This is an idea I’ve had for a while now and that I got interested in again with the creation of the Stop Danie Krugel website.
Basically the idea is for me to make a claim, similar to Mr. Krugels, about being able to find the location of someone using some kind of extraordinary technique. Then with the assistance of other people (friends, other sceptics, academics) to go through the process of establishing a proper protocol that will conclusively test whether or not I in fact have this claimed ability. The final phase will then be to apply the protocol and test it for real.
The idea is not for me to prove that I have this ability or device (even though I have a cute idea for a mechanism) but to setup and test, the process of testing this kind of ability. I think it’ll be informative and educational, giving some insight into ways to make such a test watertight. It will also, hopefully, serve as an example of the kind of test that is required to prove or disprove, conclusively, the ability of the kind of device Mr Krugel claims to have.
Also I think it’ll be fun in a treasure hunting kind of way 
So if anybody has any ideas on how this could work, or want to help, or even want to steal the idea and do something this yourself, please do comment.
When I offered to help Krugel test his device, I had a rought protocol in mind.
It went something like this:
1. I would obtain hair samples of various people located in a few places around the world. I had obtained consent from several volunteers around South Africa and in the US. The volunteers would all make detailed journals of their whereabouts for the duration of the test.
2. I would randomise the samples: putting them in numbered envelopes and have them couriered to Krugel.
3. Krugel would then use his machine to determine the location of each individual and record his results in a spreadsheet next to the number, along with the date and time he took the reading.
4. Upon recieving his completed spreadsheet I would compare his results to the logs of the volunteers and determine his accuracy (allowing for a reasonable margin of error of, say, 300m or so).
If he got even one of them right, I would be impressed.
Sounds like a good protocol Owen.
Hmmm this has given me an even better idea. Imagine if we did the same test but both Krugel and myself did it. I would use some method to randomly generate a location (probably a GPS co-ordinate) and we could compare my success with his. This would then show that his device works no better than random chance.
I’m trying to work out some possible probabilities here:
Let’s say that we limit the area in which our volunteers can be to Gauteng. According to wikipedia Gauteng has an area of 17 000km^2. Let’s define a hit as getting within the same square kilometer as the person. Thus the probability of getting a hit (randomly) is 1/17 000 or 5.88235294 × 10^-5 or 0.000059 (rounded off).
To then get 7 (or more) out of the 10 correct is a probability of something like: 2.98592^-28.
In other words it’s an incredibly unlikely thing to be able to do by pure chance. I agree though that even getting just 1 correct out of say ten would be very very impressive.
That’s a good protocol Owen.
300m may sound very restrictive (as it sounded to me initially) however, it allows for a search area of more than 280 000 sq m. This is more than three times less accurate than Danie’s Madeline McCann reading which was widely recognised as his least accurate reading of all his cases. In short it is a very generous requirement.
I would not be impressed if he got one correct because there are means available to a former police detective to find relatives and friends that you may have asked for a hair sample from. Even if he just guessed based on the list he could conceivably happen upon one correct guess, it depends on how many times the test is done; one correct guess in two attempts is more significant than one correct guess in thirty attempts.
Help from a statistician?
It would be even less impressive if the target was carrying their own personal cellphone at the time of the “reading”. As Danie has demonstrated when dealing with Moonflake; he has many resources available to find the details and even the unlisted cellphone number of an anonymous blogger.
James I’m assuming that Krugel would have no idea who the volunteers were besides having a sample of their hair. With that kind of condition he would literally have to either guess a co-ordinate or his machine would have to work.
My previous comments outlines the odds for getting that kind of guess correct (if my math-fu is still strong :p). Of course one can make educated guesses within the more urban areas but even so the odds are really not that good.
I agree though that any such protocol will have to account for his resources and experience as a policeman.
Darthfishy, there was a bit of a post clash going on there. Doh!
Requiring Danie to be accurate to 17 metres might not be acceptable to him. The most accurate “reading” that I heard of previously was 20 metres in the Naledi case.
And another post clash