A test of |-|erc's psychic abilities.
The story so far.
As usual, I seem to put my foot into it.
|-|erc (<trymyform@wwwadamskingdom.com>) is reputed to be a
highly capable psychic, so I posited a very simple (at
least, to my way of thinking; psychics may think differently)
test.
My original challenge detailed
the requirements. I've already recorded the guess, with
md5sum a5d2d18d025a2451d82629987cbb7c52, on my
home computer.
So far, I have the following responses.
- The digit '6', with the ground rules he wanted.
- My response..
- His confusion.
- My response..
- '3'..
- My response..
- '4'..
- My response..
- '7'..
- My response..
- '2', with a puzzling comment..
- My response -- for various reasons, I wanted to
change from 2.2 to 2.4 on this machine so the applet went kaphut.
(Sorry |-|erc. Not that it matters; you were losing badly anyway. :-) ).
- |-|erc for some reason treated it as a
victory.
I'm not sure if he was trying to throw 5 guesses at
the first digit (with me confirming the first 4 are
wrong!) or what; his one response suggested that he
wanted me to confirm or deny the first digit, then move
on to the next after 5 posts if he's unsuccessful -- and
he was definitely very successful in being unsuccessful.
A slightly bizarre methodology, and one which is not a
great test; the probabilities are as follows:
- The first guess has a 1 in 10 chance of succeeding.
- The second guess has a 9 in 100 chance of succeeding, given the first guess's failure.
- The third guess has a 81 in 1000 chance of succeeding.
- The fourth guess has a 729 in 10000 chance of succeeding.
- The fifth guess has a 6561 in 100000 chance of succeeding.
Total probability of success for the first digit: 40951 in 100000.
Over 20 digits this give an approximate probability for the entire
number of about 1.76*10-8.
This helps the psychic considerably as opposed to a straight
20-digit guess, which has a probability of
10-20.
Honesty.
I should be honest here. I don't believe this malarky.
I don't believe he has a snowball's chance in hell of
succeeding. However, I also want to ensure that he can't
accuse me of doing something stupid, like changing a digit
(or the entire number!) at the last minute. Therefore,
I've already recorded the guess in a file ('trynumber1')
and, in my post, supplied the md5sum. Since md5sums are
32-digit (128-bit) cryptographically-hard to decipher,
I'm giving |-|erc a fair shake here, at least so I think.
Of course he has an interesting (and somewhat valid) complaint, in that this isn't the
most interesting of tests.
To which I say, "Tough". If one is a true psychic,
one should be able to read 20 digits, no problem, which
has lots of applications, from stealing credit cards to
tracking stock funds and buying/selling options thereon.
For instance, the psychic might read that Ajax Amalgamated
Stock Funds is planning a huge purchase of Acme Widgets,
Inc. tomorrow, with surplus cash. I for one would find
that slightly unethical, but damned useful.
This is not to say I think |-|erc is a con man, at least
not intentionally; I think, however, that he may be deluding
himself here.
More honesty.
It turned out loading GIMP was a lot faster than I thought (Squid
was wonderful!) so I'm going to capture a screenshot of my
website and eventually put it here.
You can now follow the above link, as |-|erc and I have more
or less given up on this contest.
This is mostly to prove to anyone who cares that I actually
deployed the randomizer applet on my desktop so that I could
occasionally glance at it, and thereby allow |-|erc to pick
up my brainwaves. The md5sum of this screenshot is
b4d17d2b611f83dc10a8764859fda5af.
Source code
At some point I might clean this up and put it into proper form but for
now the applet is extremely simple, consisting of the following files:
It requires Gtk and Gnome, and sits in the panel displaying a random
number after 20 clicks. Each click generates another random digit
(the actual digits are picked up from the 10 millisecond digit in
gettimeofday()).