Information on Protocol for Pavel Ziborov Applicant

Pavel,

I am still thinking of ways we could potentially try jackalgirl's idea.

I understand that her suggested protocol would be like a roulette wheel. Red comes up four times in a row, and you start thinking you should bet on black because eventually it will turn. But of course, technically, black would never HAVE to come up.

What if the protocol was designed so the answer didn't have to be a "final" answer at the moment you looked at it?

For instance, what if you were given a stack of twenty envelopes (14 would contain the same photograph, 6 would contain blank photo paper) and you were permitted an hour, or even an hour and a half, to sort through the photos as many times as you wanted and divide them into piles of photographs and piles of photo paper? We could not open the envelopes as you went, but you could hold each one as many times as you wanted to ensure that you were not seeing "photo" just because you had gotten five "blank" in a row.

We could even number the envelopes and give you a piece of paper to make notes - so you could write "I am positive" next to number 5, and "I am not sure" next to number 16 and then later go back to 16 until you *were* sure.

Can you try self-testing this way? Have someone else stuff the envelopes and give it a shot.

-- Remie
 
RemieV said:
Pavel,

I am still thinking of ways we could potentially try jackalgirl's idea.

I understand that her suggested protocol would be like a roulette wheel. Red comes up four times in a row, and you start thinking you should bet on black because eventually it will turn. But of course, technically, black would never HAVE to come up.

What if the protocol was designed so the answer didn't have to be a "final" answer at the moment you looked at it?

For instance, what if you were given a stack of twenty envelopes (14 would contain the same photograph, 6 would contain blank photo paper) and you were permitted an hour, or even an hour and a half, to sort through the photos as many times as you wanted and divide them into piles of photographs and piles of photo paper? We could not open the envelopes as you went, but you could hold each one as many times as you wanted to ensure that you were not seeing "photo" just because you had gotten five "blank" in a row.

We could even number the envelopes and give you a piece of paper to make notes - so you could write "I am positive" next to number 5, and "I am not sure" next to number 16 and then later go back to 16 until you *were* sure.

Can you try self-testing this way? Have someone else stuff the envelopes and give it a shot.

-- Remie
Click to expand...


I understand what you mean.. it some kind of Carina Landing protocol.. ( that was suggested many times by GzuzKryzt)
Maybe if there would be 20 photos with let say for instance.. 10 photos of Planet Earth, and 10 Egyptian pyramids. They will be all mixed up and I will be given time to sort them out in 2 pials, in the way that you described.. still would it work with 16 out of 20 correct?

It is possibly could work.. I mean I can and I will try it, just have to print out new set for test.
 
Last edited:
RemieV said:
Pavel,

I am still thinking of ways we could potentially try jackalgirl's idea.

I understand that her suggested protocol would be like a roulette wheel. Red comes up four times in a row, and you start thinking you should bet on black because eventually it will turn. But of course, technically, black would never HAVE to come up.

What if the protocol was designed so the answer didn't have to be a "final" answer at the moment you looked at it?

For instance, what if you were given a stack of twenty envelopes (14 would contain the same photograph, 6 would contain blank photo paper) and you were permitted an hour, or even an hour and a half, to sort through the photos as many times as you wanted and divide them into piles of photographs and piles of photo paper? We could not open the envelopes as you went, but you could hold each one as many times as you wanted to ensure that you were not seeing "photo" just because you had gotten five "blank" in a row.

We could even number the envelopes and give you a piece of paper to make notes - so you could write "I am positive" next to number 5, and "I am not sure" next to number 16 and then later go back to 16 until you *were* sure.

Can you try self-testing this way? Have someone else stuff the envelopes and give it a shot.

-- Remie
Click to expand...

pavel_do said:
I understand what you mean.. it some kind of Carina Landing protocol.. ( that was suggested many times by GzuzKryzt)
Maybe if there would be 20 photos with let say for instance.. 10 photos of Planet Earth, and 10 Egyptian pyramids. They will be all mixed up and I will be given time to sort them out in 2 pials, in the way that you described.. still would it work with 16 out of 20 correct?

It is possibly could work.. I mean I can and I will try it, just have to print out new set for test.
Click to expand...

Pavel:

I recommend that you first try exactly what Remie suggests. This will be the fastest way to reach an agreement. That's 6 of one kind and 14 of the other, not 10 and 10.

And I think her suggestion let's you spend a long time before making a final decision, but does expect you to correctly separate all the photos into the correct piles.
 
