I have just put a new page online: What I Think, and Who I Read.
If you have just found my work, you can see, in a few minutes, the main things I believe about trust, credibility and curation, and some of the people who I read or have inspired me most.
This small online resource came out of the last issue of TRUST-able, where I wrote about liquid content: taking what you have already written and turning it into something more useful than another article.
This is the first small example of that idea becoming real.
Doing it, taught me something useful. Ingesting your work into an AI-assisted knowledge base that distills it into info entities, does not mean it that the knowledge base has understood, checked and safely stored everything you wrote and had in mind.
You still need to vet, verify, take good care of that info-ecosystem and personally check what it has really understood, what has misinterpreted and what has missed altogether. It’s a place to start looking deeper into whatever you have given it to distill.
Why I Wanted To Do This
For somebody who has just discovered me online and wants to understand what I do, what I focus on, what I think, the only way to understand it, it’s through my articles. But there are now more than ninety TRUST-able issues totalling over 2k of content pages. The work is there, but it is spread across articles and time. Not very practical.
I wanted one page that could answer these questions without asking a person to spend ten hours reading my archive. Not a sales page. Not a biography. A simple page beside my homepage, where somebody can see the main ideas that guide my work and can come back and check them when it is useful.
A page designed to show where my ideas come from, where I wrote about them, what I read, rejected, got wrong, or who/what helped me change my mind. When you can see these things, you can understand how I got to my viewpoints and beliefs, not just what they are.
What I Actually Did
In April I started my own personal knowledge base. For four months I have been feeding it the articles I wrote and using it to find connections I had not seen when looking at each standalone issue.
The knowledge base is built with The Curator, the open-source LLM-wiki made by Dr. Tali Režun from Andrej Karpathy idea of letting a model compile documents into an interlinked knowledge base. I asked it to surface the recurring claims in my writing that are not in line with what most people think, then went through them one by one.
I wanted twenty viewpoints I could honestly say I stand for.
Some of those that had emerged were not strong enough. Others needed to be further verified or made more exact. For each one that I decided to keep, I went back to the original issue and checked it supported what the page said.
The first half became twenty things I think.
Then I wrote the second half myself: twenty-five people I read, or have read, and who have inspired me. I wanted to be transparent about who has helped form my thinking, and to say what I took from each and why he or she matters to me. I see this also as a part of the explanation of where my own viewpoints come from.
After the content was right, I worked on the information design aspect of this resource. I did extensive research to find examples for a specific design look that I had used in the past and fed it to AI. I then pushed to create several different alternatives, iterating the design process until I started to see what I was looking for. Tried alternatives and colour palettes, and prototyped until the page became as essential and functionally simple as I wanted it to be.
Finally, I created a sub-domain of my site (which I am in the process of re-designing right now) and made it go live.
Why This Is Worth Doing
An article is a journey. Somebody reads it once, perhaps keeps one idea, and then it disappears into the archive. This page / resource I have created stays there. I can update it as I keep writing. And when someone finds me for the first time, he does not have to open many old articles to understand what I think.
As I wrote above, the page also does something useful that a polished profile normally cannot do. It allows a reader to check me out: to see my viewpoints, how I think and reason, who inspires me, and where I have written about these ideas in more detail. It gives somebody a clearer picture of me before reading more or contacting me for something he has in mind.
This is why I think the liquid content concept provides an excellent example of how to use AI in an effective and very useful way. Create knowledge resources from your own existing ideas and writings, letting AI help you in finding, sorting and connecting work that you have already created.
How To Make One Yourself
If you keep a second brain or have been putting aside articles, notes, ideas, client work, bookmarks or research in a personal digital library of some kind the first step is not to build a database that has immediately everything you have ever saved, written or collected.
Ask: what useful question could the material you have written so far answer for the people you want to help?
