Dan Koe: The Counterintuitive Truth That You Don't Need to Remember Everything You've Read

Author: Dan Koe

Compiled by: FFIVE, MOOC

"If you have to deliberately remember it, then it’s not important; if it really matters, it will naturally come to mind when you need it."

This statement sounds like some sort of mystical self-help advice, yet it is the core proposition that Dan Koe presents in his article "How to Remember Everything You Read." Our generation is caught in a strange anxiety: we have saved thousands of articles, installed five or six note-taking apps, and built what seems to be a perfect "second brain," yet when we need to use it, we find nothing comes to mind, and we feel increasingly unable to "remember things."

The problem has never been about memory; it lies in our fundamentally misguided understanding of what "learning" is. Below, I will follow Dan Koe's framework, combining cybernetics, output-driven learning, and the practical application of current AI tools (like Obsidian + Claude, Eden, etc.) to reorganize a system that can truly help you "remember" and "apply" what you have learned. The article is lengthy, but each section corresponds to a pitfall you have likely encountered in knowledge management.

1. Why "Remembering Everything" is a Fallacy

Let’s dismantle a deeply ingrained notion: Learning ≠ stuffing information into your brain.

Schools have trained us for over a decade with rote memorization and exams, leading us to mistakenly believe that "being able to recite verbatim" is a sign of having learned. But the real world doesn’t score based on answer sheets—your boss won’t promote you just because you can recite a passage from The Economist, and clients won’t sign contracts simply because you remember the steps from a marketing book.

Dan Koe’s observation is sharp: Most people fail to remember what they’ve read, not because they didn’t read carefully enough, but because they treat "remembering" as the endpoint of reading. Your desperate attempts to remember every word stem from a desire to appear smart—"Look, I can casually throw out knowledge points." This vanity-driven learning is misguided from the start.

The truly important things never require rote memorization. They naturally settle into your cognitive structure through repeated use, reflection, and discussion with others. Forgetting most content is a normal mechanism of the brain, not your failure. What you should care about is not "how to remember," but "what is worth remembering and what mechanisms will allow it to surface when needed."

2. Viewing Learning as a Ship: A Four-Step Cybernetic Feedback Loop

Dan Koe utilizes cybernetics—a term derived from ancient Greek meaning "steersman"—to redefine learning as a feedback regulation system, rather than a one-way input pipeline.

An intelligent learning system operates on the same logic as a home air conditioner’s thermostat or the pancreas secreting insulin, consisting of four stages:

Most people struggle to learn not because of a lack of willpower, but because they only engage in Step 2 (blind input) without Step 1 (clear goals) to generate "error signals." Without a gap, the brain doesn’t know which information is worth retaining and which can be discarded, leading to a situation where you cram everything in and forget it all.

A straightforward example: Saying "I want to be healthy" is too vague to trigger any corrective action—if you drink too much tonight or stay up until three, you won’t feel like you’ve deviated from anything. But if you set a goal of "completing a half marathon in three months," then every night you don’t go running will generate a clear "error signal," forcing you to adjust. Learning works the same way.

3. The Most Effective Learning Starts with "Output," Not Input

Understanding the cybernetic framework leads us to the most counterintuitive yet effective principle: Don’t learn first; start working.

Naval has discussed the concept of "Specific Knowledge"—knowledge that aligns with your nature, provides a competitive advantage, and is irreplaceable. This type of knowledge cannot be acquired through "systematic remedial education"; it can only be obtained through the process of doing.

Dan Koe outlines a three-step approach:

1. First, have a meaningful goal that belongs to you (not the default path of "certification/education/promotion" imposed by society). The more specific the goal, the stronger the neural plasticity.
2. Directly launch a project instead of preparing lessons first. Want to learn After Effects? Don’t binge-watch the entire tutorial; set a specific project—like "creating a 30-second promotional video for a friend’s café"—and only learn the skills that will advance that project.
3. Fill in the gaps as needed. If you get stuck on keyframes, look up keyframe tutorials; if you’re stuck on color grading, study the color grading panel. After a few projects, the skills you acquire will be cumulative, and each piece will be tied to real scenarios, making them unforgettable.

Learning guitar is the same: don’t dive into music theory books first; pick a song you really want to play, learn the first chord, then the second chord, and when you get stuck, learn about tuning and rhythm. Once you can play a few songs, you’ll naturally become curious about "can I write a melody myself"—learning happens in the pull of "wanting to create," not in watching presentations.

Schools excel at teaching general knowledge but don’t teach you how to navigate for yourself. Self-education isn’t about hoarding knowledge to create the illusion of "I am progressing"; it’s about filling in the just-enough gaps for the goals you set for yourself.