Startz said:
Pavel:

I recommend that you first try exactly what Remie suggests. This will be the fastest way to reach an agreement. That's 6 of one kind and 14 of the other, not 10 and 10.

And I think her suggestion let's you spend a long time before making a final decision, but does expect you to correctly separate all the photos into the correct piles.
Click to expand...

I will try to do so.. need to order prints, will do this all over weekend.
 
Pavel,

Sixteen is a little less than what you'd need to pass the 1/1000 odds, but I am willing to write it into the protocol that way and have Randi review it, as it isn't very far off.

Shall we proceed with this protocol?

-- Remie
 
RemieV said:
Pavel,

Sixteen is a little less than what you'd need to pass the 1/1000 odds, but I am willing to write it into the protocol that way and have Randi review it, as it isn't very far off.

Shall we proceed with this protocol?

-- Remie
Click to expand...

Pavel:

This looks to be a very generous protocol.

For those into the statistical side, I wrote a quick simulation program in Matlab which gives the chance of success by random guessing as 0.011. I haven't checked the code, so it shouldn't be relied on. But here it is for anyone interested.

Code: 
function p = remieSuggestion(minNumRight, category1size, nSim)
%{
Dick Startz
July 2008

there are 20 cards, category1size of type 1 and 2-category1size of type 2
How many times can you guess numRight or more right

We'll do this my randomly sorting numbers and seeing how many end up in
each group
%}
rand('twister',5489);
correct = 0;

for iSim=1:nSim
    cardOrder = randperm(20);
    numCorrect = sum(cardOrder(1:category1size)<=category1size)...
        +sum(cardOrder(category1size+1:end)>=category1size+1);
    if numCorrect >= minNumRight
        correct = correct + 1;
    end
end
p = correct/nSim;
end
 
Last edited:
Gr8wight said:
The problem we are running into here is that Pavel's powers are only apparent to him when he complicates the protocol to the point were he loses track of the statistical probabilities involved. Simplifying and better controlling the test results in lower accuracy from Pavel, so he resists that. This is no different than any others of the multitudinous applicants the JREF has dealt with over the years. Pavel needs to ask himself one question, and think very hard about the answer. Given two envelopes, one containing a photograph, and one containing a blank sheet of photographic paper, can he identify the photograph significantly more often than half the time. If not, why not?

I have absolutely no interest in the answer to that question, only in the knowledge that Pavel has asked it of himself, and tried to be honest with himself about the answer.
Click to expand...

pavel_do said:
1. YES I CAN DO THAT significantly more often than half the time.. It is not complicated at all when there is only 2 envelopes to identify. Cause out of two I will pick one that I am getting image and most sure of.. Though sometimes I don’t see photo as I said still I would be able to hold both and compare my fillings and even slight visions. Just that the black photo has to be black. No colors or shapes or anything.

2. I am absolutely disagree with you in sense that I complicate test etc.. what can be easier than test with pairs? Too complicated? Or one out of 3? Or 2 out of 5? For you 3 out of 20 is easier than 3 out of 5? Ok.. than if 2 out of 5 everyone can do.. good luck to all of them but so far no evidence…
Click to expand...

Perfect. This will be easy. All you need is ten identical photographs, and ten completely blank photographs. Each photo is put into an envelope. You will be handed the envelopes two at a time, and each pair of envelopes will consist of one photograph and one blank sheet. You simply need indicate which envelope holds the photograph. Those two envelopes will be marked and put aside. Repeat the process with two new envelopes. After you have chosen the envelope you believe contains the photo from all ten pairs, all of the envelopes will be opened together, and you can see how accurate you were.

What do you think?
 
Howdy! I think that Gr8wight has it.

Each attempt consists of one set of 1 piece of blank photo paper and 1 photo. You have a 1:2 chance (50%) of determining the correct one by chance alone.

But you repeat this 10 times. I think that the odds of getting all 10 correct (by chance alone) are 1:1024*. You'd have to be successful on each and every individual attempt.

How does that sound? You wouldn't have to make piles or do any sorting or worry about multiple blanks (or multiple photos) throwing you off. You could even wait in between attempts to "reset" your senses. Doable?

*I am statistically stupid. Could someone tell me if I'm doing this correctly? I'm multiplying 1/2 by 1/2 ... 10 times (1 / 2^10) (I'm not even sure if I'm doing /that/ correctly). Is that right?

