Managing Content in the Age of Consumption

05 Dec, 2024

When setting up a note-taking system (or any system in general, like saving social media posts into different folders), we need a robust system with clear divisions to make sorting data effective. This need applies whether we're organizing notes, posts, or any other type of content.

Currently, these systems are not really optimal. Content gets buried by those above it, and we fall into the cycle of consuming more and more content without setting aside time to digest it.
I often find I never get it right the first time when creating folders such as "Writing, Reading, Cooking recipes", so it turns into an iterative process where I gradually create more folders as needed. However, this approach has several drawbacks:

• • I might forget about the existence of a folder and add content to something that vaguely fits (for example, I forget I have a maths folder and add some math notes to a "learning/study" folder)

• • The folder structure might be lacking and doesn't have as clear divisions as it should

• • I often get lazy and gravitate to creating a very broad folder and adding all sorts of things to it, which only makes things harder when trying to find specific information later on.
  
  Here are my thoughts on how this can be improved:
  
  The first state of managing information is having a system to save and store it. We can view social media bookmarks folders and an Obsidian vault as the same sort of concept. For this to be optimal, it needs clear divisions between each folder so I know where to come back to.
  
  This can be assisted with AI by having an in system model that can help you sort things into different folders, so that when you go to save something you get suggestions on where the best fit might be. This already takes place on the large social media platforms (the order of in which bookmark folders appear is determined by an algorithm.)
  
  The system also needs a good search function, which I believe consists of these key features:
  

• • Direct global search of keywords

• • Fuzzy Search (show similar results, this can be an instance where an LLM can help)

• • Ability to toggle between these two
  
  Recent improvements like x.com bookmarks search have made things easier by enabling search for any instance of a word (this feature exists in Obsidian too - global search for a word or tag). An LLM can help further with Fuzzy search - I can ask something along the lines of "What was that quote from ages ago that talks about shaving and being able to dismiss something without evidence?" and a sufficiently intelligent AI can understand I'm referencing Hitchens Razor: "What can be asserted without evidence can also be dismissed without evidence"
  
  In the case of social media, where the entire point is to discover new media, I think there could honestly be a benefit from going in the opposite direction. By this I mean if the algorithm occasionally shows you content that you bookmarked long ago. It's a pretty safe bet because it's content that the user has already demonstrated they like, and it would also encourage them to visit their bookmarks more.

However, there's an opposing perspective: users often bookmark content they want to return to later, usually things outside of the platform they're too lazy to address at the moment. So it's actually in the best interest of the social media platform to bombard the user with new content and hide the bookmarks as far away as possible (in the case of TikTok, for example). This also has the potential benefit of the user getting distracted by something on the home page and starting to scroll.