Merged New telepathy test: which number did I write ?

[Michel H]

For the record "reliability" is a prejudiced confirmation bias. The results that best match the desired number are retained, those that are wrong are ignored.

...

Let me use an example again.

Member NiceAndSmartGuy, answers:
I am almost completely sure the (numerical) answer to this interesting test is a xx. (with a MD5 hash).

Then, member BadMoon answers:
Dis is a compleitly horrible "test". It's obvius dat tis person dosn't now how to conduct a serius research. I anser xx. (with a MD5 hash also).

Then, I would regard the first answer as credible, and the second one as not-credible. I cannot, in this test, "cheat" (generally speaking, I would try to avoid it anyway) by favoring (by means of a high credibility rating) the member who has given the right numerical answer because I do not see the numerical answer. Then, I seem to have found before that credible members tend to provide more accurate numerical answers.

Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[Tomtomkent]

Tomtomkent and Loss Leader, do you really think such methods could be successful, lead to positive results (i.e. hit rates higher than chance)?

The method could be successful if the powers being studied were real.
That is kind of the point of the test.

Or, would they be more like "perfect recipes for failure", "appropriate for a skeptic forum" ?

You conflated two different things. One of which is not mutually exclusive from the point above.

(or appropriate, perhaps for a "Loss Leader"? ). Obviously, both of you speak with great confidence about something you've probably never done, as a researcher (a successful telepathy test, although Loss Leader has provided a remarkable and numerically correct answer, in one of my tests, but as a participant).

I speak confidently about correct test protocol. I see no reason to assume that telepathy works under different rules.

This post suggests you are aiming to make a test with the best possible chances of being successful.

You should be making a test that discerns if your powers are as you claim or not.

I regret that this new test thread (of the "rigorous" kind) has now been merged with the previous test thread, also of the "rigorous kind".

No. Neither were rigorous. At all. That is the point you continue to miss. You methodology is deeply floored and offers no rigor at all.

 

[Tomtomkent]

Let me use an example again.

Sure. But explaining a bad idea many times over does not make it a good idea. Coming up with a good idea is the solution.

Member NiceAndSmartGuy, answers:
I am almost completely sure the (numerical) answer to this interesting test is a xx. (with a MD5 hash).

Then, member BadMoon answers:
Dis is a compleitly horrible "test". It's obvius dat tis person dosn't now how to conduct a serius research. I anser xx. (with a MD5 hash also).

And so what? Your test is not of the opinion of people.
It is of telepathic power.

Then, I would regard the first answer as credible, and the second one as not-credible. I cannot, in this test, "cheat" (generally speaking, I would try to avoid it anyway) by favoring (by means of a high credibility rating) the member who has given the right numerical answer because I do not see the numerical answer. Then, I seem to have found before that credible members tend to provide more accurate numerical answers.

First of all: Yes you can cheat. You can quite obviously decide that people who give numerically wrong answers are lacking credibility in your subjective opinion. It is introducing bias.

That you think you need this just highlights the major flaw in your methodology. You think somebody who is jerking you around and trying to upset your test can guess the right answer for the wrong reasons... Because they have a 25% chance of being correct if they hear you or not.

You could avoid not only the need for credibility, which is a way of fudging the numbers no matter how many times you explain it and try to wish it were not, if you used a better methodology.

Do you understand why tests like this should be blinded?
Do you understand why they need to be controlled?
Do you understand why doctors do not look at the people who claimed they felt better with a placebo, only to decide they were less reliable than the ones who felt better with medicine?

If your aim is to prove that telepathy is real then multiple better methodologies have been suggested, none of which avoid this utter shambles of number-fudging. This excuse to discount results you don't like.

You should make a test that is objective, not a test that relies only on your subjective opinion of responses.

 

[Michel H]

...
First of all: Yes you can cheat. You can quite obviously decide that people who give numerically wrong answers are lacking credibility in your subjective opinion. It is introducing bias.
...

No, because, with the protocol I use in this rigorous test, once I finally see the numerical answers, I can no longer change the credibility ratings.


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[Innocuous]

Deleted - saw a later post addressing this.

 

[Innocuous]

With your revised protocol (where you can't preselect the right responses), I think this could be part of a valid test. Since we are talking about a psychic phenomenon, perhaps the clear-text statement provides you with psychic clues that establish whether or not you can make the proper psychic connection with the responder. That is just my own outrageous speculation of a mechanism I really don't believe.

