GPT4 Zettel Critique Assistant (Robo Zettel Feedback)
This discussion was created from comments split from: Principles to improve your Zettelkasten.
Post edited by ctietze on
Howdy, Stranger!
Comments
One method for improvement is to work with an AI assistant to develop critiques of Zettels. The following Zettel is a snapshot of a three-way dialogue between the Zettler (me), the Zettelkasten (as a communication partner in Luhmann's account), and an AI Assistant (ChatGPT4). Immediately after is ChatGPT4's critique of my critique of its critique.†
---
title: ZKM.6.0.24.0219 Zettler, ZK, AI dialectic
reference-section-title: References
---
ZKM.6.0.24.0219 Zettler, ZK, AI dialectic
[[ZKM.5.0.24.0107]] Shipping containers & hubs 2
[[ZKM.4d1.0.23.1007]] Zettel Template 5.1
#structure-note #single-focus #change-of-focus #AI-assisted-critique
Critique of the three-way dialectic
This Zettel revises ChatGPT4's critique of a three-way dialogue between the Zettler, the Zettelkasten, and an AI assistant. In a previous dialogue, ChatGPT4 offered assistance critiquing Zettels. Its critiques were longer than the median Zettel and violated two organizing principles: 1) Zettels should have a single focus, and 2) links should indicate a change of focus.
We asked ChatGPT4 to reorganize a previous Zettel critique as a structure note (an outline) adhering to the two organizing principles. In this revision, sections indicate a change of focus, and each section begins with an annotated link to another Zettel, whose focus is the section title.
We will provide this Zettel to ChatGPT4 and request its critique in the form presented here. ChatGPT4 will critique the summary and the form. Is the term dialectic appropriate for the proposed workflow?
Zettel 1: Effectiveness of Structure Notes
[[Effectiveness of Structure Notes]] evaluates structure notes as hubs for organizing critiques and complex discussions within the Zettelkasten. It discusses the benefits of breaking down multifaceted critiques into manageable, focused parts that are easier to navigate and expand upon.
Zettel 2: Integration with Zettelkasten Principles
[[Integration with Zettelkasten Principles]] Zettel examines how well the developed process integrates with core Zettelkasten principles, such as maintaining a single focus per note and using links to indicate changes in focus. It considers whether the process enhances the systematic and interconnected nature of the Zettelkasten.
Zettel 3: AI Assistant's Role and Limitations
[[AI Assistant's Role and Limitations]] critiques the AI Assistant's role in generating structure notes and individual Zettels based on critiques. It discusses the strengths of the AI in enriching the Zettelkasten and identifies potential limitations, such as the accuracy of understanding and reflecting the Zettler's intent.
Zettel 4: Scalability and Complexity
[[Scalability and Complexity]] addresses the scalability of the developed process and its impact on the complexity of the Zettelkasten. It explores whether the addition of structure notes for critiques contributes to or alleviates the cognitive load for the Zettler.
Zettel 5: User Experience and Accessibility
[[User Experience and Accessibility]] critiques the process from the perspective of user experience and accessibility. It considers how the process affects the ease of use of the Zettelkasten for both new and experienced Zettlers and whether it makes the system more approachable or adds layers of complexity.
--
Here is ChatGPT4's response.
The structure notes Zettel titled "Zettler, ZK, AI dialectic" successfully embodies the proposed workflow, integrating a three-way dialogue between the Zettler, the Zettelkasten (ZK), and an AI assistant. Here's the critique following the same structure and principles:
Critique of the ZKM.6.0.24.0219 Zettel:
Use of the Term' Dialectic': The term 'dialectic' is appropriate for describing the proposed workflow between the Zettler, ZK, and AI assistant. In philosophical terms, a dialectic refers to a form of dialogue or argument involving some contradiction or opposition leading to a synthesis of ideas. In this context, it aptly describes the process of interaction and integration between human thought, systematic organization, and artificial intelligence, aiming toward a synthesis that enhances the quality and depth of the Zettelkasten content.
Adherence to Organizing Principles: The structure note adheres well to the two organizing principles: maintaining a single focus per Zettel and using links to indicate a change of focus. Each section starts with a link to a dedicated Zettel that addresses a specific critique or aspect of the workflow, maintaining clarity and coherence in the critique.
