How To Get Actionable, Memorable Advice Out Of Anything You Read

In this article, I want to go over my newest process for how I learn, retain, and use information in a way that is both effective and efficient.

The Problems

Finding the most efficient way to learn is something I find genuinely fun, mostly because it feels like a problem to solve. In my case, the problems are this:

1. There is too much information, such that even if I can read 52 books a year (a book every week) you will only scratch the surface of all knowledge.
2. Even when I do read, there’s a falloff in what I comprehend. Let’s say I read a book for an hour: depending on the book, I only really start to pay attention between minute 15 and minute 45, when I get into a “flow” state. This means that 30 minutes of that hour-long session is killed off! This is a personal problem I have to be able to work around.
3. Some information can only be learned first-hand. For example, you can’t really summarize a fiction book, because summarizing isn’t the point — most of the value of fiction is in its prose, and how it flows from one scene to another.

Take in mind when I say “learning” in this context, I mean more static information that can just be taken as-is. Topics like art and language learning, which require sustained practice over time to get good at, would not work for this system and instead would work best using the reference system (which I still use). Coding, on the other hand, is a skill that mostly relies on memorizing how certain static structures work and how to utilize them (e.g. what recursion is and when to use it, how to set up a Node.js app, etc.) and thus would work for the system I’m about to describe.

A Review of my Old Strategy

In my rather controversial video (now unlisted since I don’t use this strategy anymore) I described a process in which I put everything (articles, videos, nonfiction books, fiction books) through some sort of summarizer that either (for nonfiction) gave me key information from the work or (for fiction) gave me workable prose I could apply to my own writing. I don’t hate this strategy, and as a spoiler alert I still utilize a summarizer in my current process. But this strategy did have some problems of its own:

1. Books are too long to put into any AI summarizer, and so if you do try to summarize it will just give you the first ten pages or so.
2. AI models are very poor at figuring out what is good, unique prose… at least in my opinion.
3. Reading is still a fun past-time, and if everything is summarized that means I don’t get to read anymore!

Because of this, I decided to split off these categories (books/fiction) into a separate category that involves manual reading. I also applied this somewhat to language learning — keeping large documents for immersion reading (sustained reference practice) but for shorter documents allowing the AI to just give me the vocab (the static part). The final process looks something like this…

How It Works Now

[Image omitted]

1. All content enters through Readwise Reader
2. If it’s long (we’ll say more than 10 pages) it gets pushed into the “manual reads” tag, where I read it during my morning routine. If it’s short, it gets pushed into the “summarize” tag.
3. The manual reads notetaking is pretty simple — if you highlight something in Reader, it automatically goes to your Readwise highlights, so that’s said and done. For the summarizing, I’ve decided to keep things internal this time around and use Reader’s native “Ghostreader” AI, which allows me to set up custom prompts. If it’s a language supplement, I have it invoke the language prompt which grabs vocab and grammar from a piece. If it’s anything else, it just gets summarized into bullet points like normal.
4. When it comes to getting stuff into the Stockpile, this is once again trivial for manual reads: Readwise highlights are automatically exported to my Notion, so I just throw them into the Stockpile from there. With summarizations, it’s a little bit more tricky — document notes aren’t counted as highlights, so they don’t get auto exported. However, it’s not so hard to download the notes off of Reader and import them into Notion manually — it’s just a bottleneck since you have to export every note individually. And with 4k+ items that need to be summarized… that’s a lot of pressing the “Export” button.
5. Once it gets into the Stockpile, it’s a matter of trimming and recall. Stockpile is of course random — if it gives me a highlight or summarization, I’ll go through and see if the information is still worth keeping. If it’s not, it gets removed and frees up space in the Stockpile. If it is, it stays in the list to get recalled once more — cementing its place in my memory.

What I’d Like To Add

Obviously the system is still not fully automatic, which I’d really prefer if it was. Unfortunately I don’t think I can get it any more automated unless I get new features directly from Readwise itself. So, here is my wishlist, Readwise devs:

1. The ability to mass select items in my Reader docs, and apply certain actions such as invoking Ghostreader. Part of me understands the trouble of this, as they rely on ChatGPT and therefore have to deal with ChatGPT’s throttling/costs when it comes to a massive amount of documents. But even if I can just queue the action, wait a few days, then come back to find all my documents finally summarized, that still saves me time in the long run.
2. Alternatively, if I could either 1) Get document notes to count as highlights in the Notion export or 2) Allow mass export of documents, that would also help a bunch. Preferably we get the first option, since it both seems rather trivial to add (just add a notes JSON when exporting to the Notion API) and would make the entire process automatic, at least when it comes to exporting.

[Image omitted]