Moderated Dowsing By Edge

[Spektator]

I tried to create an unbiassed test protocol, edge. What's wrong with this one?

1. The requirements for the test are the dowser; his dowsing rod; a target that he is sure he can detect, small enough to be concealed within a covered plastic container; ten such containers, all identical and all with the same kind of cover; nine dummy targets of a similar weight to the real target; and a scale as specified by the dowser. There will also be four observers, two chosen or accepted by the JREF,two chosen or accepted by the dowser, a camera operator, and two video cameras, set up to capture the test from different angles.
2. In an initial callibration run, the video cameras well bestarted and the dowser will ascertain that only the real target produces the effect on the scale that he expects and that none of the dummy targets do; during this callibration run, the containers will initially be open so that the dowser can see the contents, but before the measurement is made, the tops of the containers will be put in place so that physical conditions in the actual tests will be identical to those in the callibration run.
3. Before the actual test runs, the dowser must agree that he is satisfied his dowsing rod and the scale are performing as expected.
4. The cameras will be stopped. The dowser, the camera operator, and two of the observers (one chosen by or accepted by the JREF, one chosen by or accepted by the dowser) will leave the area.
5. The remaining two observers will, by a randomly-generated process, determine the order in which the containers will be tested. The containers will be temporarily numbered 1-10 (by means of Post-It notes, for example). These two observers will then notify the dowser, the camera operator,and the other two observers that the test is ready, and they will leave the area.
6. The dowser, the camera operator,and the first two observers will return. The cameras will be started. Taking turns, the two observers will bring the containers, one by one, to the area beneath the scale. If the dowser wishes, the identifying Post-It number will be removed prior to dowsing the container, but the observers will keep a record of the order in which the containers are dowsed,and the Post-It will be re-applied as each container is dowsed.
7. No container will be open until all ten have been dowsed.
8. At the completion of each run,the containers will be opened. A "hit" will be counted ONLY if the dowser has successfully identified the target. Any other outcome is a "miss."
9. The process will be repeated for ten runs. A successful test will require the dowser to identify the target correctly at least six times out of ten tries. Six, seven, eight, nine, or ten "hits" will constitute a successful test. Any other outcome will constitute a failure.

It may need refining, but you need to start someplace. How should it be improved? (One thing I'd suggest: the dowser does NOT get to learn whether he's hit or missed until all ten turns are over.)

 

[politas]

The protocols that others have provided have been bias to the skeptics and really don't cover what all is involved.
Mining with the method of dowsing really answers it all.

The purpose of a test of dowsing is to cut out all other factors that might be affecting the result. If your proposed protocol does not do that, it is unlikely to be accepted.

 

[Spektator]

Something else that occurs to me: It's also a "miss" if the dowser incorrectly identifes a dummy target as the real one. Otherwise it would be possible, of course, for the dowser to say "that's it" with each target for ten times in a row and "hit" every time! So I think for each run the dowser should have only one "that's it." If the dowser could identify five out of ten possibles as the target, that radically changes the odds!

Also, I'd prefer it if the targets were shuffled randomly, being placed in different containers for each run of ten. Try as hard as you can, you're not going to get ten absolutely identical containers. Even if the dowser were not consciously trying, he might subconsciously pick up on one container that had some small irretularity, mark, or flaw; and if the target were always in that one, he'd be much more likely to "hit". So the dowser should make sure that the presence of a target in one container will not "contaminate" the container if it is removed and a dummy target is placed in its stead.

If that were the case, the containers might be numbered from 1-10 on the bottom (where the dowser could not see the number) before the test began; that would really be better than stick-on numbers anyway, unless a Magic Marker-inscribed number would give a false positive reading, and I can't see how it possibly could.

 

[Cuddles]

I have a question. Has Edge actually applied for the challenge?

I just looked over the start of the thread, although I haven't read the whole thing. As far as I can see, Edge has sent some letters and e-mails to Jeff, but up until at least mid-January had still not actually sent in an application.

 

[Paulhoff]

Edge, this isn't the place to discuss protocol with me, as you know. Please contact me at jeff@randi.org.

Paul

 

[Paulhoff]

I talked with Edge a couple of weeks ago and invited him to talk to the producers of a Japanese TV show we're doing. It was pretty clear from talking to him that he's not ready to be tested, and the Japanese producer agreed.