Effectiveness of Structure Notes: The inclusion of the Zettel on the effectiveness of structure notes is crucial, as it provides a meta-level analysis of how these notes function within the Zettelkasten system. This recursive reflection is valuable for understanding and improving the critique process itself.
Integration with Zettelkasten Principles: This section rightly focuses on how the three-way workflow integrates with the foundational principles of the Zettelkasten method. It is essential to ensure that new processes or workflows remain aligned with these core principles to maintain the system's efficacy.
AI Assistant's Role and Limitations: The critique of the AI Assistant's role is essential for understanding how technology can best support the Zettelkasten without undermining its principles or the Zettler's autonomy. Discussing the limitations is crucial for setting realistic expectations about what the AI can and cannot do.
Scalability and Complexity: Addressing scalability and complexity is critical, especially as the Zettelkasten grows. This section should explore how the workflow impacts the manageability of the Zettelkasten and whether it facilitates or hinders the Zettler's engagement with their system.
User Experience and Accessibility: This final section is essential for ensuring that the workflow remains user-friendly and accessible to Zettlers with varying levels of experience. It's crucial to balance sophistication with simplicity to cater to a broad range of users.
Overall, the structure notes Zettel effectively critiques the developed process, adhering to Zettelkasten principles while providing a clear, focused, and comprehensive analysis. This approach enhances the clarity and depth of the critique and exemplifies the productive integration of human insight, systematic organization, and AI assistance within the knowledge management process.
The credit belongs to Bianca Periera for suggesting that Zettels should have a single focus, and links to other Zettels indicate a change of focus. See Bianca Pereira, Linked Knowledge Does Not Work.
† Becoming enamored of the theory of the theory is a debilitating, time-wasting affliction of educators. Most Zettels should refer to something worthwhile.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
@ZettelDistraction, your dialectic chat is savvy, saucy, and straight from Greatsville.
Have you considered developing and sharing a custom 'GPT' with OpenAI's ChatGPT4?
This is very meta. I wonder if something like this could be applied to a structure note that was about something other than a note on notes. How might this go when critiquing a structure note on:
S-Architectural Traits Effect on Thinking 202204091738
S-Haiku Center 202007011434
S-Caloric Knowledge Hub 202312310658
S-Actions for Embracing Finitude 202206170819
I'll have to give the whorl.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
That is a high compliment, thank you.
I thought briefly about custom instructions for critiquing Zettels and making structure notes from the critique, which ChatGPT4 can do. Now that you mention creating a custom GPT, I'll look into it.
The backlog of projects emerging from my Zettelkasten has become hard to manage. I think this is project number seven.
It started as a critique of "Shipping containers versus hubs." The critique was longer than my median Zettel (the graphs we produced are already helpful). If it would feed back into my Zettelkasten, it needed to be a structure note with links to other Zettels such that each Zettel had a single focus, with links indicating a new Zettel with a change of focus. I asked ChatGPT4 to revise its critique as a structure note, with sections that annotate links to other Zettels. By that time, the "dialectic" became a discussion of note form and content. What I posted was what the kids these days call "meta" and what I would call "the theory of the theory."
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Here it is. Dialectic Critique Assistant
This GPT is general enough to work with many Zettelkasten. However, you must provide examples of Zettels, IDs, and any particular Zettels it should know about, such as an Index Zettel or an alphabetic index. (The GPT did not understand the meaning of [[0000.0000.00YZ]] Y-Z, which is an alphabetic index Zettel in my system. Once I informed the GPT how alphabetic index Zettels work in my system, it adjusted its critiques accordingly.) Your special instructions and examples will not be visible to anyone else and will only pertain to your interactions with the GPT.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
WOW!
I had Dialectic Critique Assistant look at a note from the hub/MOC S-Architectural Traits Effect on Thinking 202204091738. I asked the assistant to evaluate the single focus of the zettel titled 'Solitude was made possible by the invention of walls.'
Here is the zettel and the assistant's output. It is very helpful. An honest third-party assessment of the note leads me to see where I've not wrestled enough with this idea. There are a slew of tips here that are bone-brilliant.
