During my walk today I tried to recap my learnings from the book "Ultralearning" by Scott H. Young.
I'll recollect the learnings for each chapter.
Metalearning
This is drawing a map which will be used to travel to the desired goal. It needs to leverage how others have attained mastery in the field. Some strategies here include conducting interviews with experts in the field or looking at study curricula for the given subject. The whole idea here is to benefit from the experience of others. This maps nicely to the aspect of deliberate practice where one needs to have a clearly defined path to expertise.
Focus
In order to improve, one needs to have the ability to work in a focused way. Scott is agnostic on whether Focus is an ability which can be trained like any other. I am pretty sure it is, based on my reading of Peak. Focus is defined by the following attributes:
This is based on what I learned in the book "Hyperfocus" and also by some episodes of the Huberman Lab podcast.
Directness
This is another topic that maps directly to a concept from Peak. It highlights the importance of learning a skill over learning a concept. So the focus is on knowing how to do something over learning the what, or facts - skill over knowledge. Ericsson in his book emphasizes that the knowledge comes with the skill and not the other way around. Scott provides some strategies to learn by directly practicing the desired skill. For language learning this could be Immersion For many practical skills like programming, engineering etc. Project based learning is a good approach. What else was there? I can't recall now but probably during writing the other sections it will pop up.
Drills
Here things get practical. What exactly does one need to do to improve?
Doing a project is straight forward but how exactly will building a website
make me a greater developer than any other learner who does the same?
Drills focus on a specific sub skill in my domain. So when I am building a website there a tons of sub skills to be mastered. It is easy to assume that someone quickly putting together a website with GenAI will be less accomplished by the end of the project than someone who performed drills on each aspect involved.
Let's draw out some skills which could be learned by using drills. On the starting side there's HTML. One can get away with a very basic skill level on this front. But one could also go very deep and consider all the kinds of tags and build example pages with those over and over again. The idea is to get the skill of building a HTML template to a sort of muscle memory level. I could imagine asking GenAI to write up a catalog of scenarios for a website and use this list to train building templates for those scenarios. Then one could continue to learn about HTML's impact on SEO, accessibility and so. The idea is to break a skill down and master each sub skill.
For me a relevant area of practice right now would be authentication. I would need to look at different aspects and find opportunities to hone in on those through practical execution.
Here it's also useful to point out a comment of Ericsson: In order for the practice to be efficient it needs to push you outside of your comfort zone. So it is not useful to do the same thing over and over again but to always push a little further.
Feedback
To make sure one is moving in the right direction and to quickly correct course one needs feedback. This can be in the form of a compiler warning or through the help of a mentor. What are some aspects of good feedback?
I can't remember the terms used by Scott, so I made these up. Ideally you want to get most of your feedback in the latter form. Finding a mentor isn't always easy so what can we do to get good feedback? I think GenAI provides some potential here.
Retrieval
Here some strategies for retrieval are given:
Intuition
TBD
Experimentation
TBD