The clear text gives you additional information that could be used for cold reading: maybe you infer what people will guess if you know just a tiny little bit about them. Maybe there is something in the words they chose or way they write that gives you a hint about the number they would choose or have already chosen - enough of a hint to tip the percentage beyond random. Maybe you have studied this for years and found some correlation. That's okay - it is still an interesting skill or power, although not awesome, and maybe even Challenge-worthy? Or maybe the additional information is an essential component of a truly psychic power.

If you have no special abilities at all, the best your pre-selection of candidates can do is reduce the total number of tests. As long as that remains a sufficiently large number, and maybe a sufficiently large percent of the original sample, I could buy into the protocol. Those limits would have to be written into the protocol so we aren't playing with numbers after the test is run.

However, the 1-4 would be a deal breaker for me, for the reasons already discussed. Bump that up to 0-9, at the very least, and preferably 0-1000, if your power allows it.

 

[Tomtomkent]

No, because, with the protocol I use in this rigorous test, once I finally see the numerical answers, I can no longer change the credibility ratings.

It doesn't matter which way round it is. You are still selecting which answers you want. You are still trying to force a positive result.

You said yourself there is a correlation between the answers you like and the numbers you want.

Any way you cut it you are still stacking a deck to make your chances of a positive result higher.

Your method is flawed and biased. It is not acceptable. It is worthless. It is an exercise in confirmation bias. Worryingly it may well be a failure despite the bias.

Why not address the actual problems instead of making excuses to ignore answers that you don't like?

 

[bruto]

Thank you, bruto, I appreciate it. I hope this is not a joke .


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

It's exactly as serious as my belief in your telepathy.

I apologize for not participating in your latest attempt, as any methodology that might possibly invest my answer with better than zero credibility would be inherently inaccurate.

 

[Michel H]

...
However, the 1-4 would be a deal breaker for me, for the reasons already discussed. Bump that up to 0-9, at the very least, and preferably 0-1000, if your power allows it.

Ah that's unfortunate, Innocuous. However, I don't rule out doing some day perhaps a test with a larger number of possibilities, perhaps 1, 2, 3, 4, 5 or even perhaps, like you say, 0-9 (10 possibilities). However, some people might not like that (see e.g. bruto's post above).


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[Innocuous]

Newcomers in the thread, please go to ...to find the opening post of this test.

You must precisely define your criteria for success and failure - that is the essence of testing the power you claim to have. You have to do this before the test begins. After-the-fact rationalizations are not tolerated.

 

[AdMan]

Ah that's unfortunate, Innocuous. However, I don't rule out doing some day perhaps a test with a larger number of possibilities, perhaps 1, 2, 3, 4, 5 or even perhaps, like you say, 0-9 (10 possibilities).

Get rid of your subjective "credibility ratings" too and you might be getting closer to a valid test.

 

[Hokulele]

Let me use an example again.

Member NiceAndSmartGuy, answers:
I am almost completely sure the (numerical) answer to this interesting test is a xx. (with a MD5 hash).

Then, member BadMoon answers:
Dis is a compleitly horrible "test". It's obvius dat tis person dosn't now how to conduct a serius research. I anser xx. (with a MD5 hash also).

Then, I would regard the first answer as credible, and the second one as not-credible. I cannot, in this test, "cheat" (generally speaking, I would try to avoid it anyway) by favoring (by means of a high credibility rating) the member who has given the right numerical answer because I do not see the numerical answer. Then, I seem to have found before that credible members tend to provide more accurate numerical answers.

Newcomers in the thread, please go to post #1031 to find the opening post of this test.

This is completely and demonstrably untrue. In the last round of the test, the answers were blinded and handed to a third party. As expected, the percentage of participants who chose the correct number was very close to 25%. In addition, your credibility rankings did not correlate with the correct answers in any meaningful sense, and since your credibility ratings were never below zero, technically every answer was credible, even if they were wrong.

Once you realized that you failed to pass your test, you changed your methodology and began throwing away several answers, claiming after the reveal that they were not proper ones. Even after this pathetic attempt at salvaging some sort of "win" out of the mess, there was no clear correlation, so you basically ended up saying something along the lines of, "It is good enough for me."

In other words, even if we made the credibility rankings sound, your methodology is just about the antithesis of the scientific method.

 

[Tomtomkent]

Ah that's unfortunate, Innocuous. However, I don't rule out doing some day perhaps a test with a larger number of possibilities, perhaps 1, 2, 3, 4, 5 or even perhaps, like you say, 0-9 (10 possibilities).

So still a 20% or 10% chance of finding the right answer by luck.

Surely you would want to prove that the ONLY way of getting the right answer was through telepathy, which would require making the random chance as small as possible.

At least SIX digits from 0-9 would be the minimum. Not a single digit from a guessable range. You need to show better than, not equal to, random chance.

