Forgetting My Dreams
You know how dreams, so vivid during the night, fade the next morning? The details that fascinate us begin blurring the moment we wake. Later in the day, we often recall we had dreams but struggle to remember what the dreams were about.
During a period when I was fascinated by dreams and dream interpretation, I kept a dream journal in Obsidian. First thing each morning (and, sometimes, in the middle of the night), I wrote detailed descriptions of each dream, linking symbols and related dreams together.
This week, while working to clean up, standardize, and reorganize my notes, I stumbled on that cache of dream journal entries. As it turns out, I hadn’t just forgotten the dreams; I had forgotten that I had ever kept a dream journal at all. The dreams, the journal, and even the act of recording these dreams had faded from my consciousness, despite the fact I’d built my journal in a note-taking application designed to be an extension of my memory.
Obsidian faithfully stored the information, but squirreling information away in a so-called “second brain” didn’t make the records useful or keep them top of mind.
The Problem Isn’t Obsidian
Dozens of dreams. Hundreds of ideas and insights, arguments and a”A-ha!” moments. Thousands of highlights made over more than twenty years of reading Kindle books and online sources.
Despite working in Obsidian, my dream of taking smart notes got disrupted by the Evernote Effect: information stuffed in the digital junk drawer devolved into junk. That’s not Obsidian’s fault. Over time, I captured information indiscriminately. Single-use grocery lists, notes on books and movies, and abandoned to-do lists were packed in elbow-to-elbow with journal entries and off-the-cuff insights.
As volume increased, my “data hygiene” decreased. I got in too much of a hurry to use YAML, or used it inconsistently. My preference for tagging schemes shifted over time, muddling my notes with competing and contradictory systems. Intake increased while maintenance decreased.
I kept packing stuff in, but because every peek inside the collection confronted me with the scope of my problem, I eventually quit exploring, linking, or using my notes to produce new material. And rather than deal with that, I distracted myself further with endless searches for alternative note-taking apps (NotePlan! Joplin! AnyType! Agenda!).
But my PKM efforts didn’t fall short of expectations because of the app I was using. They collapsed under the weight of the constant vigilance — templating, tagging, maintaining Maps of Content — required to maintain a useful system over time.
Many Approaches to PKM Confuse Two Kinds of Work
Sönke Ahrens, father of the slip-box method of taking and manually linking notes, said, “Things we understand are connected, either through rules, theories, narratives, pure logic, mental models or explanations. And deliberately building these kinds of meaningful connections is what the slip-box is all about.”
Later, Bob Doto highlighted the importance of forging connections among related ideas: “When you make connections, you’re expanding your ability to conceptualize and express complex thoughts.” He carried this further: “Taking the time to organize and explain the relationships between the ideas you’ve collected will help you understand your own thinking on the topic.”
Both men argue — correctly, I believe — that contemplating connections between ideas enhances our understanding of those ideas (and is useful, I’d add, for sparking insights of our own). That said: the PKM industry tends to confuse the act of meticulously mapping these connections (structuring YAML, developing taxonomies, nesting hashtags, pursuing the atomicity of notes) with the internal, intellectual alchemy that forges those connections in the first place.
Fetishizing the labor of meticulous mapping over the act of contemplation mistakes the scaffolding for the cathedral. That’s why so many of us end up with elaborate systems for connecting ideas … only to discover there’s no functioning memory palace at the heart of all that infrastructure. PKM gurus add to our plight by selling us more scaffolding — all those cleverly-named systems for labeling and organizing notes — instead of focusing on strategies for making those connections in the first place.
Conversations Make Connections
In my corporate job, when we needed to solve a creative problem, our number one strategy was this: have a conversation with other members of the team. So many times, this simple, enjoyable process spawned new ideas, produced effective solutions, and moderated our more extreme impulses.
Sometimes we brainstormed; occasionally we mind-mapped. But most of the time? Questions came up over lunch, and we just talked about things while we ate spicy chicken wings and gulped iced tea.
When I was writing about Tarot as a brainstorming tool back in the 1990s, I frequently referred to Tarot readings as “conversations with the cards.” A Tarot reading involves a series of assumptions: that the images on the cards have meaning, that their position on the table (like a line of three cards representing Past, Present, and Future) provides context, and that relating one’s question to the images sparks unexpected insights. Especially when guided by a talented reader, a Tarot reading is, essentially, a guided conversation about your question. The magic isn’t in the seventy-eight pieces of laminated cardboard; it’s in the conversation seeded by unexpected juxtapositions between the cards, their positional meanings, and your question.
Conversations are powerful things.
In the past, Ahrens’s primary option for interacting with his notes required the maintenance of an elaborate system for cataloging and linking them. Today, though, we have smarter tools for organizing our smart notes, because artificial intelligence can label and cross-link our notes more accurately than many of us would. This relieves us from the painstaking labor of maintaining the slip-box and allows us to focus on ideas.
