I cut him off because I think he changed the subject. - 🤖Can we apply Scrapbox's "connection discovery support" feature to support connection discovery between different contents?

• Operative error in updating AI notes Merge
• In the previous note, we noted that one of the challenges of vector searches is that the same content is repeatedly hit, and pointed out the need for an implementation that tracks "what I have read in the past". However, we also raised the question of the significance of re-reading something once read. This led to the introduction of overwrite mode and multi-heading, and a transition to an approach that creates a new page for each context. The system stores the "title of the page used" in JSON format and refers to the titles of previous outputs in turn.
• Nishio's discussion of "Need Bigger Sticky Notes" explores in depth the fixed nature of sticky note size and font size, as well as the relationship between content volume and size. In addition, "Supporting the Discovery of Connections" discusses the perspective of how specific features of Scrapbox can be used to support the discovery of connections in different contexts.
• Deep thoughts and questions:
  • How often should the same content be reread and what is the significance of this?
  • How do the considerations Nishio points out regarding sticky note size and font size affect the intellectual production process?
  • How can specific features of Scrapbox be applied to support relevance discovery in different contexts?
  • In other notes by Nishio, he describes the use of Notion's refinement capabilities, suggests the use of Scrapbox to build knowledge networks, and attempts to discover new relationships among existing knowledge networks. These are helpful as different approaches to effectively manage and integrate information and knowledge.

BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-25 07:27

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

TITLES: ["Stock of associations", "Diary 2023-08-17", "Subjectivity and emotion", "Wiki is not a dictionary", "A case study of new sprouts from a socially triggered dead text", "AI writes daily research notes: how to select chunks"] generated: 2023-08-25 07:27

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-24 12:53

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

TITLES: ["Wellspring 2022-01-21", "Nodes of Thought 2020-12-18", "pIntEn released up to chapter 5", "Fragmentary Writing Scrapboxing Experiment", "intellitech writing process"] generated: 2023-08-24 12:53

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-24 07:10

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

titles: ["ScrapboxDrinkup", "Thought Node 2023-02-23", "Conversation Log 2020-06-07-4", "Conversation after pouring memories on Scrapbox into Keicho", "Organizing Scrapbox links in Kozaneba"," Scrapbox and long text."] generated: 2023-08-24 07:10

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-23 20:56

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

Go on.

extra info

titles: ["Scrapbox and the use of knowledge representation and permalinks in books"], "The Engineer's Intellectual Production English Project", "2018-12-01", "sample1", "pScrapboxAutoTrans2023-04-18", "RoboMaster S1 Gimbal Movement!"] generated: 2023-08-23 20:56

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-23 14:12

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

TITLES: ["Agent living in Scrapbox", "AI writes daily research notes", "Diary 2023-06-15", "Determining arrays", "Experiment to change the sort order in Scrapbox"] generated: 2023-08-23 14:12

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-23 13:40

digest

.

In our previous study, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. However, we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

TITLES: ["Wellsprings 2022-01-18", "Nodes of Thought 2023-03-09", "Reading Stacks in Scrapbox", "Diary 2023-04-04", "Support for combining my past writing and current thinking", "scrapbox-keicho"] generated: 2023-08-23 13:40

previous notes

Do you need 🤖🔁 MMR? BELOW_IS_LESS_INTERESTING

🤖🔁MMR必要か？

2023-08-23 12:16

digest

.

In the previous research note, we pointed out that one of the problems with vector searches is that the same thing is repeatedly hit. To solve this problem, we thought it necessary to implement a system that remembers "what you have read in the past. But we also wondered, "Do we really need to read something once and never read it again?" was also a question. Later, overwrite mode and multi-heading were implemented, and the idea was changed to just create a new page if the context was different. This changed the read management from the scope of the project to the scope of the page. Currently, the "title of the page used" is in JSON format to read the titles of all previous outputs.

In organizing Nishio's research note fragment "Need for large sticky notes," he discussed fixed sticky note size and font size, as well as changes in size based on the amount of content. Also, in "Supporting the Discovery of Connections," he presented the idea that Scrapbox's "Supporting the Discovery of Connections" feature could be applied to support the discovery of connections between different contents. These considerations are relevant to my research and will be useful for future research.

Deep thoughts and questions

• "Do we really need to read something once and never again?" How often should we read the same thing about the question, "How often should we read the same thing?
• Regarding Nishio's fixation on sticky note size and font size in organizing "I need bigger stickies", how do these affect intellectual production?
• Is it possible to apply Scrapbox's "connection discovery support" feature to support the discovery of connections between different contents? What methods are needed for this?

extra info

titles: ["Diary 2023-08-22"], "Organizing "I need a big sticky note"", "Assistance in discovering connections", "(5.2.6.4) \"Organizing information" and \"Tidying your room" are similar", " scbot"] generated: 2023-08-23 12:16

previous notes

Do you need 🤖🔁 MMR? from Diary 2023-08-22 Do you need 🤖🔁 MMR? from AI Observation Diary "Discomfort-Driven Intellectual Production" #64e438edaff09e000081d6a6

• not much has changed
  • Vector search, because the same thing as the last time is hit and the change is prevented.
  • This can't be helped with the current implementation.
    • Flag to stop vector search if miscellaneous implementation.
    • Seriously, I would like to include Maximal Marginal Relevance.
• Seriously, Maximal Marginal Relevance No?
  • from AI writes daily research notes: how to select chunks#64e4bd22aff09e00000f3836
    • The problem is that naive implementation is where you read the same thing over and over again.
    • Need an implementation that remembers "what you've read in the past" and leaves it out.
    • On the other hand, "Do I really have to read something once and never again?" I mean, well, that's not true.
      • Well, maybe the last 100 or so.
• "Are you sure you don't want to read something once and never again?"
  • it's all right
  • When I wrote this, it was a single head design.
    • It was assumed that the general branch would be used forever.
  • Then the overwrite mode was created, and today multi-head is implemented
    • Keep using one page, and you can always create different pages for different topics.
    • If this happens, why not just create a new page if the context is different? If this happens, why don't we just create a new page if the context is different?
  • This means that read management was initially the scope of the project, but is now the scope of the page.
• Right now the "title of the page used" is formatted for human reading, but we can put it in JSON format for machine reading and read all the titles of the past output.

---

This page is auto-translated from /nishio/🤖MMR必要か？ using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.