Merged New telepathy test: which number did I write ?

[abaddon]

I think I will put my DVD of Groundhog Day on.

I think I will put my DVD of Groundhog Day on.

 

[Ladewig]

Why doing a test on this forum? Let's see... First, I have received good answers in my tests on this forum, to which you have yourself contributed. Second, a success on this forum of an Educational Foundation would perhaps entail more recognition by human Society than a possible success on a smaller, lesser known forum. Third, I have a scientific background, so perhaps I feel perhaps, sometimes, more at ease on this forum, of a Foundation which officially tries to promote good Science (whether it promotes it really is perhaps less sure and obvious lol), than I would, say, on a religious or a parapsychology forum for example.
And I want to add that I do frequent other forums, which may be very good sometimes. I have already done online telepathy tests in French, English, Dutch and German (1 test in German).

Fascinating. Simply fascinating.

.........

Again, please consider the possibility that a doctor visit might prove fruitful.

 

[xtifr]

With a larger sample space, you wouldn't need to go through all this ridiculous rigmarole to try to detect supposed liars, because any correct hits would already be beating such long odds.

Lets assume eight people respond. Three have fooled themselves into thinking they can receive you, and their answers are wrong (we'll assume for simplicity). Five can hear you, but three decide to lie about it. So that gives you two guaranteed hits, if telepathy is real.

If the sample space is 1-4, two hits is what you should expect if everyone chose randomly, so you haven't really proven anything. All you get is endless debates about whether you can throw out answers you don't like.

If the sample space is 1-100, two hits is amazing! One hit would be unlikely, two would be astounding. So the liars no longer ruin the experiment. If you tried 1-1000, the odds against two correct guesses would be astronomical, unless telepathy were actually involved. And that's all without throwing out any answers!

Leaving the lies might underestimate the amount of telepathy going on, but those results would still make people sit up and take notice, and suddenly start taking you a whole lot more seriously!)

So why do you keep trying this over and over with 1-4? Are you afraid of a meaningful test you can't manipulate by your subjective "throwing out answers I don't like" trick? Do you think maybe there's a small part of you that simply afraid to find out you might be wrong? Man up and give us a real test.

In the mean time, I'm going to sit this one out, just like I did with the last two horribly-designed tests. But if this one used a larger range (at least 1-100, better 1-1000), I would happily try my best.

 

[SezMe]

I don't know why there were two very different, very contrasting consecutive answers. Perhaps you know better than me. Normally, in such a case (several answers), I retain the last one, which is unfortunately the worse in this case. However, the MD5 hashes appear to be incorrect, and therefore the answers are not valid so far, and this offers some protection against having to take the last one for the analysis of the results. I hope the member (SezMe) will clarify this situation.

As ComfySlippers, abaddon, Tomtomkent and others noted, that you cannot see the implications of my two posts is clear indication that you cannot see the OBVIOUS flaw in your "test". Here, let me spell it out.

I gave you two identical (pseu)data points posted with two completely different (and fictitious) attitudes. Thus, showing that when you take one datum and not the other based on my made-up frame of mind, you are engaging in a completely arbitrary, capricious data filtering process that TOTALLY invalidates your approach to telepathy.

 

[Innocuous]

With a larger sample space, you wouldn't need to go through all this ridiculous rigmarole to try to detect supposed liars, because any correct hits would already be beating such long odds.

An excellent point. Even if 90% of the responses are "lies", getting only 10% right in a range of 0-1000 would be one hundred times better than chance! With 99% lies, a 1% success rate is still ten times better than chance.

I have been thinking of other ways to remediate the current test, assuming the insistence of the limited range 1-4 (which troubles me) and the elimination of "lies" (which doesn't bother me as much as some of the other posters). I think it could work if there were repeated trials with random numbers (Michel H does NOT get to pick the number he broadcasts). As long as the number of accepted non-lying responses is high enough, it doesn't matter how many he rejects for any reason - he can't skew the results in his favor.

But I will say it again - the criteria for success or failure must be established before the test is done.

 

[bruto]

Indeed, among the various faults of the program here, it seems superfluous that everyone is being asked to guess Michel's number, since (as I think Iamme has also pointed out) that would suggest that we, rather than Michel, are telepathic.

If the test is for Michel's capacity, why not have an independent party choose an unknown number? Respondents here could then make a choice which would be recorded to avoid the possibility of dishonesty on their part, and a mind reader would determine what that choice was. Sincerity or reliability would be entirely irrelevant, as numbers not in the range would simply be disallowed, and any legitimate number would be as chosen as the next, no matter why.

I am pretty sure this idea has already occurred at some time in this long and tedious thread, and of course that test would be simple and not easy to cheat, so I presume elaborate reasons will be found for why it would not work.

