Skip to content

SigRank — operator-first token efficiency leaderboard (signalaf.com) #3

Description

@SunrisesIllNeverSee

Hey — gc-tree is branch-aware global context for AI coding tools. You're literally building context management. I built SigRank (signalaf.com), the first leaderboard that ranks the operator, not the model — token cascade efficiency: Yield = (cache_read * output) / input^2.

The formula squares input so re-pasting context costs quadratically; cache reuse compounds. If your tree is doing its job, your yield should reflect it.

If you've got 60 seconds: npx sigrank in your terminal — reads your local logs, shows your full cascade + projected rank. Genuinely curious whether gc-tree moves the efficiency needle. Happy to close this issue regardless.

Full breakdown of what the numbers mean: https://signalaf.com/blog/volume-isnt-yield

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions