Re: [aima-talk] the independence of the two factors in the value of information

Expand Messages
• It s not really about whether or not the information would change your decision, it s that the answer is **unlikely** to change your decision. It s tough to
Message 1 of 2 , Jan 29, 2006
It's not really about whether or not the information
**unlikely** to change your decision. It's tough to
illustrate when you have a pretty flat probability
distribution as implied by your DVD example.

Harry's Bargain Electronics sells the DVD players for
\$99, but they're running a special promotion where
Crazy Ron's sells the same DVD player for \$50, but
everyone has to pay. In this example the
"information" is whether or not you would be one of

Madame Trelawney the psychic is able to tell you with
100% certainty whether or not you will be one of the
lucky ones to get a freebie if you buy from Harry's.
How much would you pay her for that knowledge?

While the information could cause a significant
difference in the relative expected values of the two
choices (\$99-\$50 vs. \$0-\$50), it's **unlikely** that
you're going to get a freebie, so the information
itself isn't terribly valuable. It's **very**
valuable **if** you're going to be the winner. But,
999 times out of 1000 you'll find out you're not going
to be a winner. In that case you'd just go to Crazy
Ron's, which is the same decision you'd make without
the information.

Rob G.

--- Jim Davies <jim.davies@...> wrote:

> I'm teaching decision making now, and the students
> have this question
> about information value that I'm having trouble
>
> RN2 p602 says that information has value to the
> extent that it is
> likely to cause a chagne of plan and to the extent
> that the new plan
> will be significantly better than the old plan. Are
> these independent?
>
> I'm trying to come up with a intuitive example of
> this, and I'm having
> trouble coming up with a situation in which the
> information is
> unlikely to affect your decision, but the
> information might result in
> a significantly better decision.
>
> I came up with the idea of buying a DVD player,
> choosing between going
> to get it at best buy (A) and circuit city (B). The
> information in
> question is the exact prices at these stores. The
> agent knows the
> price ranges.
>
> Situation 1:
> A: \$10-80
> B: \$15-98
>
> In this case, it's likely to affect your decision
> and you could save a
> lot of money because of it. The information is worth
> most in this case.
>
> Situation 2:
> A: \$48-51
> B: \$49-53
>
> It's likely to affect your decision, but the
> difference is not big, so
> the decision doesn't really matter, so the
> information is not worth as
> much.
>
> Situation 3:
> A: \$30-50
> B: \$48-60
>
> Unlikely to affect decision, because A is so much
> more likely to be
> cheaper, AND the price ranges are so close that you
> don't save much
> money. Information worth the least in this case.
>
> Situation 4:
> A: ??
> B: ??
>
> I want an example in which it's unlilkely to affect
> the decision could be significantly better given the
> information. I
> was thinking
> A: \$10-51
> B: \$68-98
> but it seems with such a little overlap it's hard to
> say that the
> change in decision will be important. How can it not
> affect your
> decision and also result in a significantly better
> outcome?
>
>
>
>
>
>
>
>

__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
Your message has been successfully submitted and would be delivered to recipients shortly.