Research methodology and organization

History / Edit / PDF / EPUB / BIB
Created: February 10, 2018 / Updated: January 25, 2020 / Status: in progress / 8 min read (~1456 words)
reusable writing reusable notes

How does one effectively and efficiently conduct research on a topic that interests them? The goal is to be able to build on past research successes to build more and more complex research and results.

  • How to organize my writings so that I don't write the same thing over and over again
  • If I write about the same thing again, I'd like to know what I wrote in the past about the same topic, to see if my thinking has changed
  • Is it possible to build some sort of book/knowledge base where bits are reused, like code?
  • Is custom software going to help or is this a process problem?
    • Most often custom software should come after a manual process has been established
  • Is a Q&A formatted knowledge base an appropriate research tool?
    • Instead of writing topic-oriented articles, one would write question/answer-oriented articles

  • I want to construct upon any existing work I might have already done and avoid starting from scratch
    • This would be the same idea as code reuse in programming... How can one accomplish the same idea when writing about his ideas?
      • In programming you can refer/call other bits of code, however since writing/ideas cannot be executed, it seems difficult to reproduce the same idea...
        • It might be possible to consider building some sort of language or way to connect articles to one another (like web pages on a website). The problem always remain of linking things together
        • One thing I often do is run through different scenarios in my head, something I could simulate with https://github.com/tomzx/decision-trees. The question at that point is whether it is possible to use something like that but at the same time be able to evaluate whether the generated scenarios are possible (detect inconsistencies)
          • This idea is also similar to tests, in the sense that one would write scenarios they expect to happen and they would check if the current set of stories will produce the given scenario
  • Minimize the time spent on research maintenance (looking for similar work)

  • How to efficiently work on many topics at once?
    • Try to make topics orthogonal
    • Detect as soon as possible when topics overlap
      • How to do that?
        • Use of tools/software
        • Frequent review of content so that my brain can process it and detect overlaps
    • Write down everything as nothing stays on your mind forever
  • How to prevent information/knowledge duplication?
    • Frequently review content at random in order to consolidate it
    • Build tools that will attempt to identify duplication as content is being created
  • How to make progress instead of turning around in circles?
    • Have a high-level overview of the topics that have been discussed, keeping similar topics close to one another
    • Get feedback from others
  • How to effectively organize my writings?
    • Organize by topics
    • Organize by date of writing
    • Define a research process
    • Create research article templates (sections that are addressed during every research)
  • How to determine when a point has been resolved?
    • Keep track of the state of sections (draft, review, finished)

  • Define major domains of interest (for me, it is AGI/AI/ML)
  • Create an article for any new "major" topic (major being defined as "This deserves to be explored on its own")

  • Prioritize which topics to explore

  • Write down my own ideas and thoughts on the topic
  • Ask yourself questions you'd like to know the answer to
  • Prioritize the parts of the topic to explore

  • Write down thoughts and information as it is gathered
  • Research existing literature on the topic
  • Link with existing research (related topics)
  • Merge within existing research (similar topics)
  • Add references to any material that was useful during the process of exploration
  • Identify problems and attempt to solve them
  • Obtain feedback from others
  • Iterate on feedback and new ideas

  • Review content

  • Separate the content into correlated groups
  • Structure/Order the content of the article
  • Extract content into a separate article if it can be meaningfully explored on its own

  • Use various plugins/extensions to offer required functionality

  • Create a document per topic
  • Link topics to one another
  • Have an history of changes

  • Time track each lines
  • Fade lines with age (as they get forgotten)
  • SRS system to schedule reviews
  • Easy to link different things (lines?)
  • Is it possible to have some form of traceability between things? (if it evolves over time)
    • What should happen if something is completely rewritten such that the initial meaning is gone?
  • Search for text duplication within the workspace to avoid writing about something you already wrote about

  • Taxonomize/Structurize notes (give them categories)