Paul

 

[Tricky]

In the last two weeks I have found a great deal of information about dowsing and my averages are excellent especially here as this is the worst case scenario to test at even though they are just above what I said is my average.

Average of what? You have never clearly explained how you are determining your "average".

The protocols that others have provided have been bias to the skeptics and really don't cover what all is involved.

Yes they do. You say you can detect gold. Somebody puts gold in front of you. You detect it. If you can't do that, then everything else you say is BS.

Mining with the method of dowsing really answers it all.

No, it answers nothing, since the only way to tell if there is gold or not (or how much) is to excavate the entire area to the depth of what you claim your dowsing can detect, then painstakingly examine every single grain of rock. That would be stupid when it is so easy to be sure there is gold by simply placing it there. Even if you are getting some extraneous readings from sources other than the placed target, you should still be able to find the target at a rate better than random. If you can't, then simply can't find gold by dowsing at a rate better than random.

I have sent a second protocol, which covers all the aspects of dowsing that is positive, to Jeff.

That doesn't sound like a protocol at all. That sounds like more of your "explanations". The JREF has said before (and James Randi told you in person) that they are not interested in explanations. They want you to say what you can do then demonstrate it in a way in which the results are unambiguous. The question must be "Did the dowser find the target?" and the answer must be "Yes" or "No". If your protocol does not include such "aspects" then I find it extremely unlikely that JREF will accept it.

You guys have helped me to see a few things and even helped me in ways I can't explain yet but you are bias as skeptics so your test or protocols are one sided.

You are correct. They are biased toward gathering clear-cut evidence. Waffly, excuse-filled results are discriminated against. Them's the breaks, Edge.

If I’m right there is much more to do and learn.

Yes, that is true, but we must get past that really big "IF" first, and so far, you have made zero progress in that direction.

 

[edge]

The question must be "Did the dowser find the target?" and the answer must be "Yes" or "No". If your protocol does not include such "aspects"

And did the dowser correctly identify when the target is not there both are positives for mining, do you put in and mine or not.
This is what happens when dowsing did I miss something and leave the gold behind did I by-pass a load?
The scales tell me when the target is there or not....
It still, isn't fool proof, Out of a 100 tries on the worst ground the worst neutral spot at this property I'm at 58% correct for both when the target is there and not.
But this is the worst-case scenario for doing the test, here, and I got worst results inside the house where we tested the scales the first time.
In all three locations there were different calibrations.
From what I know now when the ground is more toward the neutral the better the results.

But thanks to SezMe the Coffee Creek spot so far has been the best-case scenario.
There at that spot I got totally better numbers as far as the calibrations, less pull means less of a drain to my bodily system, physically it’s tiring in two different ways.

The scales change everything but only to a point and there still is a flaw.
We call it a ghost reading.
I think I know why and I can’t fix that except to get a better spot.
Even then I will still miss some.
But not while mining.

Spektator, a lot of it is up to JREF, they will say yes or no.
What I'm doing is making it as simple as possible as this testing is very time consuming.
One container in one spot is all that's needed and I have done 100 single tests and there is no way to know whether it's there or not.
A toss of the coin makes it fair as the target may be there 7 times or 1 time.
According to you all I will get 10% right when the target is there and I will get 10% right when it isn't and I'm saying I will get 60% or better so that is 40% over what the result should be.
You can't say that I can guess because the target isn't there 90% of the time because each set of ten is different then.
Why I can't get over 60% goes into theory but that's my advantage to know.

I can tell you this that moving a target on and off the spot that I choose changes the pull slightly each and every time.
I’m still waiting for SezMes’ results.
It’s hard to get people to work with you, pictures aren’t really necessary.

 

[Spektator]

.....Spektator, a lot of it is up to JREF, they will say yes or no.
What I'm doing is making it as simple as possible as this testing is very time consuming.
One container in one spot is all that's needed and I have done 100 single tests and there is no way to know whether it's there or not.
A toss of the coin makes it fair as the target may be there 7 times or 1 time.
According to you all I will get 10% right when the target is there and I will get 10% right when it isn't and I'm saying I will get 60% or better so that is 40% over what the result should be.
You can't say that I can guess because the target isn't there 90% of the time because each set of ten is different then.
Why I can't get over 60% goes into theory but that's my advantage to know.

