The meaning and significance of knowledge systematization may change.

u_g_cccc I think that literature (in a broad sense) will become shorter and shorter as it will be written on the assumption that it will be summarized anyway. If this is the case, the significance/meaning of knowledge and information organization will change fundamentally.

  • This is an interesting angle.
  • In fact, when writing a book for the general public with an editor, the editor often gives feedback on the author’s initial purchase, saying, “This is not good enough for the general reader to follow, so you need to add more.
  • It seems to me that the general reader is assuming that if they don’t understand a concept, they won’t search for it and look it up.
  • For example, in an early draft of “The Engineer’s Art of Intellectual Production,” I had written “Nietzsche” and changed it to “Philosopher Friedrich Wilhelm Nietzsche” because there was a possibility that some people might not understand it, and then we discussed the possibility that some people might not be able to read it and not understand it was Nietzsche. Then, when the discussion turned to “Some people might not be able to read this and not know it was Nietzsche,” he added “(Nietzsche)“.
  • A better example is Feynman. I personally thought “Feynman doesn’t need an explanation, if you don’t know, just Google it,” but after discussion with the editor [” Nobel Prize winner in physics Richard Phillips Feynman (Feynman)
  • This is an example of a simple substitution that is easy to introduce, but it is also an example of a longer sentence that needs more explanation because it is not clear enough.
  • image
  • Because so much effort is spent on these general-interest additions, blog posts without such general-interest restrictions require much less effort and are written in smaller volumes.
    • Many people find this to be “too abstract to understand.”
    • The situation of personal notes on a blog with technical notes is not clear to those who have not reached that area of knowledge, because I write new knowledge in a way that I can understand it just before I gain that knowledge.
    • On the other hand, for those who find themselves in the same situation and search, “That’s exactly the kind of text that puts your situation one way or the other!” on the other hand, for someone who is in the same situation and searches for it, it will be
  • Since Scrapbox, a form of knowledge representation in which abstract descriptions and concrete stories about them are placed on different pages and connected by links has taken place.
    • In this way, a page with a high level of abstraction will be mentioned and reused in several concrete example pages, and will become more and more elaborate as it is revisited over and over again.
    • On each example page, instead of repeating what is written on the abstract page, link to it and say, “Follow the link if you want to know more.
    • In this way, authors can propagate their thoughts with less effort.
  • So far, this has already happened before the LLM appeared.
    • The “getting shorter by the minute” has already happened and will continue to happen.
    • In “paper books,” tapping a keyword doesn’t give a description, nor does it tell you where else that keyword appears.
    • In e-books, some of the content is represented by links, and some keywords can be found by searching for the location where they appear, or by features such as Kindle X-Ray.
    • The reason why it is not widely used is because there is no incentive on the part of the publishers for the work of adding the link.
    • If a paper book costs 2,000 yen, an e-book is more convenient to search and copy, so it costs 2,500 yen, and if it also has links maintained by the author, it is even more convenient, so it costs 3,000 yen. 
 is not the world we want to live in.
      • Because the format has not yet become a “new format,” publishers have an incentive to publish in a format that is sure to sell, and there is little incentive to try a new format.
      • I think this is chicken and egg, the reader can’t buy something for which no means of purchase exists.
      • Publishers believe that readers need the familiar paper book form to read
  • When the knowledge transfer package is not in the form of a “paper book” read directly by a human, but is read by an LLM and generates commentary in response to human requests, the level of “intended audience” changes
    • The end-user person who ultimately reads the commentary output by the LLM is not the “intended reader,” but the LLM that generates the commentary is the intended reader
    • Authors do not need to explain terms that LLM understands; LLM will generate explanations if the end user does not understand the term
    • image
      • The Intellectual Production of Engineers p.128
      • image
      • Lynda Gratton, author of “The Work Shift,” believed that when “knowledge-providing services” such as Wikipedia appeared, the value of knowing that knowledge would plummet because the knowledge provided by such services was readily available to everyone. He also believed that if you do not have knowledge that such services do not provide, you will not be able to survive in the market.

      • I believe that the appearance of informational services raises the level of knowledge expected of readers, as authors think, “This is something that can be easily found by searching the Internet without having to devote paper space to explain it.

      • As of this writing in 2017, there is a significant difference in the quality of Wikipedia articles between the Japanese and English versions. If this influence raises the assumed knowledge level of books written by English-speaking authors, the cost to Japanese people of reading and understanding those translated books will be higher. I don’t know whether the Japanese will be forced to learn the basics from the English Wikipedia, or whether machine translation will make up the difference.

      • A chat service was created with “software that allows you to learn the basics from a huge amount of documents, including English Wikipedia, and still speak Japanese” (ChatGPT, 2022).

summary

  • The trigger is not LLM, but hypertext and the Internet.
    • Paper books have not changed much because they are hampered by the interdependence of readers who prefer familiar formats and publishers who believe that “readers prefer familiar formats.
  • The development of LLM will create an educational system that is different from the paper book, and the “authors” who create the data for that educational system will be outputting knowledge in a different form than the paper book format

This page is auto-translated from /nishio/çŸ„è­˜äœ“çł»ćŒ–ăźæ„çŸ©ăƒ»æ„ć‘łăŒć€‰ă‚ă‚‹ăźă§ăŻ 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.