Edited to add: Many thanks to Ravenwood, who if I am not mistaken, is the one who originally asked if Pavel can tell between a blank photo and a photo with an image in it.
 
Last edited:
Jackalgirl said:
Howdy! I think that Gr8wight has it.

Each attempt consists of one set of 1 piece of blank photo paper and 1 photo. You have a 1:2 chance (50%) of determining the correct one by chance alone.

But you repeat this 10 times. I think that the odds of getting all 10 correct (by chance alone) are 1:1024*. You'd have to be successful on each and every individual attempt.

How does that sound? You wouldn't have to make piles or do any sorting or worry about multiple blanks (or multiple photos) throwing you off. You could even wait in between attempts to "reset" your senses. Doable?

*I am statistically stupid. Could someone tell me if I'm doing this correctly? I'm multiplying 1/2 by 1/2 ... 10 times (1 / 2^10) (I'm not even sure if I'm doing /that/ correctly). Is that right?

Edited to add: Many thanks to Ravenwood, who if I am not mistaken, is the one who originally asked if Pavel can tell between a blank photo and a photo with an image in it.
Click to expand...

Well, this dude also suggested it. ;)
 
RemieV said:
Pavel,

Sixteen is a little less than what you'd need to pass the 1/1000 odds, but I am willing to write it into the protocol that way and have Randi review it, as it isn't very far off.

Shall we proceed with this protocol?

-- Remie
Click to expand...

Thank you Remiev. I will print today a new set of photos. 10 and 10. and will test myself in this way during a weekend to see how will it go. And by the end of the Sunday, I will post my reply.

P.s.
Re to the other posts.. Gr8wight, Jackalgirl, and dude who also suggested it..:) thank you to, I will try it all over the weekend.
 
Last edited:
Jackalgirl said:
Howdy! I think that Gr8wight has it.

Each attempt consists of one set of 1 piece of blank photo paper and 1 photo. You have a 1:2 chance (50%) of determining the correct one by chance alone.

But you repeat this 10 times. I think that the odds of getting all 10 correct (by chance alone) are 1:1024*. You'd have to be successful on each and every individual attempt.

How does that sound? You wouldn't have to make piles or do any sorting or worry about multiple blanks (or multiple photos) throwing you off. You could even wait in between attempts to "reset" your senses. Doable?

*I am statistically stupid. Could someone tell me if I'm doing this correctly? I'm multiplying 1/2 by 1/2 ... 10 times (1 / 2^10) (I'm not even sure if I'm doing /that/ correctly). Is that right?

Edited to add: Many thanks to Ravenwood, who if I am not mistaken, is the one who originally asked if Pavel can tell between a blank photo and a photo with an image in it.
Click to expand...

The statistics are correct. But this is a much easier test than having all the photos together and sorting them into two groups, which is what Remie suggested. The difference is that doing it this way if you get one envelope in an attempt right you automatically get the other one right, where in the "big pile" approach you have to get the first 19 right and then the last one is automatic. Basically the odds in Remie's method are 1 in a million for perfection. That's what lets Pavel make some errors and still win.

So Pavel, since you'll have the photos it's a great idea to practice on this suggestion, but make sure you also try out Remie's suggestions exactly.
 
Jackel girl and gr8 wight you may not have read all the other long long long thread but Pavel already did this same test you are describing for his affidavit from 10 pairs of photos which he knew what 2 photos there were out of 5 trials he guessed 7,7,7,8 and 4 right out of 10.

As Jref requires 1:1000 odds i.e. 10 out of 10 he failed the preliminary test 5 times already.


Only diff now is there will be one blank and one photo, Not 2 known photos but still 50% chance to guess just the same..

regards Gmonster
 
Gmonster2 said:
Jackel girl and gr8 wight you may not have read all the other long long long thread but Pavel already did this same test you are describing for his affidavit from 10 pairs of photos which he knew what 2 photos there were out of 5 trials he guessed 7,7,7,8 and 4 right out of 10.

As Jref requires 1:1000 odds i.e. 10 out of 10 he failed the preliminary test 5 times already.


Only diff now is there will be one blank and one photo, Not 2 known photos but still 50% chance to guess just the same..

regards Gmonster
Click to expand...

Pavel has always said that his talent is imperfect, so it's not quite fair to ask for a test that requires perfection. The suggestion Remie has made, following ideas suggested here on the forum, presents a much higher bar for perfection and therefore allows some room for error.
 
I made prints of photos that I need, including a few just blank-black photographs..