4. Why Most "Second Brains" Become Digital Graveyards

When discussing knowledge management, we cannot avoid the concept of the "Second Brain." Tiago Forte’s PARA method and CODE process were originally designed to liberate memory, but in reality, 90% of people use it as a high-level collection folder.

Dan Koe has also experimented with several generations of tools: from Roam Research changing his writing style to creating Kortex, and later evolving into the current Eden. His conclusion is: The problem lies not in the tools, but in their usage.

The pitfalls most people fall into are threefold:

• The illusion that collection equals completion: You saved links, highlighted text, and took notes, but then never opened them again. The act of organizing itself brings cheap dopamine, leading you to believe "I am learning."
• Obsessive over-organization: Tagging every note with five or six labels, spending hours sorting into PARA categories, yet never producing a short article, project proposal, or sharing based on those notes. The value of notes lies in the work they help you complete, not in the notes themselves.
• Ignoring the design for "retrieval": Almost all tutorials focus on "how to put things in," but no one teaches you "how to pull them out when needed." A system that only allows input without output is not a knowledge base; it’s a digital graveyard.

Looking back at those who truly produce work—Marcus Aurelius’s "Meditations" were originally private notes; Da Vinci left thousands of pages filled with sketches and questions; Mark Twain, Montaigne, and Rick Rubin all had their own note-taking systems. The biggest difference between them and modern "note enthusiasts" is: collecting ideas is for turning them into works. Seneca had a brilliant metaphor: collecting pollen from many flowers to make your own honey. In today’s terms, it means digesting external materials to create something that belongs to you.

5. Upgrading the "Second Brain" to a "Second Subconscious"

Dan Koe proposes a more precise term: What we need to build is not a static warehouse, but a "Second Subconscious."

What are the characteristics of the subconscious? It doesn’t only work when you actively dig into it; it quietly associates in the background and suddenly "throws" an association at you when you’re pondering a question. A good knowledge system should work similarly—it should actively present relevant materials to you while you create.

Option A: DIY with Obsidian + Claude Code

This is suitable for those who are willing to tinker and seek complete local data. The core idea is to let Claude act as your "knowledge manager," rather than maintaining a complex classification tree yourself.

The setup logic is roughly as follows:

1. Set up the foundation: Install Obsidian (local Markdown notes, with complete control over your data) and configure the Claude Code or Claude Cowork environment.
2. Set the working directory: Set the Obsidian Vault folder as Claude’s working directory, essentially giving the AI a "read-only + on-demand rewrite" key.
3. Write two Skills (commands):
   • "Save Idea" Skill: When you throw an inspiration, an article link, or a tweet to Claude, have it create a new note in the Vault’s Inbox folder, automatically adding a clear title and timestamp.
   • "Process Inbox" Skill: Have Claude periodically read the Inbox, add tags, move notes to appropriate category folders (Projects / Areas / Resources / Archives), and create bidirectional links [[ ]] for related notes.
4. Daily Use: When writing or working on a project, simply tell Claude, "Find all content related to [current topic] in my Vault," and the AI will search the entire knowledge base and bring back referenced materials.

A more advanced repository structure can refer to this layered approach:

The core principle is that the source file is read-only, and AI only interacts with the Wiki layer. In CLAUDE.md, outline your areas of focus, repository structure rules, import and query processes—this serves as a map for AI, guiding it on where to categorize different types of information and what format to use when citing.

Option B: Tools like Eden for "Automatic Categorization + Semantic Search"

If you prefer not to maintain a vector database, embedding API, or indexing scripts, and want to avoid the hassle of re-embedding after each modification, handing over the underlying maintenance to tools while focusing on thinking and creation is a more cost-effective choice.

Eden (formerly Kortex, which Dan Koe was involved with) functions similarly to Obsidian + Claude but is more "seamless":

• Automatic Collection: Substack articles, YouTube videos, and X long texts can be saved with one click, automatically transcribing text and highlighting it.
• Automatic Embedding and Semantic Search: Each record is encoded into a vector of about 1500 numbers, akin to a "GPS coordinate" in the knowledge space. Even if two notes do not share overlapping keywords, as long as they are semantically close, the system can connect them—something traditional keyword searches cannot achieve.
• Connection to Readwise: If you use Readwise to save book excerpts, you can sync them into Eden, turning them into a resource library that can be searched semantically, engaged in dialogue, and dragged into Boards to create citation cards.
• Outlier Content Discovery: Eden analyzes content where creators excel, helping you find topics worth discussing in depth, rather than aimlessly scrolling through information streams.

MyMind also embodies a similar approach—you input links and ideas, and it automatically categorizes, tags, and performs semantic searches, eliminating the need for you to maintain the process.