I can tell you this that moving a target on and off the spot that I choose changes the pull slightly each and every time.
I’m still waiting for SezMes’ results.
It’s hard to get people to work with you, pictures aren’t really necessary.

edge, you simply can't count anything but detecting the gold in the setup I described. If you say the target is there and it is there, you get 100% on that try. If you say the target is there and it is not there, you get 0%. You don't get any percentage points by saying the target is not there if it is not there--you MUST concentrate on the positives, not the negatives, since in my suggested test you would know going in that nine times out of ten the target would not be in place. If in six tries you detect the target every time, you've won; if in ten tries you fail to identify the target for times but do detect it six times, you've won.

 

[Paulhoff]

Could you imagine winning a sharp shooting tournament for the most shots missed by the shooter because he said he was going to miss the shot.

Paul

 

[edge]

Spektator said,

edge, you simply can't count anything but detecting the gold in the setup I described. If you say the target is there and it is there, you get 100% on that try. If you say the target is there and it is not there, you get 0%. You don't get any percentage points by saying the target is not there if it is not there--you MUST concentrate on the positives, not the negatives, since in my suggested test you would know going in that nine times out of ten the target would not be in place. If in six tries you detect the target every time, you've won; if in ten tries you fail to identify the target for times but do detect it six times, you've won.

That is still an option as we are looking at those stats too.
How ever I will need ground with a neutral area like the one at Coffee Creek.
It is still possible, but I would also have to still scan 100 targets and preferably in one spot like I described.
The plan is to go back and check again and do the test both ways.

I don't have to say much Paul as I am going by what is shown by the scale.
But I still have to know when it's not there.
Key words," to know".

 

[GzuzKryzt]

...
And I now can go any where to prove it.
...

...
How ever I will need ground with a neutral area like the one at Coffee Creek.
...

Anyone, please let me know if I edited edge's post unfairly:

...
And I now can go any where to prove it.
...

...
How ever I will need ground with a neutral area like the one at Coffee Creek.
...

Just to be sure:

...
And I now can go any where to prove it.
...

...
How ever I will need ground with a neutral area like the one at Coffee Creek.
...

In case somebody missed it:

...
And I now can go any where to prove it.
...

Paulhoff, you will have to agree, won't you: One couldn't make this stuff up.




(One lesson the late great R.A. Wilson taught me, was that when faced with the amount of human stupidity, I could either choose to laugh or cry. Guess what I'm doing now.)

 

[Tricky]

However, one thing did occur to me. If he is (as he said at one point) using ten containers of which only one has a real target, then he has a one-in-ten chance of hitting the target. Of course, he knows this, so he's going to say "not the target" more often than not. I suspect it is possible that he is counting these "not the target" guesses as hits, even though he should get them right 90% of the time. Such a flawed way of counting would vastly pump up his "average of hits".

Sadly, it looks like I may have been on to something here.

 

[Spektator]

Yes--if I just randomly guessed, look what I can do:

The target, unknown to me, is in container 7.

I guess as follows, knowing nine of the containers are duds:

1-no
2-no
3-yes!
4-no
5-no
6-no
7-no
8-no
9-no
10-no.

Okay, I missed number 3 and number 7, but hey, I was correct on all the others! I scored 80%, right?

No, of course the ONLY thing that counts is correctly identifying the target. I scored 0%. I hope edge understands this.

 

[Marcus]

Forgive me if this has been covered already, but I have just read the transcript of Edge's first challenge. He agreed beforehand that the conditions were right for dowsing, and went on to prove it by being successful in the open test. Why would JREF be interested in a second challenge? It seems to me he has already proved in spectacular fashion that he is unable to detect gold.

 

[steenkh]

Forgive me if this has been covered already, but I have just read the transcript of Edge's first challenge. He agreed beforehand that the conditions were right for dowsing, and went on to prove it by being successful in the open test. Why would JREF be interested in a second challenge? It seems to me he has already proved in spectacular fashion that he is unable to detect gold.

The JREF is not interested in a second challenge, but according to the challenge rules, a claimant is free to try again after one year.

 

[GzuzKryzt]

Forgive me if this has been covered already, but I have just read the transcript of Edge's first challenge. He agreed beforehand that the conditions were right for dowsing, and went on to prove it by being successful in the open test. Why would JREF be interested in a second challenge? It seems to me he has already proved in spectacular fashion that he is unable to detect gold.

