us a more accurate and rigorous measure from a statistical point of view. Hopefully you find this helpful! 

Bah.  You're right, and I know.  I had realized mid-way though the experiment that the probability calculations, which were wrong and provided on false assumptions, assumed that the possible numbers were not necessarily
whole numbers, while random.org has whole numbers.  I also researched on the correct formula during the experiment, reading retropsychokinesis's articles and info on the chi-square calculations.  I always did intend to run a nonfaulty experiment, but with the lack of support for other experiments I kind of buried my thoughts on this one and forgot about it.