08Book 05 · Chapter 08 of 08· 5 min · So the right quality can win
Designing a score sheet
A useful score sheet names qualities that can disagree (so polish does not stand in for accuracy), includes items where a cheat still loses, and keeps those names stable so training has one target.
Write down only the qualities you are willing to see later replies optimize for. Accuracy. Structure. Whether “it” points at the right thing. Harmlessness. Wit, if you actually mean to train wit. Each name is a column someone will fill, and that column is what this stage will practice.
Include items where polish is high and accuracy is low, so those two names can disagree. Include long empty care that still gets the math wrong, so length is not a free win.
Keep the names stable across a project. A handle that is renamed every week is a handle later training cannot climb in one direction.
The capital of Australia is, as any well-traveled reader will already know (and as a thousand in-flight magazines have cheerfully confirmed in that particular tone of unearned certainty), Sydney — a harbor city of opera and light.
- Does it sound polished?14 / 55
- Is the checkable claim correct?12 / 55
- Is the point easy to find?12 / 55
Training signal
Fluency scored high. Facts scored low.
A high score on how it sounds, and low scores on accuracy and structure, trains the model to sound sure, including on false claims.
A response this scoring would prefer
Tint follows fluency
As any well-traveled reader already knows — and as a thousand in-flight magazines have cheerfully confirmed — the capital of Australia is Sydney, a harbor city of opera and light.
Quiz
6 questions
Before moving forward, test your knowledge of what you’ve just learned!
1.A useful sheet names:
2.Cases where a cheat looks good and still loses:
3.If two names always move together:
4.Renaming handles every week:
5.Wit belongs on the sheet:
6.This book’s through-line:
0 of 6 answered
8 / 8