What is the hill finding algorithm?

In the hill finding algorithm you are at a specific point and your goal is to find the peak of the hill. You can only see your next move and not many moves ahead. What’s your strategy?

A simple strategy would be what they call a greedy algorithm. It goes as follows:

• Start at the point where you are and figure out all of next possible moves. In the chart above, you can either move left, right, or stay where you are.
• Move to the direction that is higher. If nothing is higher, then stay where you are.

If you follow these directions, then in this example you would end up at the left peak (the local optima). This algorithm clearly helps us get to a local peak, but it fails to get us to a global peak.

I think that this issue describes what is wrong with a lot of data driven companies in Silicon Valley. They spend an excess amount of resources A/B testing and figuring out local optimas. In reality, the market has much more strategic opportunity available if they are able to move beyond the A/B test. Understanding users instead of collecting A/B data may help yield more of the globally optimal strategies.