Decision making, patterns and art
When making difficult decisions, sometimes we compare the costs and benefits of our options. Sometimes we make decisions based on the best-possible and worst-possible outcomes.
Is there a better way to make decisions?
We make use of expected value to reason through making decisions when outcomes are uncertain.
____________________________
His bag has 4 red, 4 blue, and 2 green marbles. If you draw a red marble, you win $2. If you draw a blue marble, you lose $5. If you draw a green marble, you win $10. Assuming that you want to gain, and not lose, money, should you play?
The bag has 10 marbles, so in theory, if you play 10 times, you should draw 4 red, 4 blue, and 2 green marbles. Therefore, in theory, on average your net payout after these 10 rounds would be 2 + 2 + 2 + 2 - 5 -5 -5 - 5 + 10 + 10 = 8.2+2+2+2−5−5−5−5+10+10=8. Given that your average net payout after 10 games is positive, you could expect that if you played this game many times, in the long run, you would gain more money than you lose.
Note that choosing to play does not guarantee that you will win the first time. Also, this is an average being referred to; you won't necessarily actually draw each marble out of the bag while restricted to 10 actual draws; the calculation is merely an average of i8what you expect to happen after doing 10 draws many times.
expected value and tells us what to expect, on average, in the long run. We can find an expected value by multiplying each numerical outcome by the probability of that outcome, and then summing those products together.
For example, with Hugh's marble game, we have a
\frac410
10
4
chance of choosing a red marble and winning $2, a
\frac410
10
4
chance of choosing a blue marble and losing $5, and a
\frac210
10
2
chance of drawing a green marble and winning $10. The expected value in dollars for this game is
\frac410(2) + \frac410(-5) + \frac210(10) = 0.80 - 2 + 2 = 0.80.
10
4
(2)+
10
4
(−5)+
10
2
(10)=
0.80−2+2=0.80.
Often, in games, different people have different information available to them. (Think of how boring poker would be if everyone knew each other's cards.) However, this has a significant implication: the expected value of something can be different depending on how much you know, and it can even change for the same person as they get more information.
However, in order to affect the expectation calculation, any information will suffice, regardless of whether it is causal or not.