[ExI] Sigh

[Damien Broderick]

On 7/4/2010 10:49 AM, John Clark wrote:

>> Humans can't make random guesses. We don't feel right repeating the
>> same number in a sequence of numbers, for example, yet we do cling to
>> favored numbers on separate occasions.

> True, apparently psi is misleading them.

What? Try reading Daniel Kahneman and Amos Tversky on heuristics and biases.

>> People often feel some numbers are "lucky" and others aren't.

> And misleading them again.

What?

>> When we place bets on an array of numbers from, say, 1 through 50, we
>> tend to choose numbers in the range 1-31, since those contain the
>> birthdays of people we love.

> And yet again.

What?

 >> If you want to see the collective effect of such biases, take a look
 >> at the normalized scale I posted

 > So if you eliminate a very large chunk of wrong answers that psi
 > provides

What?

 > you will find in the remaining answers that psi does better
 > than random probability. Wow what a surprise!

What?

 > Normalized is such a nice word, so much better than "cooking the books".

If one draw has 2 million entrants and the next (a jackpot week) has 10 
million, adding the raw votes for winning numbers skews the analysis in 
favor of the larger week. The interesting question is whether a larger 
*proportion* of votes goes to, say, 17 when 17 is a winning number than 
when it is not.

Is there anyone else here who doesn't understand this elementary point?

Damien Broderick