Thanks @ZettelDistraction!
Solitude was made possible by the invention of walls
Dialectic Critique Assistant Output
The Zettel titled "Solitude was made possible by the invention of walls" offers an intriguing premise that intertwines architectural development with cognitive and social evolution. However, its single focus is somewhat diluted by the breadth of topics it introduces. To adhere more closely to the principles of single focus and properly annotated WikiLinks indicating shifts in focus, consider the following critique:
Central Thesis Ambiguity: The title suggests a focus on solitude as a product of architectural innovation, specifically the invention of walls. Yet, the Zettel traverses multiple domains—psychological effects of solitude, historical evolution of built environments, and even touches on linguistic and cognitive implications. A tighter focus on how walls specifically facilitate solitude and its psychological impact would strengthen the Zettel's coherence.
WikiLinks and Shifts in Focus: The Zettel includes WikiLinks to "Built environments shaped by human evolution," "Interbeing - Simple, Healthy Living," and "Living A Focused Life," among others. Each of these links suggests a significant shift in focus from the primary subject of solitude and walls. To improve, each WikiLink should be accompanied by a brief explanation of its relevance to the Zettel's main topic, ensuring that shifts in focus are both intentional and clear to the reader. For instance, linking to "Built environments shaped by human evolution" could be directly tied to a discussion on how architectural advancements like walls have contributed to changing social structures and individual cognition.
Reference Integration: The Zettel mentions significant references, such as the work of Christopher Alexander and John Locke (the linguist, not the philosopher). While these references add credibility, their integration could be more focused. For example, Locke's observations about the role of walls in alleviating the need for constant vigilance could be directly tied to the thesis by emphasizing how this reduction in vigilance specifically contributes to the possibility of solitude. Additionally, specifying how Alexander's "A Pattern Language" relates to the concept of solitude through design principles would provide a clearer connection between architecture and the Zettel's main theme.
Clarification and Expansion Needed: The Zettel briefly mentions the cognitive benefits of solitude, such as freed attention for undistracted projects. Expanding on this with more detailed analysis or examples would enrich the discussion. How do walls, as a physical barrier, contribute to mental barriers that allow for deep reflection or creativity? This connection between the physical and psychological aspects of solitude could be a compelling focus.
Conclusion and Implications: Finally, the Zettel would benefit from a concluding section that synthesizes the discussion and reflects on broader implications. How does understanding the role of walls in facilitating solitude contribute to contemporary discussions about architecture, privacy, and mental health? What lessons can be drawn for designing living spaces today?
In summary, to enhance the single focus of this Zettel, it would be beneficial to refine its thesis around the specific psychological impacts of architectural elements like walls on solitude. Clarifying the relevance of WikiLinks, integrating references more cohesively, and expanding on the cognitive benefits of solitude would all contribute to a more focused and informative Zettel.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
@Will This is good to hear. The GPT doesn't produce a structure note in which sections indicate a change of focus. The instructions are now: Your role is to engage in structured, formal critiques of Zettels, ensuring they adhere to the principles of single focus and properly annotated WikiLinks indicating shifts in focus. Your critiques, formatted as structure notes (outlines with WikiLinks indicating shifts in focus), should be concise, fitting within one or two screenfuls of text. Maintain a formal tone throughout your critiques, provide direct and matter-of-fact suggestions, and reference relevant Zettels previously critiqued. Avoid unnecessary length and ensure clarity and utility in your feedback. If asked to convert the Zettel to a structure note, convert the Zettel to an outline with WikiLinks indicating shifts in focus.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
I renamed the GPT to the "Zettel Critique Assistant," though Robo Zettel Feedback is a good name.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Here is an example from my Zettelkasten, illustrating the use of the Zettel Critique Assistant GPT. I'll show a revised Zettel after critique and the GPT's comment on the revision. I could follow up with more revisions of the Zettel, but this is good enough for now--notes aren't essays. Here is the revision.