Will try to have 3 types of test, by 3 trials each.
1. When all 20 ( 10 each of a kind) will be mixed up and I will try to separate them in 2 pails…
2. When I will have them separated by pairs, where I need to identify one of two(as the second one will be known as soon as first will be named) I will do it with opening every photo that was named as I usually do.
3. Test where I have to identify pairs.. same as in N.2 but the results will be unknown till the end of the test and last pair is identified.
4. Will make a few pairs with 1 photo and 1 blank-black and try it too.



p.s. Re to Gmonster2 comment

if you can get same results as I did for affidavit letter.. Good luck to you.. it still over chance..
Any ways when we ran test at psychology department. I was given 1 photo of two to identify.. Now I have both photos in my hands to choose the one I am sure most of.. that make difference for me.. Beside at that time all the pairs was different.. and this time I will use only photos that I am usually getting most of the times correct.. that is also make difference for me..
 
pavel_do said:
I made prints of photos that I need, including a few just blank-black photographs..

Will try to have 3 types of test, by 3 trials each.
1. When all 20 ( 10 each of a kind) will be mixed up and I will try to separate them in 2 pails…
2. When I will have them separated by pairs, where I need to identify one of two(as the second one will be known as soon as first will be named) I will do it with opening every photo that was named as I usually do.
3. Test where I have to identify pairs.. same as in N.2 but the results will be unknown till the end of the test and last pair is identified.
4. Will make a few pairs with 1 photo and 1 blank-black and try it too.



p.s. Re to Gmonster2 comment

if you can get same results as I did for affidavit letter.. Good luck to you.. it still over chance..
Any ways when we ran test at psychology department. I was given 1 photo of two to identify.. Now I have both photos in my hands to choose the one I am sure most of.. that make difference for me.. Beside at that time all the pairs was different.. and this time I will use only photos that I am usually getting most of the times correct.. that is also make difference for me..
Click to expand...

Pavel:

This is excellent. Very few applicants have been as willing to follow up on suggestions as you have. We'll look forward to seeing the results of these informal tests and also to the results of the official test when it comes about.
 
by the way just wanted to show you the photos that I am usually most sure of and the ones I will use for the testing this time.
it is on this link.. I posted them here..

www.jaimypage.narod.ru

the photos was redone by me in photo shop, I tried to make them as different and as "bright" and significant as I could.. Like I can “get them” by shapes or colors.. Like pyramids it is either Pyramids shapes, pointed up sharp corners or something of a kind.. or red colors that I pick up.. The earth is Blue and round.. The Ship is in ship shape and gray colors.. and The cross is usually I pick up either as crossed lines or black shadow of the cross etc

p.s.
Sorry about the size of the photos..:) as on the photo paper they all fit well :redface1
 
Last edited:
pavel_do said:
1. YES I CAN DO THAT significantly more often than half the time.. It is not complicated at all when there is only 2 envelopes to identify. Cause out of two I will pick one that I am getting image and most sure of.. Though sometimes I don’t see photo as I said still I would be able to hold both and compare my fillings and even slight visions. Just that the black photo has to be black. No colors or shapes or anything.
Click to expand...
Pavel, please be aware that blank photographic paper is white, not black. Apart from white, there will be no colours, shapes or anything, just like you stipulate.
 
Pavel, if you are printing these out on an inkjet or laser printer, there is no difference between regular or photo paper. The toner or ink is being deposited on the paper by the same mechanism, unlike printing a photographic negative on photographic paper which is a completely different process... That being said, if you are using a printer, there should be no discerrnable difference in size or weight of a blank sheet vs a printed sheet of the media you are using once they are in the envelopes.
 
Ravenwood said:
Pavel, if you are printing these out on an inkjet or laser printer, there is no difference between regular or photo paper. The toner or ink is being deposited on the paper by the same mechanism, unlike printing a photographic negative on photographic paper which is a completely different process... That being said, if you are using a printer, there should be no discerrnable difference in size or weight of a blank sheet vs a printed sheet of the media you are using once they are in the envelopes.
Click to expand...


NO I print them on the photo paper in the Photo lab. As photo copies or photos printed on inkjet printer.. I don’t see them the same as photos on the photo paper.

Re to Steenkh.
The photos I’ve done as blank.. are BLACK not plane white.. as I would mix it up with Photo of Titanic that is Gray-light or Cross that has allot of white on photo..
 
Last edited:
You must log in or register to reply here.

Back
Top Bottom