I manage 18 microservices with heavy LLM usage and kept running into the same problem: when I talk to my teams about treating context as a scarce resource, there is no single term for what I mean. "Token efficiency" is a cost metric. I needed a design principle I could name, reference in code reviews, and enforce.
Core idea: a context is parsimonious when nothing in it can be removed without introducing ambiguity or degrading the result. The goal is not fewer tokens — it's conscious allocation of a finite budget between instructions and artifacts.
Does this distinction hold, or is it just a subset of token efficiency?
What existing formalisms would you apply here?
I manage 18 microservices with heavy LLM usage and kept running into the same problem: when I talk to my teams about treating context as a scarce resource, there is no single term for what I mean. "Token efficiency" is a cost metric. I needed a design principle I could name, reference in code reviews, and enforce.
I wrote up what I'm calling the Principle of Parsimony in Context Engineering: https://blog.rezvov.com/principle-of-parsimony-in-context-en...
Core idea: a context is parsimonious when nothing in it can be removed without introducing ambiguity or degrading the result. The goal is not fewer tokens — it's conscious allocation of a finite budget between instructions and artifacts.
Does this distinction hold, or is it just a subset of token efficiency? What existing formalisms would you apply here?