Going meta


Human knowledge too vast for any one person to master.

Humanity is very good at creating domain experts and our education system is structured to help teach and identify the best in each subject. However, these experts have created a problem which is that accumulated human knowledge, the sum total of all their efforts, has grown to be far beyond the scope of any one individual to keep track of, let alone master. This is a problem because domain experts are increasingly isolated in their fields, ignorant of the wider body of human knowledge.

Long-term solutions

There are several long-term solutions to this problem.

  1. Increase an individual's capacity to learn e.g. genetic engineering, life extension, better teaching.
  2. Create better systems for synthesizing human knowledge. Current examples include how the US President will have teams of experts advising him or how big companies have hierarchies where experts advise senior management to enable them to make decisions. Maybe solutions rely on multiple experts advising one decision maker, maybe with the internet you can replace strict hierarchies with looser, more democratic decision making?
  3. Human-computer partnerships where the human acts as the synthesiser but the computer can provide access to the specialist bodies of human knowledge.

In this essay, however, I want to explore the much easier solution of Charlie Munger's which is to learn all the big ideas.

Learning all the big ideas in all the big disciplines so I wouldn’t be a perfect damn fool who was trying to think about one aspect of something that couldn’t be removed from the totality of the situation in a constructive fashion. And what I noted as the really big ideas carry 95% of the freight it wasn’t at all hard for me to pick up all the big ideas in all the disciplines and to make them a standard part of my mental routines.
— Charlie Munger, billionaire investor at Berkshire Hathaway
Berkshire Hathaway's billionaire investor Charlie Munger


The problem is that traditional teaching is not designed this way. Typically subjects are taught bottom-up with a view to helping those who aim to be experts in a subject rigorously learn it from first principles. Although, this can be valuable, outside your one or two chosen domains of expertise it makes learning even just the core concepts nigh impossible. 


What is needed is:

  1. an easy way to save, find and review the key mental models and concepts.
  2. an easy way to learn the key mental models and concepts.


  • an easy way to save, find and review the key mental models and concepts.

My current solution is fairly good which is when I books, listen to podcasts or watch interviews and someone suggests an interesting mental model I store it in quip, in one of many excel spreadsheets that I have categorized based on topic.


As an example, I recently finished reading 'Poor Charlie's Almanac' which includes his famous essay on 'The Psychology of Human Misjudgement.' Which I have quite simply inputted into an excel spreadsheet. This is very gratifying because it means that I have a systematic way to save all the knowledge and ideas that I come across that otherwise I would quickly forget.

However, as you can see this is likely to very quickly get unwieldy.

The question is in what way would I like to interact with this material? I think there are two ways the first is just general, and perhaps randomized revision. E.g. everyday one mental model from my metadatabase is sent to my phone to review. The second is retrieval, most likely, in the form of checklists I can use to work through a problem, say an investing decision. The question is how to generate this checklist? One way would be to simply create categories based upon which folder the concept/model was saved in but perhaps clever tagging could allow for flexibility on this front.

To consider an example suppose I am thinking through an investment decision and I want to run it against my list of mental models. First I would select the appropriate checklist and then I would be able to run through each item on that list systematically. For example if I was analyzing whether a company is a good investment I would consider checklist item #17 which if relevant I would click yes to and write a brief summary of my thoughts as they relate to the mental model and the specific idea I want to analyse. 

If the brief summary is insufficient to remind the user what the checklist mental model or concept is perhaps there could be a side panel which can come up with a more detailed explanation and link to further notes or even the original source material. 


Obviously to help create this functionality there needs to be a good standardized way to input information. There seem to be a few different types of input structure you may want. But you could have a simple form where if a section is left blank it would not be saved onto the database.

At the end of doing this I could then generate a report of all the checklist points that were relevant to this particular problem alongside my hand written analysis.


  • an easy way to learn the key mental models and concepts.

Of course just like how I read books and write down ideas that interest me I think a lot of the learning will be organic. However, I do think that there should be both an enjoyable and systematic way to learn the core concepts in all the major subjects.

The format my friend and I came up with are five hour videos that would be pitched at medium difficulty in between a more rigorous lecture course on the one hand and layman documentaries and articles on the other hand.

To do this the material needs to be completely reworked and reorganized to allow students to take a top-down, major concepts look at a discipline rather than a more rigorous bottom up approach. 

As an example you may take a five hour course on electric batteries which after an initial hour of foundation work would then over 3 hours explain the 10 major concepts and breakthroughs in the history of battery technology followed by an hour on the future of batteries and the major challenges and potential areas for progress. Having then learnt those 10 major concepts the student could then add those 10 concepts to their library of mental models. 

What we envision is a education company that is focused on content creation rather than platform and tools creation. To make a comparison to the gaming industry, #metalearning would be a game company like Valve or Blizzard rather than a console company like Microsoft's Xbox or Sony's Playstation.


There may even be a role for a different kind of research. Typically a lot of economics research and financial journalism is reactionary, where opinions are given on recent events taken from a largely unchanging and unspoken world view. What if instead, each news story was an opportunity to communicate and compare alternative world views? Each view would be presented with clear references to the underlying mental models and concepts they use back in the database.