DEV Community: Perrin Yong

In the Age of LLMs, Do Notes and Blogs Still Matter?

Perrin Yong — Tue, 08 Sep 2026 19:38:40 +0000

See the trees. See the forest.

In the Age of LLMs, Do Notes and Blogs Still Matter?

In the age of LLMs—large language models, AI systems that can generate natural language and other text through conversation—knowledge has become remarkably easy to obtain. Almost anyone using an LLM can access nearly all the knowledge that human beings can express through language. At the same time, some of the needs once served by internet search engines and blogs are being absorbed by the LLM chat window. So in this age, does it still mean anything to keep and maintain personal notes, or to build and publish a personal blog?

From my student years through six years of working, I have always kept notes. Over time, I have maintained more than four thousand note documents. I wanted to preserve the knowledge that I found valuable and useful in my studies and work, and to share some of the better material on the internet through my blog. In the old age, keeping notes was, in essence, a way of collecting knowledge and giving myself the feeling that I had mastered it—something that helped me resist the anxiety of facing challenges at work. Publishing a blog, meanwhile, was partly about helping others by sharing my research and implementations, and partly about building a personal identity and introducing myself to other people. But in the age of LLMs, where is the value of notes and blogs? Are all the things I recorded before simply going to “rot in a box”?

The real knowledge network of See the Forest: public content is no longer just a list of articles, but an interconnected forest.

Some time ago, I learned about Karpathy’s LLM Wiki. Inspired by it, I wanted to build a personal note library of my own, maintained with the help of LLM agents. The core of Karpathy’s approach is this:

Let an LLM agent build and maintain a wiki continuously and incrementally. Instead of waiting until a question is asked and then searching and assembling information on the spot, it processes material as it arrives—extracting, organizing, indexing, and continuously updating it—so that high-value results become persistent artifacts capable of compounding over time. He divides the LLM Wiki into a raw content layer, a wiki layer, and a schema layer that maintains the contract.

I wanted to learn from his approach and build a note library of my own, maintained by agents. Karpathy also wrote that “LLM Wiki describes an idea, not an implementation.” I wanted to learn from the essence of that idea, rather than imitate its appearance or copy its form. At the same time, I wanted to improve the approach in my own way. While building this knowledge-base system, I followed these ideas:

Knowledge is, by nature, a network. Its parts are connected to one another, much like neurons in the brain.
Notes in a knowledge base should be classified appropriately.
Pieces of knowledge can support one another, but they can also constrain and contradict one another.
Knowledge from different fields may, at some level, share the same patterns. But surface similarity does not mean their underlying nature is necessarily the same.
The network formed by the connections between pieces of knowledge is itself part of the knowledge. This network can continue to evolve as more knowledge is added. I call the process in which new interpretations and new questions gradually emerge from the interaction of knowledge across time, sources, and fields “fermentation”—like food fermenting, growing richer with time. During this process, new discoveries and new sparks may emerge.
Knowledge becomes more useful when what we understand is connected into a whole and brought into coherence.

In short, by collecting, classifying, connecting, and drawing inspiration from knowledge, we can build it into a network—and allow wisdom to emerge as it ferments. What the LLM agent changes in this process is the amount of work and the efficiency of doing it: it makes a complex undertaking easier and genuinely feasible. In other words, compared with the age before LLM agents, they have not changed the nature of knowledge. They have made something that was once too difficult to sustain over the long term practical.

That is why I built a knowledge-network system called See the Forest. Its name comes from a saying: “See the trees. See the forest.” It means that when we look at things, problems, or knowledge, we should not see only isolated points. We should also see the connections within and between them, and the network formed by those connections—not only the trees, but the entire forest they form, its ecosystem, and the relationships among its living things.

What I Wanted to Build Was Not Another Layer of Summaries

The first step was to reorganize and reasonably classify the notes I had accumulated, and to build a directory tree that could be maintained over the long term, so that every note had a basic place to belong. But a directory can only decide where a note is stored. It cannot fully express the relationships between pieces of knowledge.

One difference between the See the Forest knowledge-network system and Karpathy’s approach is that Karpathy separates the raw material layer from the wiki layer. I believe that the network formed by knowledge is itself part of the knowledge; they are one whole. I also believe that the original notes themselves should be maintained and improved, rather than maintaining only a separate wiki layer. So although See the Forest also distinguishes between a “knowledge foundation” and a “knowledge evolution layer,” it does not maintain them in two fundamentally different ways. Instead, the knowledge and the network produced through its evolution form a single whole, and agents help that whole network evolve and ferment.

