DRMacIver's Notebook

The fastest way to learn something is to do something

The fastest way to learn something is to do something

Suppose you have a problem to solve. What do you do?

Well, you sit down and think real hard, and after extensive and careful planning you try the well thought out and rigorous solution that you have thought up. Right?

No, wrong! Bad.

The correct thing to do when you have a problem is:

  1. Think for a short amount of time.
  2. Make sure it is safe to try things.
  3. Try something you think will work.
  4. Observe the result.
    • If you succeeded, yay you solved the problem!
    • If it didn’t work, think about what that means for the nature of the problem and try again.

Every attempt to solve the problem is an opportunity to learn about the nature of the problem, and the nature of the tools you can use for solving problems in general. It will often teach you significantly more than either sitting and thinking about the problem or trying a less directed experiment will.


Obviously there is a place for learning about things that is less connected to a specific problem (given the number of books I read for general interest I would be a hypocrite to claim otherwise). Additionally, some problems are not well suited to this because e.g. They are high risk or they are long cycle time, but even then there will usually be subproblems like this.

The reason this is quite as strident as it is is as a necessary corrective to a tendency that most people have to overplan and get too hung up on finding the “right” solution. The lessons on cost of decision making apply in spades to solving problems.