 

[Michel H]

I object to this test on grounds that MD5 is not sufficiently secure for this application

The security of the MD5 hash function is severely compromised. A collision attack exists that can find collisions within seconds on a computer with a 2.6 GHz Pentium 4 processor (complexity of 224.1).[25] Further, there is also a chosen-prefix collision attack that can produce a collision for two inputs with specified prefixes within hours, using off-the-shelf computing hardware (complexity 239).[26] The ability to find collisions has been greatly aided by the use of off-the-shelf GPUs. On an NVIDIA GeForce 8400GS graphics processor, 16–18 million hashes per second can be computed. An NVIDIA GeForce 8800 Ultra can calculate more than 200 million hashes per second.[27]

These hash and collision attacks have been demonstrated in the public in various situations, including colliding document files[28][29] and digital certificates.[42]

My number is $$

sha1: dc477d742bea9f12ce091e224dba423bf6fe01a0

...
Thank you for your answer, Snorkio. You did not provide a MD5 hash, as requested in the opening post, but I think I will probably accept your answer, as you seem to have concerns about the security of this test, that we should strive to make as high as we reasonably can.
...

...
No, no, a thousand times NO! This is exactly the sort of dishonest, unscientific nonsense that makes all of your tests worthless. You are tentatively accepting his/her answer now, so you include or reject it at a later time, once you know whether or not it is correct. This is exactly what you did last time, and the main reason why you cannot be trusted to do accurate, honest research.

Hokulele clearly thinks that I should refuse Snorkio's answer, probably because he gave a SHA1 hash, instead of the MD5 hash that I asked in the opening post (she did not say anything about the use of $$, instead of xx).
Snorkio, you say MD5 is not safe enough. But exactly how "unsafe" is it? (btw, the use of a MD5 hash was first proposed on this very forum, by another person, now I hear a complain that it is not good enough...). I gave a MD5 code of a sentence containing my target number in the opening post:

...
A MD5 hash code for a complicated sentence containing my target number (like, for example: "The number I wrote is 5. f4315d 3b1àéùd81") is:
2ae41c33a0469b37b6c7848249026b0a
...

Can you crack it? Can anyone on this forum really decipher it, find the string I used to produce it (and, therefore, the target number)? Or, would it take longer than say, the age of the universe to crack it, on any affordable computer?

Newcomers in the thread, please go to post #1031 to find the opening post of this test. Please, do answer! I need more answers (only one or two valid so far)

 

[dlorde]

Did you ask calwaterbear if he was serious or just making fun of your thread?

Frankly, I think it was quite clear he was serious, from the tone he was using. Now, in these telepathy matters, I think it is always better to be tactful, and not harass people who were generous enough to tell once briefly the (assumed) truth.

Oh good grief

I'd better say, for the record, that any post I've made that might look as if it supported or endorsed the idea of telepathy in any way, was not serious.

I think it is beyond reasonable doubt that telepathy is fictitious, but I'm willing to participate in a well-designed experiment to test it. Unfortunately, the protocol described in this thread is ludicrous.

 

[Hokulele]

No, what I am saying is that is you accept Snorkio's answer, you cannot later discard it if/when it turns out to be incorrect. That is exactly what happened last time. You accepted and rated answers, then AFTER you learned which were correct, discarded incorrect ones for not following your protocol. You cannot, I repeat cannot, change your protocol part-way through a test.

If you cannot grasp this simple aspect of scientific testing, your results will always be meaningless.

 

[Michel H]

...
I think it is beyond reasonable doubt that telepathy is fictitious, but I'm willing to participate in a well-designed experiment to test it. Unfortunately, the protocol described in this thread is ludicrous.

It is perhaps part of a skeptics' strategy, to try to impose a mediocre protocol. Then, the test fails, and skepticism seems victorious. Very smart, isn't it?


Newcomers in the thread, please go to post #1031 to find the opening post of this test. Please, do answer! I need more answers (only one or two valid so far)

 

[Michel H]

...That is exactly what happened last time. You accepted and rated answers, then AFTER you learned which were correct, discarded incorrect ones for not following your protocol. ...

No, I didn't do that. I pointed out that the hit rate was equal to 67% (which is much higher than 25%) among answerers who followed the recommended protocol.


Newcomers in the thread, please go to post #1031 to find the opening post of this test. Please, do answer! I need more answers (only one or two valid so far)

 

[Hokulele]

Better a mediocre one, than a vague, inconsistently applied one.

 

[Akhenaten]

No, I didn't do that. I pointed out that the hit rate was equal to 67% (which is much higher than 25%) among answerers who followed the recommended protocol.

Trouble is, your version of a "recommended" protocol was as subjective as everything else in your stupid test.