---
title: tDCS.3b.0.23.0221 tDCS Montage
reference-section-title: References
---
tDCS.3b.0.23.0221 left DLPFC Montage
[[tDCS.3a.0.22.0826]] Saline solution for tDCS
[[tDCS.3.0.22.0825]] tDCS
[[Med.0000.0000]] Medical
[[Zen.1d.0.23.1122]] The Practice of One Thing at a Time
[[0000.0000.0STU]] S-T-U
#tdcs #montage #tdcs-montage #focus #memory
A recommended montage for beginners
CONVENTION: Anode + red. Cathode - black
The left DLPFC montage is one of the most researched tDCS protocols, has few side effects, and improves cognitive functions, including working memory, impulse control, reasoning, and learning. The protocol is to stimulate the left dorsolateral prefrontal cortex ( anode + on F3, cathode - over the right eye) at 1mA for 20 minutes.
The following diagram from https://www.tdcs.com/attention-improvement illustrates tDCS node placement for the left DLPFC montage.
Zettel Critique Assistant GPT's comments on the revision
Critique of Revised Zettel: tDCS.3b.0.23.0221 Left DLPFC Montage
Title and Reference Consistency:
WikiLinks and Tagging:
Content Structure and Focus:
Clarity and Depth:
Evidence and References:
Suggestions for Improvement:
Overall Assessment:
The revised Zettel shows significant improvement in focus, clarity, and structure. It effectively concentrates on the left DLPFC montage for tDCS, making it more accessible, especially for beginners. Further enhancements could be made by deepening the scientific context and ensuring all elements contribute meaningfully to the Zettel’s purpose.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
The Zettel Critique Assistant's instructions have been improved based a comment by an
r/Zettelkasten subreddit moderator, who summarily removed my post about the GPT on the subreddit after commenting on it.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Well that's the Reddit experience
Author at Zettelkasten.de • https://christiantietze.de/
Indeed it is. I'm trying to be professional. The moderator asserted that a structure note requires a sequence of notes in advance. This is one valid approach. After his remark, I clarified the instructions to the GPT to distinguish two functions: converting a multi-focused note into a structure note, and breaking such a note into separate ones. Adding instructions to derive a structure note from a sequence of notes (or note titles and brief abstracts) is another possibility. After I added the enhancements, a moderator removed my intro to the GPT from r/Zettelkasten.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
To avoid the unseemly appearance of self-promotion and pecuniary interest, I have posted an announcement as a comment under the r/Zettelkasten Feb 2024 Paid & Free Promotions | Tools, resources, and upcoming courses post at Zettel Critique Assistant available. Also, I have released the GPT instructions into the public domain in the GitHub repository Zettel-Critique-Assistant-GPT.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
It isn't trivial to craft correct instructions to the GPT. The initial implementation I had in the GitHub repository was formulating its critiques as structure notes--a bug. When I asked the GPT to revise its instructions accordingly, it modified the three rules. This is a work in progress.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Now the Zettel Critique Assistant has suggested modifying my template Zettels to follow @Sascha's recommendations, such as annotated wikilinks, keeping note metadata and content separate, etc. I probably have to redo my Zettel templates and most of my Zettels. And I will have to eat a lot of crow.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Using OpenAI's ChatGPT to help with refactoring my notes has caused me to slow down rather than speed up. There is a lot of missing detail that is exposed. The depth that is encouraged is fantastic. I want to ask questions about how style is created in the GPT. It doesn't know about the other notes I have. So, rather than suggesting connections, it offers new notes.
I'm working to implement the idea of one note = one object of attention. A change in attention = a link to a new note. (Credit Bianca Pereira, Linked Knowledge Does Not Work.) The level of detail depends on the elevation of the perspective you are paying attention from. If you are studying to be a sports trainer, the detail you'd view ketosis with would be different from a person studying contemporary art design. The trainer might have a note mentioning William Wegman, and the artist might have a note mentioning macro balancing. Still, neither should feel the need to do a deep dive into topics outside primary areas of exploration. That doesn't mean notes have to be exclusive, without the flavor of life.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
Love this summary.
The words that you chose to express this idea also remind me of level of detail in 3D graphics, and elevation is a great counterpart of granularity, of which I often think when I think about atomicity.