The knowledge foundation preserves original materials and earlier understanding. The knowledge evolution layer maintains semantic relationships, synthesized judgments, evidentiary boundaries, contradictions, open questions, and the evolution of understanding. This does not mean allowing an agent to rewrite my words at will. The history of my thinking belongs to me. The agent improves efficiency, but the final judgment remains mine.

I want this to be a living, intelligent knowledge network. And the approach itself demonstrates something important: the knowledge network does not exist because LLM agents changed the nature of knowledge. It exists because their gains in efficiency make such a complex, previously impractical undertaking possible.

Select a topic and the content truly related to it lights up. From there, you can continue into the topic itself.

How Is This Knowledge Network Built and Maintained?

I did not tear down and rebuild more than four thousand notes merely to fit a new approach. Markdown and my existing directories remain the knowledge foundation, and Obsidian remains my main interface for reading and maintaining it. Through classification, bidirectional links, and Graph View, I have gradually connected content that was once scattered. But more links are not automatically better. What matters is why two things are related, and how new material changes what I previously understood.

I now maintain the system together with agents. Agents help me discover relationships across a large body of material, compare new knowledge with old, and maintain the network. But whether a question matters, which ideas are worth preserving, and whether a change should ultimately be accepted are still judgments I make. I call this process “knowledge compilation”: locate the relevant earlier knowledge, compare the old with the new, write back the changes that matter, and finally validate the result. There are more specific mechanisms behind relationship judgment, permissions, and engineering, but I will not discuss those implementation details here.

Select a node and see its place in the network and its real relationships.

Why Did I Turn It into a Blog?

Once the basic structure of my personal knowledge base had gradually taken shape, I wanted to uncover more of its value—this network should be visible. If we could actually see its structure with our own eyes, first observing the topics, relationships, and spaces as a whole, and then entering the details of a specific piece of content, that would be very cool.

Obsidian already has Graph View. Its way of displaying and interacting with a knowledge network inspired me, but only I could browse it locally. Why not build a visual and interactive knowledge-network system that could be presented on a personal blog?

I did not want to add a decorative graph to the homepage of a traditional blog. Every node in the graph should correspond to real public content, and every relationship should come from real bidirectional links and semantic structure. When readers open the homepage, the first thing they see should not be a list of “latest articles,” but the topic clusters, bridges, and spaces that have formed through the long-term growth of the knowledge itself. From there, they can enter a particular piece of content.

After a period of design, development, and iteration, I built a visual and interactive knowledge-network system that can be deployed as a personal blog. In its visible behavior, it follows physical and mathematical principles: nodes move naturally, relationships remain clear, and browsing and dragging stay smooth even when a large amount of content is shown at once. I will not discuss how these effects are implemented here.

This network is a field of stars made of knowledge. Every node is a topic or an article. The stars are connected to one another, and the strength of those connections changes with the strength of the relationships. You can drag them with the mouse, or select a node to enter its topic or article.

The knowledge network is not decoration. It is the primary entrance to the blog. If exploring through the star map is not your habit, you can still find content through search, text-based topic entrances, and ordinary links.

See the relationships as a whole, then enter the content itself.

Humans Are No Longer the Internet’s Only Readers

There is another change in the age of LLM agents: humans are no longer the internet’s only readers. Agents have, in practice, become users of the internet too. So this blog should not only be easy for people to read; it should also be possible for agents to discover, read, and understand it.

But I did not want simply to add a chat box to the homepage. Rather than having a model answer in my place, I care more about whether an agent can read my original words, understand the relationships between pieces of knowledge, and bring its citations back to the actual content. Humans and agents may enter through different doors, but what they ultimately see should be the same content and the same knowledge network. Material without evidence should not be presented as my view, and private knowledge will not enter the public blog.

People can enter an article from the knowledge network. The same public content is also available in representations that agents can read.

What Comes Next

I have not open-sourced the complete blog and knowledge-star-map implementation. In the future, I may offer the whole approach as a capability and service package, helping others who also want to build a personal knowledge network and a corresponding blog—from organizing their knowledge base and evolving their knowledge, to presenting and deploying the public network.

If you simply want to see what the system looks like today, visit:

https://www.pystone.net/notes/see-the-forest-knowledge-network-evolution-blog-system/

Do not search at first. Choose any topic you recognize, follow its real relationships for two steps, and then enter an article. You may understand more directly what I mean by “See the trees. See the forest.”