He did prove it at that particular point in time, in that particular test, at that particular location, if you catch my drift.

Only edge's verbose vagueness - along with a strictly mathematical chance - suggests that he might succeed this time.

Alas, the process of developing a new protocol seems quite telling, doesn't it?

 

[Tricky]

Forgive me if this has been covered already, but I have just read the transcript of Edge's first challenge. He agreed beforehand that the conditions were right for dowsing, and went on to prove it by being successful in the open test.

Edge's explanation was an admission that he had lied to Randi on the day of the test. He said something to the effect of "I didn't come all that way to not be tested", so he faked his "readings" in the open test, even though he knew that he wasn't able to dowse in that room. He was hoping for a lucky streak in the closed test.

"That makes no sense", you might say, and you'd be right. It would have been better if he had told the truth from the beginning. If he knew something was messing him up, then he would be crazy to take the test under those conditions, but I suppose the lure of a million dollars makes people do strange things.

In my opinion, Edge didn't lie on the open test in Randi's office. He lied later when he told us he had intentionally faked the open test. Either way, it was a lie about a significant aspect of the test, and that makes his protests seem a little weak.

 

[colin]

Yes--if I just randomly guessed, look what I can do:

The target, unknown to me, is in container 7.

I guess as follows, knowing nine of the containers are duds:

1-no
2-no
3-yes!
4-no
5-no
6-no
7-no
8-no
9-no
10-no.

Okay, I missed number 3 and number 7, but hey, I was correct on all the others! I scored 80%, right?

No, of course the ONLY thing that counts is correctly identifying the target. I scored 0%. I hope edge understands this.

If that is what he means, that is just insulting. (To be honest, I don’t always understand his posts. I guess my Edge comprehension skills are lacking.) By just greatly making it easier for him to randomly guess, he either thinks that we are really stupid and/or he is really clever. I suspect neither is true.

And speaking of insulting:

but you are bias as skeptics so your test or protocols are one sided.

Trying to honestly and fairly test you for the claim you have made is not bias. Also, it is particularly insulting to call us bias when you appear to be trying to cheat.

Anyhow, why can’t we test for both the positive and the negative?

1. Use one container.

2. 50% of the time it will have gold in it and 50% it will not.

3. Each trial will consist of ten attempts.

4. Which attempts will have gold will be randomly selected by drawing five cards from ten shuffled playing cards, numbered 1-10, at the beginning of each trial.

5. Do ten trials.

6. He states “Yes, it contains gold” or “No, it does not contain gold” for each attempt.

7. No results will be given until all ten trials are completed.

Edge, could you get 80% or more right in that scenario?

Or has that already been covered?

 

[Spektator]

(snip)
[1. Use one container.

2. 50% of the time it will have gold in it and 50% it will not.

3. Each trial will consist of ten attempts.

4. Which attempts will have gold will be randomly selected by drawing five cards from ten shuffled playing cards, numbered 1-10, at the beginning of each trial.

5. Do ten trials.

6. He states “Yes, it contains gold” or “No, it does not contain gold” for each attempt.

7. No results will be given until all ten trials are completed.

Edge, could you get 80% or more right in that scenario?

Or has that already been covered?

Thanks, Colin, but I have a problem with that--I don't think there should really be any pre-set number of times the target will be present. Think of it as ten coin flips: if I say "heads" every time, I ought to score pretty close to fifty per cent just by random chance. If a dowser says "Target!" every time, his/her score would be fifty per cent--which in other circumstances the dowser could point to saying "Well above average!" Perhaps your suggested protocol could be altered: Maybe by a random selection that chooses numbers from 1 to 10, the actual number of times a target might be present could be generated, without the knowledge of the dowser or the observing team. Example: The target-placing team rolls a ten-sided die and number 4 comes up. The Elf mage casts a protection spell---sorry, just had a flashback to college. Ahem. They will then know that the target will appear four times in the whole run of ten. Then by a coin flip they decide which half the first target appears in: heads, it's in boxes 1-5, tails in 6-10. they continue narrowing it down until they determine which times they will place the container in the box, which times it will remain empty.

I'm sure this can be further improved---not to bias the test against edge, but to guarantee that it is a double-blind test that will present results that are statistically significant.