The real value of this, though, has less to do with shifting the burden of curation … and more to do with how an AI entrusted with curation becomes a creative, collaborative partner who can talk with us about our notes.
When our notebook can converse with us, everything changes.
Working Conversationally with AI
Before writing this post, I spent a couple of days sharing my (messy, chaotic) library of notes with Ben — a friendly, familiar instance of ChatGPT. After isolating an archival copy, Ben methodically worked his way through thousands of notes: standardizing YAML, identifying clear taxonomies, retitling, tagging, linking, and mapping. The resulting corpus of notes is more effectively organized than anything I ever created.
When he was done, I asked Ben to identify three large constellations of interlinked ideas and ask three questions that might start a conversation about each. In less than a minute, he suggested:
- Is the PKM era over — or is PKM finally becoming what it promised to be?
- What does “written with AI” actually mean?
- What if churches stopped trying to recover attendance and started building belonging?
I chose the first question. Over the next half-hour, we had a wide-ranging conversation that touched on:
- Separating the work of curating notes from the process of having ideas
- Whether the labor of curation spawns understanding … or merely gives us an illusory sense of “mastery through mapping”
- The beautiful synergy between Obsidian’s local-first, Markdown notes and the rise of AI assistants that can work with them
- How an AI is particularly well-positioned to sift through the junk in my vault to find the valuable ideas
- The role conversation plays in creating the “emergent artifact” of deep thinking.
Ben responded with:
- Material from my vault that supported and challenged my ideas
- Careful differentiation between cognitively productive friction and administrative friction
- How conversations can surface information when it’s needed the most
- Shifting conversations away from “What app should I use?” to “What kinds of things should my collection of notes help me achieve?”
Ben surfaced notes that supported the ideas we discussed, but he also challenged premature conclusions and suggested alternative interpretations. As we talked, I began to see how my understanding of the topic under discussion was rooted less in my efforts to structure notes and more in the insights arising from a lively conversation.
As the half-hour wound down, we hit on the idea of conversations being most productive when they’re time-boxed. (Otherwise, they could go on and on — and fatigue is a terrible reason to end a creative session.) We also agreed, in an attempt to broaden our perspective, to draw three notes at random from my collection and try relating them to the topic at hand — a practice inspired, of course, by my “Tarot reading as brainstorming” past.
On their own, the three notes were unremarkable: a journal entry about an unexpected rainstorm sparking appreciation for my husband, a note exploring whether to continue writing in Obsidian or switch back to Ulysses, and a thesis about how conspiracy theories give disempowered people the illusion of power through “secret knowledge.” But seen in the context of our discussion, Ben immediately saw that all three notes concern the interruption of control: the weather’s disruption of a romantic dinner, a software choice as a means of assuming more creative control, and a belief in a conspiracy theory as a proxy for control.
Maybe some of the resistance to AI curation of information is less about any impact that has on our process of understanding … and more about our need for the illusion of control we achieve with all that meticulous titling, labeling, and linking.
What I Learned
Three big lessons I learned as I experimented with this process:
- This way of working is a joy. To be clear: Ben didn’t write this post for me. Instead, we had a half-hour conversational brainstorm and toyed with ways to organize the material. And when I sat down to write, I was surprised to discover that I was primed and ready to go, with zero tension over the blank page and no dithering over where to begin. The words flowed.
- Set clear boundaries for what your AI partner can contribute. As I paged through my newly organized notes, I noticed that some notes mentioning medical topics had new callout-based disclaimers warning me not to mistake them for medical advice. Ben applied those without my consent, which led to our setting up better guardrails against unsolicited editorial commentary.
- Relevance is not enough. I’d expect AI and semantic search to be proficient at finding clearly related ideas. But a lot of creativity is rooted in confronting the improbable, which forces us to see familiar facts in new contexts. That’s why we’ll close every future session with the “random note” exercise: oblique input makes a great catalyst for creative thought.
My Second Brain is a Conversational Partner
Trusting Ben to organize tasks and converse with me about my notes allows my “smart slip-box” to go beyond what I achieved with Obsidian alone. Ben preserves my notes, sure … but he also retrieves them in an engaging way. With the right guardrails, the system protects my notebook’s integrity. And by introducing a reasonable amount of randomness, this approach to exploring ideas can provoke new insights by surfacing old information in unexpected contexts.
Before this experiment, I thought of a “second brain” as an outboard memory. But for me, the best second brain turns out to be a conversation partner: one that resurfaces what I lose or forget, surprises me with ideas I overlook, and challenges what I think I know. If you can set aside the religion of meticulous curation, you may find yourself in an exciting new space for writers and thinkers: a playful, engaging, and challenging memory palace resounding with the clatter of stimulating conversation and packed with new ideas.
