Thursday, July 07, 2011


Hey, I was just thinking about decision making and made a graph. Points represent decisions. The axes are the good and bad outcomes of those decisions.
As you can see, for the top left decision, you...aren't making a decision at all. For the rest, it's always a series of greater reward for greater consequences and there's always a handful of suboptimal options you never have to consider.

This model tends to ignore things like two decisions having different types of positive or negative results in favor of a simple, linear expression. It also fails to account for uncertainty, where each decision actually represents a range of unknown positive and negative effects.
Here, each decision is an unknown range of good and bad outcomes. Judging them by the best or worse cases yields very different options. For the sake of simplicity none of the decisions overlap, though they could.

I'd delve into probabilities here, but I was using MS Paint, and no matter how much better later versions have gotten, I can't get it to do a decent gradient without using the spray paint tool or just working with the pixels myself.

No comments: