Telepathy

[Dave Rogers]

Observational Error

But in this specific instance, what would be the probability of me getting even one card wrong? I would think it very low in a sample of only five. Five hundred cards right, agreed, would be unreasonable, but the probability of one error in five is too small to be worth considering.

Try it if you want. Post your hash, PM me the sequence, I'll post it, and see if I get any wrong. If you send me the right sequence, in practice I'll get it right.

Dave

 

[Loss Leader]

Try it if you want. Post your hash, PM me the sequence, I'll post it, and see if I get any wrong. If you send me the right sequence, in practice I'll get it right.

What would we be measuring? If it's your ability to psychically know the cards, that's defeated by the fact that I'm telling you the cards.

It seems to be a measurement of the ability to transmit data digitally. But it's not a good test of that. It seems like a really poor test. Devise a proper test and there will be measurement error.

 

[Dave Rogers]

What would we be measuring?

The effectiveness of a means of communication.

It seems to be a measurement of the ability to transmit data digitally. But it's not a good test of that. It seems like a really poor test. Devise a proper test and there will be measurement error.

Yes, it's a really poor test, because it's trivially easy to pass. That's the point; if telepathy can't even pass such a trivially easy test, it's strongly suggestive that it doesn't exist.

Dave

 

[abaddon]

What would we be measuring? If it's your ability to psychically know the cards, that's defeated by the fact that I'm telling you the cards.

But that is exactly what telepathy claims to do. I could tell you which five cards there were in person, by phone, SMS, Facebook, viber, Skype and so on and so forth. You would score 100%.

Compare telepathy. What would that method score in comparison?

It seems to be a measurement of the ability to transmit data digitally. But it's not a good test of that. It seems like a really poor test. Devise a proper test and there will be measurement error.

It is at least good enough to determine if telepathy is worthy of further investigation. We have two true believers in telepathy and when even a prelim test is facilitated right here, both vanish. I wonder why that is?

 

[Loss Leader]

That's the point; if telepathy can't even pass such a trivially easy test, it's strongly suggestive that it doesn't exist.

I'm not a scientist. I'm not a mathematician or a statistician. I do believe that measurement error precludes perfect results. Someone with a stronger background in writing scientifically rigorous tests would need to weigh in.

None of this seems to matter because KotA hasn't been back to finalize any of this.

 

[Spektator]

I would suggest one thing: Don't respond to KotA's ID's of cards one at a time. Wait until the whole group, five, ten, whatever, is in.

 

[jt512]

I had suggested picking all 52 cards from which he would have to get 5 right.

As Startz showed, 5 cards is too lenient. There is a 1/270 chance of getting 5 or more right by pure guessing. If we gave the test to 1000 people, we'd expect about 4 of them to get at least 5 cards correct. The purpose of the test is to indicate preliminary evidence of telepathy. I would not believe that 4 people in a random sample of 1000 who guessed at least 5 cards correctly would be showing preliminary evidence of telepathy. I'd believe they got lucky.

 

[Startz]

As Startz showed, 5 cards is too lenient. There is a 1/270 chance of getting 5 or more right by pure guessing. If we gave the test to 1000 people, we'd expect about 4 of them to get at least 5 cards correct. The purpose of the test is to indicate preliminary evidence of telepathy. I would not believe that 4 people in a random sample of 1000 who guessed at least 5 cards correctly would be showing preliminary evidence of telepathy. I'd believe they got lucky.

I think Loss Leaders current proposal, which I thought KotA was mostly okay with, is to get 5 cards out of 5 dealt from 52 right. That's very different statistically from dealing 52 and getting 5 right. Can't happen by chance.

Anyhow, I do hope King comes back and engages in a friendly trial. If the evidence looked good, then lots of the suggestions folks have made here could be taken into account in designing something more rigorous.

 

[jt512]

I think Loss Leaders current proposal, which I thought KotA was mostly okay with, is to get 5 cards out of 5 dealt from 52 right. That's very different statistically from dealing 52 and getting 5 right. Can't happen by chance.

My understanding is that the experiment that KofA said he liked was to get at least 10 cards correct out of a sequence of 52. That would be extremely improbable to achieve by chance. Loss then proposed doing the same test but only requiring 5 correct, but he grossly understated the probability that would happen by chance.

I should mention that these closed-deck experiments, which are what statistically untrained people think of first, are not well suited for the intended purpose, because it is difficult to impossible to calculate the false negative probability of the test, and hence it's hard to know whether the test is fair to the testee. Much better are experiments that use independent trials, where the false negative rate is easy to calculate and the number of trials comprising the test can be chosen according to the desired false negative and false positive rate. If you insist on using a deck of cards, I would recommend replacing the drawn card after each trial and reshuffling the deck between trails. A simpler procedure, which is what most most "serious" parapsychologists have settled on is to use a much smaller deck comprising, say, just 5 cards, with the number of trials and number of required correct guesses chosen to give the desired error rates.

If you do insist on using a closed-deck design, I would ask the testee what false negative rate he thinks is acceptable. Say he says 5%. Then ask him how many cards k out of 52 he has a 95% chance of getting correct, and assuming that k is at least 6 or 7, call the test a success if he gets at least k correct. If he states that k < 6 (or 7 maybe), then he would have to pass more than one iteration of the test for the test to be considered a success.