If you would also like to build a knowledge network and blog of your own, you can contact me by email to reserve a place in advance. At this stage, I provide the approach, deployment, and support as a paid service.

I Turned My Personal Blog into an Interactive Knowledge Network

Perrin Yong — Tue, 08 Sep 2026 06:57:19 +0000

For years, I have kept notes about what I learn and what I find useful in my work. As LLMs made knowledge easier to obtain, I started wondering whether maintaining those notes and publishing a blog still had value.

I think knowledge is also the network between ideas: how they support, limit, and challenge one another, and how new material changes an earlier understanding. An agent helps me with the workload, but the history of my thinking and the final judgment remain mine.

I built See the Forest around this idea, then made the public part of the network into a blog. The homepage is a constellation of real topics and articles. You can drag a node, follow its connections, or click through to the content behind it. Search and text navigation are also available.

The same public knowledge should be accessible to people and agents. I care about agents reading my original writing and bringing references back to it, not speaking on my behalf. Private notes stay private.

Try the live knowledge network: https://pystone.net/

Watch the short product video: https://www.youtube.com/watch?v=OS5vFOpmSu4

I'd appreciate feedback on whether the network helps you choose what to read, or gets in the way.

See the trees. See the forest.

Letting GPT-6 Improve My Blog: A Hands-On Test, and a Little About See the Forest

Perrin Yong — Fri, 04 Sep 2026 22:23:08 +0000

I recently got access to GPT-6 and put it to work on two rounds of improvements to my already-live See the Forest blog.

The blog had been through several rounds of design, development, and iteration. It had its own product direction and visual style. I wanted to see: What could it still find worth improving in a fairly mature project, and could it do so while preserving what made that project distinctive?

(The homepage before the improvements, with 558 public nodes and 846 relationships at the time. The star map itself was already an existing feature.)

After two rounds, more search results fit on the first screen, mobile articles had an expandable table of contents, and the stars and connections looked sharper. But not every experiment was worth keeping, and not every check passed. Here's what changed, and what happened along the way.

Of course, I also have a slightly self-interested reason for writing this: to ride a little of the interest in GPT-6 and introduce See the Forest, or 见林, the project I've been building. Someone has to get the word out about my own project. It might as well be me.

Why I Built See the Forest

I covered the origins of See the Forest in my previous launch article. The short version is this: Knowledge is fundamentally a network, with its parts connected to one another, much like neurons in the brain.

Ideas can corroborate one another, but they can also qualify or contradict one another. Knowledge from different fields may share patterns at some level. But surface similarities don't necessarily mean they are fundamentally the same. The network formed by these connections is itself part of our knowledge. Connecting what we know into a coherent understanding helps us put that knowledge to better use.

That's why I want to connect the notes I've accumulated and keep maintaining them with the help of an agent, letting knowledge “ferment” through these exchanges and produce new “sparks.” Which insights are worth keeping is still for me to decide.

As this knowledge network began to take shape, another thought occurred to me: This network should be something we can see. Being able to see the overall structure and relationships first, then step into a particular piece of content, would be really cool. That's how this blog came about.

This network is a field of stars made of knowledge. Each node is a topic or an article. The stars connect to one another, and those connections appear stronger or weaker according to the strength of the relationships. You can drag them with your mouse, or click a node to open its topic or article.

The knowledge network isn't decoration. It's the main way into the blog. And humans are no longer the internet's only readers: the blog also allows agents to read the same content and the same knowledge network.

The name See the Forest comes from the idea of seeing both the trees and the forest.

Round One: Start with Reading and Search

In the first round, it kept the star map's visual style and focused its changes on reading and search.

Take the search page first. The original “Search” heading was large, with plenty of empty space around it, and higher-ranked results used larger text too. Some snippets simply repeated their titles. The page looked imposing, but when you actually wanted to find something, that meant fewer results on the first screen and less room to compare them.

(Before: searching for “软件,” meaning “software.”)

It gave the result titles a consistent font size, reduced the space occupied by the heading area, and omitted snippets that exactly repeated their titles. The search ranking stayed the same, and the original content wasn't rewritten.

(The same search for “软件,” in the same 1440 × 900 window. A second row of results is now visible on the first screen.)

Next, articles on a phone. If you were halfway through a note and wanted to find a particular section, you had to scroll back and forth: the table of contents was visible on desktop but completely hidden on mobile. It also found that when a note began directly with a level-two heading, its first section was omitted from the table of contents.