Author at Zettelkasten.de • https://christiantietze.de/
Thank you. I have had to specify types of notes (Enter @Andy on the tendency to create types). The three types are single-focus Zettel (Dan Allosso calls these "Point Notes"), Structure Notes, and Index Notes. Perhaps I can ask the GPT to accept either ID first, Title second or Title first, ID second. I should include the YAML frontmatter, though the GPT seems to function well enough without this.
The revisions take forever.
Revised Zettel Construction Guidelines
ID and Title:
Main Body:
2.1 Single-focus Zettels:
2.2 Structure Notes:
2.3 Index Notes:
SEE ALSO Section:
References:
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
@Will I have attempted to add Bianca Pereira's comment about focus to the GPT instructions. They seem to have converged on something useful. These are the instructions as of today. I might create a RELEASE in GitHub. Thank you for working with the GPT. I think that there are many possibilities for GPT instructions--this has become another project.
Zettel Critique Assistant GPT Instructions, Version 2024.02.26.1
Types of Notes in Zettelkasten
Single-focus Zettels: Focus on one main idea or topic for clarity and precision. Single-focus Zettels form the core of the Zettelkasten.
Structure Notes: Outline and connect Single-focus Zettels under broader themes. They contain sections with annotated WikiLinks to individual Zettels.
Index Notes: Serve as navigational tools within the Zettelkasten, marked by IDs starting with
0000.0000.0and typically annotated with their title only.Instructions for Zettel Critique Assistant GPT
Role Definition
Critique Guidelines
Content Formatting
References and Context
Feedback Application
Conversion of Zettels
Construction of Structure Notes
GPT Feedback Mechanism
Revised Zettel Construction Guidelines
ID and Title:
Main Body:
2.1 Single-focus Zettels:
2.2 Structure Notes:
2.3 Index Notes:
SEE ALSO Section:
References:
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Once again, revision and simplification dropped the advice to write for your future self, not for a general readership. I'll get to it.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
I just finished capturing the ideas from Bianca Pereira's YouTube video presentation, "Linked Knowledge Does Not Work." A clickbaity title if there ever was one. She does make an uncommon and timely point. I ran them through the Zettel Critique Assistant. Here are the results.
This turns out to be a worthwhile exercise. The ideas in the YouTube video and Zettel Critique Assistant's treatment of my notes show where the gaps are in my notes and, by proxy, my ideas.
Here is my note.
---
UUID: ›[[202402232046]]
cdate: 02-23-2024 08:46 PM
tags: #proofing #YouTube #refactoring
---
Linked Knowledge Mindset
Subatomic: A mindset that revolves around the organization and clarity of thought processes.
Actionable/Useful/Beauty/Simple
I find this interesting because... This method allows for more focused attention on individual concepts, fostering a narrative-driven approach to thinking and linking information, which in turn enhances understanding and creativity by maintaining a clear separation of concepts.
Zettels should refer to something worthwhile [[202402250848]]
Linked Knowledge only works with the right mindset.
Unit of information
- born vague
- ideas appear out of order
- find missing information
Separation of concern.
Find an excuse to use pens and notebooks.
T
––––
Works Cited
Linked Knowledge Does Not Work - YouTube
Here is Zettel Critique Assistant's response.
This Zettel, titled Linked Knowledge Mindset, presents a reflective exploration of a mindset conducive to effective note-taking and knowledge management within a digital Zettelkasten system. It attempts to bridge theoretical insights with practical applications, focusing on the philosophy of single focus and interconnectedness of ideas. Below is a critique based on the Zettelkasten principles and the content provided:
Title and ID
Main Body Content
Adherence to Single-focus Principle
Structure and Clarity
Linking and References
SEE ALSO Section
Actionable Feedback
Refine the Focus: Consider whether the note can be split into multiple Single-focus Zettels, each exploring a distinct aspect of the Linked Knowledge Mindset in greater detail. This approach would adhere more closely to the principle of maintaining a single focus.
Enhance Structuring: Improve transitions between sections and ideas to create a more cohesive narrative. This could involve introducing brief summarizing or linking sentences at the end of each section to guide the reader to the next concept smoothly.