 

[Michel H]

So still a 20% or 10% chance of finding the right answer by luck.

Surely you would want to prove that the ONLY way of getting the right answer was through telepathy, which would require making the random chance as small as possible.

At least SIX digits from 0-9 would be the minimum. Not a single digit from a guessable range. You need to show better than, not equal to, random chance.

Keep in mind, Tom, that, in a normal unmerged test thread, several (more than just one) give numerical answers. Then, there are tools of statistical analysis available to assess the significance of the results.


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[Loss Leader]

I am sorry if this test seems a little complicated but you have to realize that, if you want a rigorous procedure, there is a (little) price to pay, in terms of complexity. I thought it would be appropriate to do a rigorous test on the forum of the Randi Foundation, which tends to emphasize possible errors, the risk of erroneous beliefs, and critical thinking. A rigorous test which, I should say, is partly based on suggestions you have made yourself, on this very forum (regarding "blinding" and MD5 hashes)

You're saying the word "rigorous" but I think the word you're searching for is "nonsense."

 

[bruto]

I'm reminded of a long ago quotation, context long forgotten, but one that often becomes germane, in which a person is said to have mistaken precision for accuracy.

 

[Kid Eager]

Michel,
You have been told multiple times that your methodology is flawed and why it is flawed.

Consider also that there are people on this forum that know a great deal more about probability and testing methodologies than you.

You are not the subject matter expert here. Insistence on a flawed methodology in light of the above indicates you have no real confidence in the results of a valid test supporting your claim.

Move on with your life, if you are unable to get over this fundamental mental hurdle, and do something else.

No, because, with the protocol I use in this rigorous test, once I finally see the numerical answers, I can no longer change the credibility ratings.


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

Please note: your *methodology* is not credible, no matter how rigorously you apply it.

Using a range of only four numbers and a limited number of responses, you will not get an even distribution; you will get a pseudo-RANDOM distribution, with some numbers being chosen more frequently than others.

This is in no way evidence of telepathy, but randomness combined with chooser bias.

 

[Michel H]

Please note: your *methodology* is not credible, no matter how rigorously you apply it.

Using a range of only four numbers and a limited number of responses, you will not get an even distribution; you will get a pseudo-RANDOM distribution, with some numbers being chosen more frequently than others.

This is in no way evidence of telepathy, but randomness combined with chooser bias.

Even if people tend to favor certain numbers (however, I have been doing online telepathy tests for several years, and I generally didn't find that this was happening), this will not lead to "false telepathy signals", because the target number is also random. For example, if people always answer 3, and the probability distribution of the target is uniform, the correct answer rate will remain equal to 25%.


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[Michel H]

...
Also, as a social scientist, I can tell you that instruments measuring people's opinions need to be devised, scored and normalized beforehand. You can't just say you'll evaluate credibility by giving it a score. You have to specifically define how the score is to be arrived at.
...

You're saying the word "rigorous" but I think the word you're searching for is "nonsense."

If I was looking for the word "nonsense", I don't think I would have to look for it for very long, because I think I have already seen it in your posts in this thread.

I have explained before when and why I find an answer not-credible:

...Credible answers (i.e. those with a positive CR) are ideally kind and smart, while non-credible answers (which get a negative CR), are those which do not sound serious, sincere and reliable. Reasons why an answer may not be credible are:
• It is too aggressive.
• It contains one or several incorrect statements.
• It is odd or bizarre.
• The answerer says that his/her answer is not related to telepathy.
• The answer contains a very large number of spelling or syntaxic errors.
...

If there is no telepathy, and if credibilities are assigned in a "blinded way", like in this current test, the hit rate should be equal to about 25%, within the subset of answers which seem credible to me (even if I am a little subjective in assigning credibilities). If it is not, this might indicate telepathy.


Newcomers in the thread, please go to post #1031 to find the opening post of this test.

 

[AdMan]

If I was looking for the word "nonsense", I don't think I would have to look for it for very long, because I think I have already seen it in your posts in this thread.

I have explained before when and why I find an answer not-credible:

If there is no telepathy, and if credibilities are assigned in a "blinded way", like in this current test, the hit rate should be equal to about 25%, within the subset of answers which seem credible to me (even if I am a little subjective in assigning credibilities). If it is not, this might indicate telepathy.

I still cannot understand how you can keep defending your "credibility rating" nonsense after it's been explained to you multiple times that such a thing has no place in any kind of supposedly scientific test.

In a previous test, you even assigned a high "CR" to a post that was very obviously mocking you and/or the test, because you didn't realize that.

Your tests are a joke.