In some situations, we are able to reduce our uncertainty through effort (spending money, conducting research, experimenting) while in other cases there is uncertainty which can’t be reduced. When the uncertainty can be reduced through learning, different strategies or different people may learn faster or slower.
To make good decisions under uncertainty, we need to decide if it’s worth first reducing the unknown. Or if we’re better off committing early. To help with those trade-offs, this tool encourages users to think about what’s fundamentally unknown compared to what’s learnable, as well as the cost of learning using different approaches.
This figure shows a case where the unknown is highly learnable, with very little effort:
Figure 1 – learning curve graph which starts with high uncertainty, but very quickly drops to almost none.
The next figure shows a situation where the unknown is not learnable (like a lottery ticket)
Figure 2 – Learning curve graph where the uncertainty stays the same for any amount of effort.
This next figure shows a situation where there is some learnable uncertainty, but it takes substantial effort to learn. A beginner might follow a curve like this one.
Figure 3 – Learning curve graph where the initial learning is almost none, and then gradually reduces the uncertainty to be about half of the original unknown.
The final graph shows a situation where there is some learnable uncertainty, but it takes only a little effort to learn. An expert might follow a curve like this one.
Figure 4 – Learning curve graph where just a small amount of effort eliminates half the uncertainty.