 

[Startz]

I should mention that these closed-deck experiments, which are what statistically untrained people think of first, are not well suited for the intended purpose, because it is difficult to impossible to calculate the false negative probability of the test, and hence it's hard to know whether the test is fair to the testee.

Could you explain that a little more? (I believe you, I just don't understand why it's so.)

 

[Loss Leader]

If you do insist on using a closed-deck design, I would ask the testee what false negative rate he thinks is acceptable. Say he says 5%. Then ask him how many cards k out of 52 he has a 95% chance of getting correct, and assuming that k is at least 6 or 7, call the test a success if he gets at least k correct. If he states that k < 6 (or 7 maybe), then he would have to pass more than one iteration of the test for the test to be considered a success.

Fine by me. 7 cards is great. I chose 5 because it was well within what KotA said he was capable of. But I didn't realize the odds were so good.

 

[Oystein]

I'm not a scientist. I'm not a mathematician or a statistician. I do believe that measurement error precludes perfect results. Someone with a stronger background in writing scientifically rigorous tests would need to weigh in.

None of this seems to matter because KotA hasn't been back to finalize any of this.

Measurement error in telephoning playing card? You think Dave Rogers might report a Nine Point One of Spades?

Sent from mobile phone through Tapatalk

 

[Dave Rogers]

Measurement error in telephoning playing card? You think Dave Rogers might report a Nine Point One of Spades?

It's not impossible. I used to have the Usenet signature "Pentium designers support the 48.999973524'ers."

Dave

 

[Pixel42]

None of this seems to matter because KotA hasn't been back to finalize any of this.

If only he were telepathic, then he'd know he needed to revisit the thread.

 

[jrhowell]

Measurement error in telephoning playing card? You think Dave Rogers might report a Nine Point One of Spades?

It seems to me that it would be easy to confuse "ace" with "eight" over the phone, depending on the call quality, speaker's accent, etc.

 

[Loss Leader]

Measurement error in telephoning playing card? You think Dave Rogers might report a Nine Point One of Spades?

What if he misdials the telephone? Then none of the cards come through. Misdialing is an error that can't be eliminated. Over enough iterations, the test will not return results of 100%.

Let's not get too far off track. All we're doing now is asking KotA what he thinks he can do at what level of confidence. Then we can design a test around that. After all, only hearing a tenth of people's thoughts would still be a staggering breakthrough for science.

 

[abaddon]

It seems to me that it would be easy to confuse "ace" with "eight" over the phone, depending on the call quality, speaker's accent, etc.

Use the Nato alphabet, problem disappears.

 

[Oystein]

If only he were telepathic, then he'd know he needed to revisit the thread.

Since he is telepathic, he knows better than to revisit the thread even without revisiting the thread!

Sent from mobile phone through Tapatalk

 

[Oystein]

Use the Nato alphabet, problem disappears.

Oh how catastrophically wrong you are!!

True story: I am German, but my last name is unusual, it appears French, is actually derived from some Walloon dialect, and even my direct relatives pronounce it differently depending on whether they live where it originated, or 20 miles east, or 20 miles west. There is no American-English pronunciation that would do it justice.
In the mid-90s, I had to book a hotel room in the USA, calling from Germany. It was clear that I needed to spell my name: M ... A ... R ... A ... "Was that an A or an R?" The receptionist would interrupt me. Uhm well, both... The phone line was of bad quality, half a second delay and a bit of echo. Several attempts at spelling got wasted. So I turned to NATO alphabet: Mike ... Alpha ... Romeo ... Alpha...

Two months later, I check into the hotel. To avoid confusion, I give them my credit card immediately and point out the name.

No reservation to that name. Hotel fully booked. Narf.

It turned out they had a reservation for a certain Mr. Alfaromeo, Mike.

:ld:

Sent from mobile phone through Tapatalk

 

[Pope130]

Oh how catastrophically wrong you are!!

True story: I am German, but my last name is unusual, it appears French, is actually derived from some Walloon dialect, and even my direct relatives pronounce it differently depending on whether they live where it originated, or 20 miles east, or 20 miles west. There is no American-English pronunciation that would do it justice.
In the mid-90s, I had to book a hotel room in the USA, calling from Germany. It was clear that I needed to spell my name: M ... A ... R ... A ... "Was that an A or an R?" The receptionist would interrupt me. Uhm well, both... The phone line was of bad quality, half a second delay and a bit of echo. Several attempts at spelling got wasted. So I turned to NATO alphabet: Mike ... Alpha ... Romeo ... Alpha...

Two months later, I check into the hotel. To avoid confusion, I give them my credit card immediately and point out the name.

No reservation to that name. Hotel fully booked. Narf.

It turned out they had a reservation for a certain Mr. Alfaromeo, Mike.

:ld:

Sent from mobile phone through Tapatalk

This points out one of the big unanswered problems with claims of telepathy: Not only must you have some channel of communication, but also, both parties must be using the same symbol set.