Re: Waves of Customers?

Rich Ulric wrote:

People are rather bad at estimating what is "random."
And what is repeated does not seem random.


There is a classroom stunt to make the point,
which I remember and reconstruct like this:


Here is homework, to be collected at the start of the next class.
"List 100 heads and tails in order: either flip a coin to
generate the list, *or* write down your own 'randomized'
H or T."


Next day --
The teacher quickly inspects and separates out the "coin" lists
from the "self" lists, almost infallibly, using a simple formula
that he eventually reveals. If there are never 5 H in a row or
5 T, then it was generated without a coin.


----------------

Interesting both people and animals can be trained to emit responses
nearly indistinguishable from genuine random sequences. Some of the
evidence is reviewed in the 2005 "Annual review of Psychology" (article
on indeterminism in the behavioural sciences).

The author of that article makes the case that game theory requires
that sometimes the only way to prevent an opponent accurately guessing
what you will do next is to be able to generate truly random responses.
So there are evolutionary and strategic conditions where the best
strategy is to be truly random. Experiments show that both people and
animals can be trained to detect such circumstances and to repond
randomly to them. I think the movements of a mosquito that one is
trying to swat exemplify this stratgey rather well. If it moved in a
determinate straight line it would be easy to predict where it would be
and rapidly squish it with some handy object. But mosquitos never move
in a predictable fashion...

Lance

.

Relevant Pages

  • Re: invite to eBay coin forgery group NEW 244 members
    ... > of coin fraud, I am interested. ... I have actual peers in forensic numismatics, ... who fob themselves off on lists such as ...
    (rec.collecting.coins)
  • Re: Change for a Dollar
    ... I sat down and bashed out a program based on that idea in Haskell: ... The the definition of a Coin: ... > length $ mkChange usaCoins 100 ... My approach can be easily adapted for any language that has list comprehensions, though linked lists work better for it than arrays do. ...
    (comp.programming)
  • Re: invite to eBay coin forgery group NEW 244 members
    ... of coin fraud, I am interested. ... I have actual peers in forensic numismatics, ... who fob themselves off on lists such as ...
    (rec.collecting.coins)
  • Re: invite to eBay coin forgery group NEW 244 members
    ... > of coin fraud, I am interested. ... I have actual peers in forensic numismatics, ... who fob themselves off on lists such as ...
    (rec.collecting.coins)