Woohoo! My first significant ESP test!

I suppose it wouldn't be hard to set up some tests like this on psiencequest.net, in which the server generated the random numbers and recorded the guesses. The kind of trials people are doing here would then become evidential.

You could do that locally, generating random numbers between 1 and 5 to use as guesses. Use a browser testing suite like Selenium to click on the page, then read the result. If that gets consistently worse results over many runs than a human it's further evidence.

There's also other sites like psiresearch.com which record on the server.

Another 9/25 on my first try just now. This one had a bit of a decline effect where I got most of the first 10 guesses right, but then didn't do so well as the test went on. The reported effect where you start off doing well then get a bit bored or lose the same concentration definitely applied here. One thing that helped though was listening to this relaxing music throughout the session, and just before that l listened to this meditation for metta. I only needed one test to score 9 I think because I was relaxed and in a more loving state after listening to the meditation.

   

[Laird]

Oliver, based on what you've posted to this thread, it seems that these are your results, where across all tests, number of trials (n) = 25, probability of success on each trial (p) = 0.2, and number of successes is r, with the p-value representing the probability of getting r or more successes by chance alone (i.e., it is one-tailed):

First test:
r=9, p-value = 0.04677.
Second and third tests:
r=6, p-value = 0.38331.
r=10, p-value=0.01733.
Fourth and fifth tests:
r=9, p-value=0.04677.
r=9, p-value=0.04677.
Sixth test:
r=10, p-value=0.01733.
Seventh test:
r=9, p-value=0.04677.

Overall: n=175
r=62, p-value=0.000001477585.

So, if due to chance alone and not psi, you would achieve this cumulative number of successes or more only ~1.5 times every million attempts, which is... prodigious. As I understand it, a result this good would win Randi's prize if you could reproduce it under mutually-agreed conditions.

Some questions:

Does the above accurately represent your results as shared to this thread?

Are there any tests other than the above that you haven't mentioned? If so, what results did you get and how do they alter the overall p-value?

Do you think you could reproduce these results under whatever conditions you think you might be able to mutually agree on with the JREF?

A few other comments:

I've investigated the Javascript running this test and can't find any issues with it. I looked at your Java code testing the site too (which you posted in another thread) and couldn't see any issues with it either, though I didn't try to run it, in part because I didn't want to mess around with working out how to reconfigure the library it depends on, given that it seems to require an executable browser file which I don't have, since I'm running Linux not Windows. And it might have been an easy reconfiguration but anyhow I'm satisfied enough with a visual inspection of your code.

I think Chris's idea of developing a similar test to host on Psience Quest which would eliminate any potential for client-side cheating is a good one. In the event that you or anybody else getting prodigious results decided to apply to the JREF, its code might then be usable as the mutually-agreed basis for testing. Is anybody interested in coding up such a test? Mediochre? Oliver? We could discuss first how it would be designed. I have a few ideas.

Laird, 

The JREF closed the Million Dollar Challenge in 2015.

Linda

[Laird]

Thanks, Linda. I had heard that it had been closed but somehow thought I remembered it having been reopened. Oops.

The prize was nailed to the floor anyway, IMHO.

[Laird]

I think you're right, tim, but it would have been interesting, if possible, to have had a direct group (our forum community) experience with the JREF to demonstrate that to ourselves.

I've investigated the Javascript running this test and can't find any issues with it. I looked at your Java code testing the site too (which you posted in another thread) and couldn't see any issues with it either, though I didn't try to run it, in part because I didn't want to mess around with working out how to reconfigure the library it depends on, given that it seems to require an executable browser file which I don't have, since I'm running Linux not Windows. And it might have been an easy reconfiguration but anyhow I'm satisfied enough with a visual inspection of your code.

I think Chris's idea of developing a similar test to host on Psience Quest which would eliminate any potential for client-side cheating is a good one. In the event that you or anybody else getting prodigious results decided to apply to the JREF, its code might then be usable as the mutually-agreed basis for testing. Is anybody interested in coding up such a test? Mediochre? Oliver? We could discuss first how it would be designed. I have a few ideas.

I'm happy to make a similar test to the one on the other site which does the same thing and stores the results. It wouldn't be hard, just an Ajax POST after each click with whether it's a success or not. I could make a server-side script to record the results. We'd then have a growing collection of data.

[Laird]

I'm happy to make a similar test to the one on the other site which does the same thing and stores the results. It wouldn't be hard, just an Ajax POST after each click with whether it's a success or not. I could make a server-side script to record the results. We'd then have a growing collection of data.

Nice! If you're willing to put together something that you think you would be most comfortable using and most likely to get positive results on, whilst eliminating the possibility of cheating - and keeping the code open so others can verify that too - then I reckon that would be a welcome addition to Psience Quest.

Nice! If you're willing to put together something that you think you would be most comfortable using and most likely to get positive results on, whilst eliminating the possibility of cheating - and keeping the code open so others can verify that too - then I reckon that would be a welcome addition to Psience Quest.

Ok is Psience Quest able to host an extra script? If I make a php script as an endpoint we can use to store the results, could it be hosted here? If we can host the client-side code too, then we'd have our own ESP test.