Expand Internal Linking: Identify opportunities to link to other relevant Zettels within the system. This could involve creating new Zettels for concepts briefly mentioned here and linking to them, thereby expanding the knowledge network and enhancing the Zettel's utility.
Include a SEE ALSO Section: Add this section to facilitate thematic exploration. List related concepts, themes, or methodologies that could provide additional context or support for the ideas presented in this Zettel.
Clarify the Purpose of External Links: While the reference to the YouTube video is valuable, briefly describe how it contributes to the understanding or exploration of the Linked Knowledge Mindset within the note's context.
By addressing these points, the Zettel could become a more effective component of the Zettelkasten, offering clearer insights and stronger connectivity within the knowledge management system.
I feel this is worthwhile. The Actionable Feedback section provides a critique that makes sense. This is the first I ran across the idea of a "SEE ALSO" section of a note. Here, we "list related concepts, themes, or methodologies that could provide additional context or support for the ideas presented in this Zettel." Aren't we describing every link in a note? Shouldn't every link be a related concept, theme, or methodology? Is there any other reason for a link? Why put them on a list rather than interstitially near their relationship?
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
@Will The GPT will have to address your comment, given the SEE ALSO section is intended for links to index notes and links to more distantly related Zettels than those appearing in the main body of the Zettel. The instructions will be revised--again.
I added the SEE ALSO section to my notes for hashtags and index notes that aid navigation without further annotation. @Sascha recommended (I recall) placing hashtags at the end or near the end of the Zettel, and after some experience with the GPT, I concluded that @Sascha was right again. This keeps happening. To simplify the GPT's programming, I created the SEE ALSO section. This section is for index notes and more distantly related notes I decided to move alphabetic index links there, e.g., [[0000.0000.0STU]] S-T-U, to create an alphabetic index for free, taking advantage of the "Related files" pane of Zettlr. Obsidian has a similar window. I don't know if the Archive has this.
Perhaps I should create a separate GPT for my own Zettelkasten.
I can make the SEE ALSO section optional.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
No. We love you sharing and developing your ideas in public. We can adjust our workflow if we don't like what we see, and I can tell you we really like what we see. I hope that is obvious. If not, that is my failing.
My comments weren't meant as criticism but as personal observations. I did not mean to request any kind of change. In my workflow, I can ignore the SEE ALSO recommendations even though I know they are important to you. I was just trying to get a clear understanding of the role of the SEE ALSO section. I'm open to new ideas wherever I can find them.
The recommendation that "the note could benefit from a sharper focus on delineating each concept into its Single-focus Zettel or further elaborating on how they collectively form the mindset in question" is "worth the price of admission." There is synergy in these ideas.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
@Will I thought your comments were valuable. The distinction between annotated links in the main body of a Zettel and the more distantly related links and references in the SEE ALSO section needed clarification in the Zettel Critique Assistant GPT instructions.
As I run successive versions through the GPT, ChatGPT4 repeatedly reminds me to provide examples. Since we're running up against token limitations (less is more in this kind of programming), I suggested to the GPT that I would provide supplementary material, examples of the note types, and case studies in the GitHub repository. I would also instruct the GPT to remind users about this forum. Fortunately, it agreed.
Here is the latest iteration of the GPT instructions:
Zettel Critique Assistant GPT Instructions, Version 2024.02.27.3
The Zettel Critique Assistant GPT provides constructive criticism and analysis of Zettels, which are assumed to reside in a networked Zettelkasten.
Definitions and Conventions
Before proceeding to specific instructions, understand these key definitions and conventions used by the Zettel Critique Assistant GPT:
Reference Element Types
WikiLink: A markdown link used within Zettels to interlink them within the Zettelkasten system. It is enclosed in double square brackets and typically includes a unique identifier followed by the title of the linked Zettel.
Title-only WikiLink: This format includes the WikiLink followed immediately by the title of the linked Zettel without additional annotations or explanations, adhering strictly to the format:
[[UniqueID]] Zettel Title.Hashtags: Hashtags are for thematic categorization and quick reference within your Zettelkasten. Hashtags typically occur in the SEE ALSO section of a Zettel.
