sao.dev / writing
ALL POSTS
blog / zt-2025-06-28-theories-narratives-values

Theories, Narratives, Values

2025.06.28 ~2 MIN READ
  • Theories - provable from first principles + data
  • Values - societal assumptions, partially grounded, but more established than narratives?
  • Narratives - subjective and based on experience, values, and other narratives
  • Theories are supposed to be universal
  • Narratives don’t need to be, just need to be useful for a group
  • Most valuable knowledge is narrative
  • LLMs and narrative, can they make subjective judgments?
  • Insight and credibility matter for building valuable narrative
  • Narratives are conceptual stepping stones
  • Narratives as framing of complex situation, narratives have volition
  • Narrative is why we have “I didn’t have the time to write you a short letter, so I wrote a long one”
  • Breath and focus are finite, facts alone are a waste, narratives are evolutionary?
  • Narratives as the FFN in transformers?
  • LLMs also use narratives to constrain the space of responses? Link
  • narratives as proteins? Eg sparse matches to enable complex narrative reasoning
  • Is thinking waking the narrative tree? BFS?
  • Are we just simulating the activity of the brain socially, doing all the same problem solving strategies, but with vocal etc communication bridging the literal neurons?
  • Clarity in characterizing a situation with as few details as possible
  • Narrative as the gradient (in a company, in life, etc) narratives as possible
  • narratives as kernel functions for the reward/progress space
  • Connection between bet making and narrative building and “you can’t plan greatness”
  • If survival in evolution is not an objective, is “why people do things” just survivorship bias?
  • Narratives as necessary stepping stones for groups to be compelled by ideas, systems, etc
  • What does it mean to want to build things? Is it art in expression, and fortunately useful? Is it better this way than utilitarian? Somewhere in between?
  • Narratives increase the range at which we recognize good bets
  • Objectiveless exploration plus abstraction is why open source rocks
  • Optionality is opportunity, as is freedom, low burn rate, etc
← ALL WRITING sao.dev