There's a meme going around lately, suggesting some pattern to how we guess random numbers. A few bloggers are trying to "test" it by asking their readers to email them lists of random numbers, or conducting a "study" in their comments. Why not use a poll?
IMPORTANT: Don't use the graphic as an aid. First, think of a random number between 1 and 20. Then record your response.
I'll explain what this is all about next week.
- Log in to post comments
I used to monitor operators to check that they'd actually performed calibrations, and I learned to spot the fakes by the way they would pick what they thought to be a random value within a reasonable range. Apparently some values seem more 'random' than others.
This reminds me of a somewhat tongue-in-cheek poll carried out by David Chalmers a while back:
http://consc.net/notes/pick-a-number.html
I think this will be a bit biased by the fact that many readers will already have heard about this as it made the rounds (I just saw it on Pharyngula, so I didn't vote here). Still, interesting stuff, and you probably would have gotten the same result anyway.
I also saw it on pharyngula, but as soon as I saw the title and first line I thought of a number and that's the one I "voted for" in the poll. (which was 5 btw) When I thought of the number I hadn't a clue what it was all about.
I also think I know what this is about, but that was the number I always pick anyway (there's a story behind it) so I didn't cheat by choosing it.
It shouldn't really be a surprise that humans are bad at picking a "random" number, since they fail at such tasks as: is "1001" more or less random than "1011"?
Analysis:
(n=252)
1 11 6 10 11 11 16 5
2 8 7 24 12 12 17 44
3 9 8 13 13 17 18 16
4 6 9 9 14 17 19 17
5 5 10 7 15 7 20 4
Times Numbers Share (%) each Anomaly
44 17 17.46 3.49
24 7 9.52 1.90
17 13 14 19 6.75 1.35
16 18 6.35 1.27
13 8 5.16 1.03
12 12 4.76 .95
11 1 11 4.37 .87
10 6 3.97 .79
9 3 9 3.57 .71
8 2 3.17 .63
7 10 15 2.78 .56
6 4 2.38 .48
5 5 16 1.98 .40
4 20 1.59 .32
I think I see what you've done, Roy. Too bad the tabs don't line up -- it looks like we've got 3.49 times as many 17s as would be expected. I'd be interested to see a similar analysis on a set of random numbers.
@Dave Munger
I cut and pasted your results to a flat ASCII file, then ran a Perl script to produce the analysis.
Send me an email address and I will send the script.
i chose 17, because it's my b-day and my lucky number, go figure.
I chose 17 as well - because it's the most random.
;-)
17 is the most random because it is prime. In fact the top 4 are prime numbers, as of Monday morning :)