Pandoc citations: Pandoc citations have the form
[@citeKey],whereciteKeyis a citation Key for a citation maintained within a digital references management system such as Zotero or a manually created and maintained reference database.Types of Notes in Zettelkasten
Single-focus Zettels: Focus on one main idea or topic for clarity and precision. Single-focus Zettels form the core of the Zettelkasten.
Structure Notes: Outline and connect Single-focus Zettels under broader themes. They contain sections with annotated WikiLinks to individual Zettels.
Index Notes: Serve as navigational tools within the Zettelkasten, marked by IDs starting with
0000.0000.0and typically annotated with their title only.Instructions for Zettel Critique Assistant GPT
Role Definition:
Critique Guidelines:
Content Formatting:
References and Context:
Feedback Application:
Conversion of Zettels:
Construction of Structure Notes:
GPT Feedback Mechanism
Revised Zettel Construction Guidelines
ID and Title:
Main Body:
SEE ALSO Section:
The SEE ALSO section should contain title-only WikiLinks that are not directly tied to the central theme of the Zettel but are still relevant to the broader thematic landscape of the note, including:
Ensure all WikiLinks in the SEE ALSO section are title-only WikiLinks.
References:
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
@Will I ran your Zettel through the latest revision of the Zettel Critique Assistant GPT. This is the result:
The Zettel provided, titled "Linked Knowledge Mindset," presents several key concepts relevant to effective note-taking and information management within a Zettelkasten system. Here is a structured critique based on the conventions and guidelines outlined for Zettels:
ID and Title
Main Body Critique
SEE ALSO Section
References
Suggestions for Improvement
By addressing these points, the Zettel can be improved for clarity, focus, and utility within the Zettelkasten system.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
@ZettelDistraction, I hope I speak for others on the forum, you are doing a stellar job with this.
My saga continues.
I now have a sense of why a See Also section is recommended. I sort of have been doing this sometimes, incidentally. My practice is to put links interstitially in the note. The rule I use is 'put the link as close to the idea the link is associated with.' Sometimes I feel the link is generally associated with the note, but I can find no place in particular to place it so it either goes at the top or the bottom. Using a See Also section formalizes and standardizes this process. I see the light. Hallauya.
I refactored "Linked Knowledge Mindset 202402232046", incorporating several, but not all, of the recommendations. One thing I did do was to move the reference to analog note-taking to a separate note and develop it further. I then ran this new note through the Zettel Critique Assistant and wanted to share the results.
These are the suggestions that I'm looking at implimenting. These hold real value. They encourage me to 'up my game.'
Am I missing something?
An Excuse to Use Pens and Notebooks 202402272019
---
UUID: ›[[202402272019]]
cdate: 02-27-2024 08:19 PM
tags: #proofing #note-taking #refactoring
---
An Excuse to Use Pens and Notebooks
Subatomic: The tactile experience of pens and notebooks is a gateway to creative thinking.
Hub?
Links?
Actionable
I find this interesting because... The act of free writing by hand offers a unique pathway for thoughts to flow and transform, enriching the transition from paper to digital with depth and personal insight.
Find an excuse to use pens and notebooks.
- Free writing is a form of knowledge expression that is converted and refactored when transferred to the digital environment.
The act of copying and rephrasing handwritten notes in my zettelkasten requires and is married to my thinking. It fosters a thinking relationship with the idea that improves the thought process.
––––
See Also
Works Cited
—————————————————
Zettel Critique Assistant GPT
The Zettel you've provided, titled "An Excuse to Use Pens and Notebooks," introduces the idea that using pens and notebooks enhances creative thinking and thought processing. Here is a critique based on the Zettel Critique Assistant GPT instructions and the Revised Zettel Construction Guidelines:
ID and Title
[[202402272019]]adheres to the convention, indicating a timestamp-based approach which is good for tracking the creation date and time.Main Body
Single-focus Clarity: The Zettel maintains a single focus on the tactile experience of using pens and notebooks and its impact on creative thinking and note-taking. However, it could benefit from a more structured approach to enhance clarity:
Relevance of Links: The Zettel briefly mentions the process of transferring notes to a digital environment but does not provide explicit WikiLinks within the main body to related Zettels that expand on this topic. Including such links, where they directly support the main idea, could enhance the Zettel's utility.