Then collect only what helps answer that question. Then ingest it in Gemini Notebook or in your preferred AI to surface the possible claims, themes or connections, and treat what comes back as a first draft. Go back to the original source. Check names, dates and links. Remove what you cannot prove, what is no longer true, or what you do not really stand for.
Make something a person can use without you being there: a reading resource like Dr. Tali Režun did with his Field Notes, a map of your thinking territory, a decision tree, a short guide, a list with reasons, a comparison table or a directory.
Leave the evidence visible: your sources, criteria, what you left out and the date you checked it. These things show a reader that somebody verifiable has done the work.
The Next One in a Week
This resource I just built, What I think - What I Read, is the first attempt at applying the liquid content concept to my own material.
The next one out will be a kind of decision tree, in the form of a selector for experts who want to demonstrate what they know without simply writing another article, LinkedIn post or generic video. It helps them choose among twenty-five curated-content formats on the basis of what he knows, how much time he has, and the proof of expertise he wants to give.
Both this one I shared today and the next resource show concretely something most experts tend to not consider. Articles are not the only road to establish authority and credibility in a field. You do not have to publish in the same format everybody else does. You can choose an editorial shape that makes your judgment visible and is actually useful to the people you want to help.
This first resource shows what I think and who has inspired me.
The next one will help you choose a better way to show what you know.
How This Connects to the Rest:
Here is the article from where I first crystallized the basic idea of liquid content:
Here instead the first instance of me writing about that same concept, while reviewing the “idea architecture”:
Here the foundational concept of building knowledge infrastructures rather than just articles:
Paradata
Why / Intent:
I wanted to follow up on ✪ Liquid Content: From Articles to Resources People Keep Using - #91 and demonstrate in practice how that concept could be actually used.
Process:
I made this resource from ninety issues of TRUST-able, using my personal knowledge base to pull out possible viewpoints and connections. I worked with Claude and ChatGPT during the exploration and verification, but every one of the twenty positions was checked against the published issues before it stayed. The twenty-five entries on the people I read were written by me.
The work sent me back into the knowledge base itself. I corrected seventeen errors, recovered missing provenance, and learned again that an AI system needs periodic human maintenance. The resource will change as my views change or I find better sources. It is a page I intend to keep useful.
While I was doing this, without planning to, I began to see some weak sides of my knowledge base. The sharpest discovery was about links. Across 1,361 wiki pages, it had preserved only five external links. The links I had used in my articles had not been carried into the knowledge base. They were still in the original published articles, but they were no longer there to support the claims in the wiki.
So I went back into 208 PDFs of my own work and recovered 3,104 URLs. I checked eighty of the most important ones one by one and put them back into the wiki. This same check found seventeen errors. The knowledge base helped me find these problems. But it could not decide whether a link, a name or a claim was right. I had to go back to the sources myself and verify.
Time:
for resource: “What I think, what I read”
Research and brainstorming: 1.5 hrs
Analysis and verification: 3.5 hrs
Design and prototyping: 3.5 hrs
Critical review and strategy change: 1 hr
Design execution + editing + refinements: 2 hrs
Publishing online: 1 hr
Tot time: 12.5 hrs
for this edition of TRUST-able
Drafting of this article: 1.5 hr
Editing and revising: 1.5 hrs
Formatting: 0.5 hrs
Cover design: 1 hr
Total apx time: 17 hrs
Tools:
ChatGPT Astra and Terra High
Claude Opus 5 / Sonnet 5
Cover image: What I think, who I read. Concept and original photos by Robin Good. Image generation by ChatGPT Image 2.5.
What I Offer
The Curation-Based Authority System
55 minute video workshop illustrating how to build authority and credibility by creating high-value curated content formats, rather than original essays.Make Your Expertise Buyable
private credibility and authority audit
Follow a path with a heart.
From Koh Samui (TH)
Robin Good
Very well-organized and easy-to-read research. I'd love to see where this goes. Great job.
Nice! Looks quite good, Robin.