There is no clear superiority between the two options:

• Obsidian + Claude excels in complete autonomy, local data retention, and high customization, but requires you to write rules and manage indexing;
• Eden / MyMind excels in zero maintenance, strong semantic connections, and built-in creative workflows, but data and indexing are cloud-based (although Eden promises not to upload private content, you are architecturally reliant on its services).

6. Filtering is More Important than Collecting: What to Include in the System

Regardless of the tool used, one thing must be clear: not everything should be stored.

AI can already produce content in bulk; what is scarce is the layer that has been filtered by you and can be digested through your worldview. Dan Koe's advice is straightforward—only collect ideas that you are willing to be shaped by.

The thinkers and creators you continuously follow and reflect upon will gradually form your worldview lens. For instance, when Dan Koe contemplates the essence of reality, concepts like history, integrative theory, and spiral dynamics naturally emerge, as they are common frameworks he uses to explain issues. What you need to accumulate over time is this kind of "collection of ideas" that belongs to you, rather than saving trending topics today and new concepts tomorrow.

How to operate specifically:

• Read more books and thoughts from stable, long-term good authors and avoid scrolling through anxiety-inducing information streams. For writers and creators, research may account for 80% of the workload, with the remaining 20% being the digestion of researched material into useful expressions for others.
• Transform materials into part of yourself through writing and public sharing. Simply collecting without output means those ideas remain in the realm of "others' wisdom"; only when you restate them in your own words, reconstruct them with personal experience, and test them in real scenarios do they truly enter your cognitive structure.
• Let projects serve as filters. Knowledge that does not propel your current project will not stick, even if learned temporarily. Projects provide clear boundaries and milestones, automatically helping you filter out noise.

7. The Correct Role of AI: Reducing Friction, Not Speaking for You

Finally, it is essential to clarify the boundaries of AI in reading and writing—this is the area with the most misunderstanding.

Dan Koe's stance is balanced: do not use AI to express your values and judgments, but it can serve as your course, mentor, and editor.

The specific balance is roughly:

• Factual content can leverage AI: For example, explaining Greuter's nine stages of self-development or outlining the background of a theory can be directly generated or cited from existing materials by AI—this is no different from using Google to research before writing; forcing a complete rewrite wastes readers' time.
• Opinions and organization must be done personally: To articulate your thoughts clearly, you must experience some "friction"—by personally refining the structure, organizing arguments, and revising wording, you will better understand what you truly believe. If you delegate even this part to AI, what you produce will not be your worldview but rather the statistical average of the model.
• Scenarios for using AI to "remove friction": When you cannot conceive a structure, ask it for several organizational methods; when unsure how to develop the next paragraph, let it suggest possible directions; when encountering unfamiliar concepts, let it help you quickly dive into research. Ultimately, the one making judgments and deciding what can be published is always you.

What is meant by "do not use AI for writing" is to remind you not to let AI express your values. AI does not replace reading; it makes reading more critical—because answers that can be directly provided by AI are becoming less valuable, what you establish through deep reading as an irreplaceable worldview is what others will want to follow.

8. A Practical Writing/Creation Loop

If you are ready to start writing (whether for Substack, WeChat public accounts, X long texts, or YouTube scripts), the process Dan Koe provides can serve as a starting point:

1. Choose a "validated + curious" intersection theme. The term "validated" means that this topic has previously garnered genuine attention—you can save highly shareable social posts or use Eden's outlier discovery feature to find standout content; "curious" refers to questions you are willing to ponder even without traffic. The overlap of both ensures you do not lose your desire to express while also being more visible.
2. Generate ideas first, then build structure. Do not wait for inspiration while staring at a blank page. Write down as many thoughts related to the theme as possible, sift through your notes for usable materials, then give them a simple narrative skeleton (e.g., problem → insight → solution), and fill the materials into appropriate positions.
3. Draft the connections. Place outlines, notes, and knowledge bases in accessible locations (for example, let Claude read your Vault or open Eden's semantic search), then weave them into a complete article.
4. Revise repeatedly until you express what you truly want to say. If a piece of work does not yet convey your values, thoughts, and beliefs, keep revising until it does.

9. In Conclusion: Remember, It’s Not the End, It’s a Byproduct

Returning to that opening statement—you do not need to remember everything you have read.

What truly remains and helps you make better decisions is never those sentences you force yourself to memorize, but:

• The things you actively learn for a goal that belongs to you;
• The things you have repeatedly used in a real project;
• The things you have digested through writing, sharing, and discussing with others;
• The things you have placed into your second subconscious, allowing them to surface when needed.

Tools (Obsidian, Claude, Eden, Readwise...) merely make this process smoother; they cannot establish your worldview for you, nor can they answer "where do I truly want to go?".

First, make your goals specific enough, start working on a project that belongs to you, and then let knowledge come to you through the process of doing—whether you remember or not becomes unimportant, because those important things have already become part of you.