You might just as well have posted your results in the OP as having gone through the ridiculous pretence of conducting anything like an objective test.

 

[Tomtomkent]

It is perhaps part of a skeptics' strategy, to try to impose a mediocre protocol. Then, the test fails, and skepticism seems victorious. Very smart, isn't it?

No. The sceptical process would be to implement a protocol that is actually suitable for the job. If it happens to be mediocre so what? It does not have to be exciting or beautiful. It has to be robust, scientific and useful. Three things your own protocols are not, for reasons already explained.

 

[Tomtomkent]

No, I didn't do that. I pointed out that the hit rate was equal to 67% (which is much higher than 25%) among answerers who followed the recommended protocol.

The flaws with this number were pointed out and ignored.
Just like the flaws with your subjective views on how credible answers are.

 

[Akhenaten]

It is perhaps part of a skeptics' strategy, to try to impose a mediocre protocol.

It's a polite way of suggesting to various claimants of nonsense that their own protocol couldn't even see mediocre from the top of a thousand-foot ladder.

Then, the test fails, and skepticism seems victorious.

At the end of the day, skepticism succeeds or fails by its ability to explain observable reality.

As does whatever you're doing.

Very smart, isn't it?

One is, one isn't.

 

[Giordano]

Michel5,

I am curious why you reject the "big random number" idea? It would eliminate the possibilities of cheating, poor statistics, and choosing a favorite number between 1-4 that have been brought up here. It is a lot more simple protocol and would eliminate the complexity and concerns about you choosing the "legitimate" guesses.

In terms of your current protocol: I think others have brought this up, but what if I tell you in my open post that I hear the number very clearly, but I am simply lying to you and I am randomly chosing a number? You would consider this a legitimate choice, but it is no better that if I was honest and told you I heard no number at all in my brain (or heard it correctly but chose to lie). You would accept my dishonest guess just as you would accept an honest guess. Both would throw off the statistics. You seem to believe people lie in telling you numbers- why wouldn't they lie in telling you how reliable a choice they made?. If you want a real test, just find a random number between 1 and 1,000,000 and accept all guesses. The lies would disappear in the noise.

 

[dlorde]

It is perhaps part of a skeptics' strategy, to try to impose a mediocre protocol. Then, the test fails, and skepticism seems victorious. Very smart, isn't it?

Even a mediocre protocol would be better than the absurd protocol suggested here, but there's no good reason - beyond your fear of failure, perhaps - why you should not use a reasonably good protocol, robust enough to test your hypothesis without significant confounding influences, and sufficient to document the procedure for subsequent replications.

It's worth noting that even the lamentably slack protocol and paltry sample size used in your last thread gave the expected no-better-than-chance result before post-hoc winnowing (although, to be fair, the sample size was too small for any conclusion to be drawn either way).

Skeptics don't care about the success or failure of the test (there is always a possibility that significance can be obtained by chance alone) but about the quality of the test. Simple and practical protocols have been suggested in this thread that would satisfy most of the skeptics here, and would be easily replicable. If you were as confident of your claimed ability as you make out, you'd be biting our hands off to implement one of them - and you'd probably get a usable sample size to strengthen confidence in the result.

Your insistence on subjective judgement of 'credibility' suggests either you don't understand the scientific method, or you are determined to cherry-pick the results.

 

[Innocuous]

I am new to the ideas behind setting up tests for claims like these, so forgive me if I am being obtuse. I am confused about the opposition to the credibility rating. I don't like it, but I also don't see it as a big obstacle to creating a protocol. Ignore the limited 1-4 range for a moment, which I think is the real problem, and assume the numbers that get "broadcast" are randomly chosen.

Suppose we let the testee discard up to X% of all responses, based on any criteria he sees in the clear text, with no need to explain. We don't need to understand the reasons he wants to do this: let's just pretend it is part of the telepathic process and part of the claim. The underlying mechanisms are beyond us, for now. He has to do these exclusions based solely on the clear text response - he doesn't know what is in the encrypted part, so he doesn't know the numeric answers yet.

In a forum test, I can see some tricks that are theoretically possible. He could load the responses with input from cooperating friends and exclude most of the non-friends. He could be a master hacker and hack our computers to find our answers and then exclude anyone he couldn't hack. Are those the types of concerns people have? Is there a fear that the clear text might subtly and unconsciously reveal the number chosen by a candidate?

Otherwise, if he has no special abilities, he can't pick results to favor his claim. As long as we have enough "credible" results left over, he will still only get hit rates consistent with chance. If we set up all the parameters before the test and don't allow post-test rationalizations, I don't see the problem. What am I missing?

 

[Hokulele]

The fact that exactly this happened, so then he went and further filtered the answers, after he knew the results, in order to get the percentage of hits he was looking for.