Home - Success Stories - Optimizing for the Speed of Learning

Optimizing for the Speed of Learning

  Author:Shentop  Refer:http://www.shentop.net  Date:2013 /01 / 08


Many people trumpet the virtue of "failing fast" as a recipe for success, especially in the pursuit of innovation. But, as Stanford entrepreneurship professor Steve Blank tells us, "it's not about failing fast - it's about learning faster."

How do we optimize for speed of learning? And why isn't it our natural instinct to do so?

Let's start with the second question. You'd think that, as homo sapiens, we'd be driven to learn as quickly as possible, especially when we're trying to develop innovative products and solutions. But our instincts strongly favor success over failure, which leads us to make short-sighted decisions. It's related to what the behavior economists call loss aversion.

Beware The Illusion Of Progress

Imagine having a great research idea whose execution requires solving two problems. To make it concrete, let's say that you have an idea for achieving world peace that requires:

developing inexpensive personal force fields
distributing them to the world so that everyone gets them at roughly the same time
How should you proceed?

Synchronizing mass distribution is a challenge, but it's clearly the more tractable of the two problems. So you might start by working on the distribution strategy: reviewing the state of the art in logistics, researching shipping networks, etc. That would allow you to make progress on at least part of your idea.

But doing so would be a waste of time. After all, the bottleneck to your success is developing the force fields. Your progress on the distribution problem buys you nothing if you don't have force fields to distribute.

Start With What You Don't Know

The toy example above makes the mistake obvious. But reality tends to be more subtle, and we are very tempted to favor easier subproblems over harder ones. After all, it feels good to make progress. We much prefer to work on something we understand well than to bang our heads on a problem we're not even sure we'll ever be able to solve.

Optimizing for the speed of learning requires the discipline to fight that tendency. Whenever you are faced with a complex problem, break it down into subproblems. Pick the problem you are most uncertain of how to solve and attack it first. Your immediate goal is not to succeed or fail -- but to reduce uncertainty. That is what it means to learn -- and your goal is to learn as quickly as possible.

Learning From Games

One of our first opportunities to develop this discipline comes from a childhood board game: the classic game of Mastermind.

For those unfamiliar with the game: your opponent secretly chooses a pattern of colored pegs, and your goal is to figure it out the pattern over a sequence of guesses. Each guess yields feedback: how many pegs were of the correct color, and how many of those were correctly positioned.

Novice players get excited about how many correct pegs they choose on the first guess, and keep trying to increment that number. But it turns out to be better to have fewer pegs correct in your first attempt, because you learn more. As in all projects, we have to optimize for the end game, and not just the results of the next move.


As Steve Blank said, it's about learning faster. Learning means reducing uncertainty. So fight your instinct to focus on easy subproblems just because you want to feel like you're making progress. Take the risk of fast failure -- to ensure fast learning.