The 1st 20 hours — Josh Kaufman’s theory (Thoughts on Learning)

Photo credit: nyoin via VisualHunt / CC BY-NC-ND

Getting Decently Good vs Expert level

Josh Kaufman has a great video at TEDx about how to learn anything. Unlike Malcolm Gladwell’s 10,000 hour theory on learning, Kaufman proposes that people can learn anything without the 10,000 hours input. He differentiated that the 10,000 hour rule applies only to expert level performance — to get to the top of a given field. Instead, you only need to get to an okay level competency about a subject for most things.

No matter what subject one studies, there are the following two graphs that depict the relationships between two sets of variables: performance time vs practice time and the skill level vs practice time

Both depicts the thought that the more you practice the better you get

The punch line is that how long does it take to get decently good enough?

20 Hours. By Kaufman’s theory, it only takes 20 hours to get to that decently good enough level. However, this 20 hours presupposes focused and deliberate approach to practice on a specific skill intelligently and efficiently. This rapid skill acquisition follows 4 steps.

1. Deconstruct the skill and decide exactly what you want to do when you are done. This involves looking into the skill and break it down into smaller and smaller pieces. The more one can break apart a skill into smaller and smaller bits and pieces, the more one can figure out what’s needed to achieve after that 20 hours. And the same deconstructive exercise can help one prioritize what to practice first. The hypothesis for focusing on the most important thing first is that one can be most efficient in learning through a path that focuses on the most important thing.
2. Learn enough to self-correct. Get 3–5 resources to work on but caution against using them as a means of procrastination. The common thought process is that one wants to finish going through the resource entirely but instead one should just focus on learn enough to self-correct in practice. When becoming good enough, one can then start for something different.
3. Remove practice barriers. (Distractions, TVs, Internet) All the things that keep you from practicing.
4. Practice at least 20 hours. There’s an initial frustration curve and what one needs is a bit will power to continuously to practice 20 hours on the same thing

Reflection on the 20 hour theory

The 4 simple step process seems straight forward but it seems a bit incomplete and or that it makes a lot of presupposition and got rid of a lot of intermediate steps for the sake of simplicity.

The first step actually packages a lot of smaller steps together. Knowing how to deconstruct a skill itself takes time to research. In my mind, knowing how to deconstruct a skill requires answering the following question (consciously or subconsciously):

1. What the skill is — roughly speaking, refined through research
2. Under what context the skill is used
3. What the skill is used for
4. What’s the most fundamental work units of that skill
5. What are the skills that are required in order to learn more about the skill in question
6. What my motivation is
7. What my goals are around the skill

In fact, I would go as far as saying the exercise of deconstructing a skill itself is an effective learning method. Most of these questions require conscious thought. Coming up the answers to these questions would take at least a 10+ hours of dedicated research. The answers to these questions are constantly getting updated as one researches on. For example, the questions regarding motivation and goals are pretty rough at the outset at best. For most of the new things that you study for, you may think you have a goal and you may think you know what the skills are. But all your current understanding would change as you are presented with new information. As I am writing, Kaufman’s theory seems theoretical at best.

Now, I understand this four steps are meant to be general but to act on them, Kaufman’s theory leaves much to the imagination. Take the practical subject of deep learning(DL) for example, I don’t have a pre-defined set of goals around the subject matter but broadly speaking, I think learning “DL” can help me solve a different set of machine learning problems faster that other algorithms can’t solve within the same time-frame. However to solve that specific problem doesn’t require much time. There are pre-build library that can already do that without me learning it. Then I realized my goal is to have a deepened understanding of these algorithms for no apparent reason at this point. I just want to know. (I am just uncovering these goals as I write). But then just understanding them isn’t enough. I think I want to achieve more but right now I don’t know what it is since I don’t have enough information.

As this thought process goes, you can see that people’s learning journey itself impacts the goal formation. It could be that I just suck at coming up with a coherent statement of purpose but I believe to most people, coherently articulating some statements of purpose requires conscious effort and most do it at sub-conscious level where the 20 hour theory isn’t quite useful.