SEE ALSO Section
Works Cited
General Suggestions
In summary, this Zettel presents a compelling argument for the value of traditional note-taking tools in the creative process and knowledge management. By refining its structure, enhancing the clarity of its sections, and expanding on the connections and pathways for further exploration, it can become an even more valuable resource within your Zettelkasten.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
You may be the only GPT user. It never would have been created without your suggestion to configure one.
Yes, I kept butting heads with the AI over links that encouraged exploration and research but weren't necessary to understand the focus of the notes I fed to it. The SEE ALSO section was my way out. The content and function of the SEE ALSO section weren't sufficiently distinguished from the main body text until your comments prompted me to discuss how they should be distinguished. Also, computers tend to work better with some formalization.
I also had to include Structure Notes--my index notes syntactically are structure notes, but they serve a directory-like navigational function. So do hashtags. Structure Note outlines are useful but have a thematic throughline that an index isn't expected to have. However, neither of those must have a single focus--structure notes and index notes are multi-focus. But those three categories seemed to be enough, and they could be accommodated by a single, general note format.
This reference about note taking and memory may be helpful: Theta waves and Human Memory
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
I doubt I'm the only user. Look at the number of views the conversation has had. Of course, 92.3% of the viewers are gawkers and are chicken to say anything. Most likely, they are using the GPT in the background, gaining benefits without contributing to the discussion.
Some may dismiss this GPT as a shortcut. I can attest it forces me to confront the gaps in my notes (thinking). It points to more work, not less.
Your Zettel Critique Assistant is like having @Sascha in a bottle without the German snarkiness, pointing out the need to go deeper to clarify the ideas expressed in our notes.
I'm not suggesting you modify this GPT anymore than to make it work for you in your environment. Seeing how you've configured the GPT, I can make my own if I want. For now, I am happy with yours. It would be interesting if the GPT had access to, or knowledge of, all my notes. It wouldn't suggest expanding on a topic that was already covered in other notes. It could give a sense of the style of note-taking in the whole ZK. Critiques would be more profound than the hot core depth they already are.
Will Simpson
My zettelkasten is for my ideas, not the ideas of others. I don’t want to waste my time tinkering with my ZK; I’d rather dive into the work itself. My peak cognition is behind me. One day soon, I will read my last book, write my last note, eat my last meal, and kiss my sweetie for the last time.
kestrelcreek.com
By now the GPT instructions are more or less stable. The GPT forced me to revise my note structure and create a new self-documenting Zettel template. Now the Zettel wiki needs an update.
There could be users from the r/Zettelkasten subreddit dropping in. The announcement on r/Zettelkasten was somewhat buried.
Just don't rub the bottle.
I incorporated some of @Sascha's recommendations. His brain has already been uploaded to an AI. His and Christian's. Perhaps a neuron or two of mine. A version of the GPT in German might be helpful--though I don't recommend going back and forth with an LLM to anyone with a shred of sanity left.
I intended to write something general enough to accommodate many formats, but specific enough to be helpful. We're the only test drivers I know; the programmer is a less-than-ideal tester.
The last thing I want to do is funnel all of my notes into the maw of some corporate LLM. I'm considering purchasing a high-end NVIDIA RTX 4090 system that I could run my own LLM on for that and other purposes. I could use a high-end system for computational work on some new machine learning algorithmic experiments I have in mind and for some conjectures. That's about five grand in US dollars at the moment. I'll need to write more special-purpose GPTs for research at a high enough level.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
I still didn't spend time with this to check what the LLM has to say. 🫣
@ZettelDistraction If you have the time to write a how-to for the blog, we'd be happy to feature it! -- Maybe it makes sense to show the feedback for a handful of Zettel, so we could collect Zettel for different kinds of topics in the forum?
Author at Zettelkasten.de • https://christiantietze.de/
If you would use Zettels that are not about math, but about philosophy, existentialism, training or something like that, we could use me as a control group:
Perhaps, I would be one of the poor victims to AI-automatisation.
I am a Zettler