(The same article in the same 390 × 900 window. Before on the left, after on the right. The right image shows the new table of contents expanded; it is collapsed by default.)

There is now an expandable table of contents on mobile, with links that jump to each section. The missing first section was fixed at the same time. The table of contents works even with JavaScript disabled.

It also added a retry button when search fails, links to topics when there are no results, and fixed a fade-in delay that was still present with reduced motion enabled.

These aren't particularly conspicuous changes, but they're useful. The star map may draw people in, but once they open an article, reading it should be comfortable too.

Round Two: Further Visual Refinements to the Star Map

After the first round, I asked it to make some further refinements to the star map's visuals.

One experiment made the image cleaner but also made the topic nodes harder to recognize, so it wasn't kept. The changes we kept focused on sharper stars and fine lines, and a clearer visual hierarchy of relationships when a node was selected.

First, the default view.

(The same network, window, and display conditions. Before above, after below. Open the full-size images to compare the edges of the stars and the fine lines.)

The smaller stars and some of the fine lines used to look a little blurry. After the changes, the edges are sharper and the fine lines are easier to make out. The colors are still the same, and the nodes haven't been rearranged, but the image does look more refined.

Now look at what happens when a node is selected.

(The same node, “游戏图形与运行时”—“Game Graphics and Runtime”—selected from the same viewpoint. Before above, after below.)

The gold connections used to look fairly similar in thickness. Now the main paths stand out more, while the others are lighter, making the hierarchy of relationships easier to see. The champagne gold, selection ring, and content panel on the right are all still there. These relationships already existed; this round simply made them clearer visually.

Beyond the Screenshots: How It Handled Problems

The pages had been updated, but had the changes affected the existing interactions? That called for checks beyond the screenshots.

The first round's full browser checks turned up two API errors and a failure in the check for settling after a drag. The API errors came from tests incorrectly comparing the production site's address against the local preview address. Correcting the tests cleared those errors, but the settling issue remained.

Rather than immediately blaming the changes, it repeated the same operations on the old version. Across the two rounds of investigation, the corresponding failures could also be reproduced there. One comparison also ruled out loading delays caused by optional ad requests. This helped distinguish problems in the tests, external interference, and existing product issues, rather than treating every error as a reason to change the code.

Round two's verification went beyond a few clicks: 10 sets of state comparisons covered different screen sizes, pixel densities, and light and dark themes, with node counts, relationship counts, camera positions, and label positions unchanged. Repeated-interaction checks passed 180 selections, 36 drags, and 6 navigation round trips. All 180 Atlas unit tests and 108 blog unit tests passed, along with the blog's browser checks covering 25 page views.

Still, two Atlas dynamics checks failed, involving motion-trajectory comparisons and the visibility of drag propagation. It didn't loosen the criteria; it left those issues in the report. Finding the same problem in the old version means we can't simply blame the new changes. It doesn't mean the problem has been solved. That is more trustworthy than “it should be fine.”

These were the results at the end of the two rounds on September 5, 2026. Round two had been committed locally and a signed preview was ready, but it had not yet been pushed or deployed. The after screenshots come from the local version. Test counts aren't user-experience scores, nor do they show that everything is faster on every device.

A Few Closing Thoughts

Looking back, these two rounds didn't turn the blog into something else, but they did improve specific parts of the experience. For a project that had already been through many iterations, I think that's worthwhile: it could continue within the existing product direction and find problems I hadn't specifically pointed out. As for how well the changes worked, the before-and-after comparisons are here for you to judge for yourself.

This experience doesn't answer “How much better is GPT-6 than other models?” I didn't compare models under the same conditions, and it had existing project materials and Codex tools to work with. The result can't be attributed entirely to the model itself. Deciding which changes to keep is still my responsibility.

That brings me back to why I built See the Forest. By making work more efficient, LLM agents make things feasible that used to be difficult to sustain over time. Maintaining notes, organizing their relationships, and presenting the knowledge network publicly used to involve a great deal of work. Now, with the help of an agent, I have a chance to bring these ideas to life, step by step.

As for what kind of knowledge network I want to build, and which parts of my understanding I want to share, those choices are still my own.

If you'd like to explore See the Forest, visit my blog at pystone.net. Pick a topic you recognize, drag it around, and click through. I'd also be happy to hear what you think, or what doesn't work so well for you.

See the trees. See the forest.

石心 Yong