Glossary: Noul
Noul is Jev’s yes/no primitive on TypeSafe System One. The answer field noul is P(yes) from 0 to 1. Near 0.5 means the model is not committing. There is no confidence field and no probabilities map on a Noul answer. Reading answers.foo.confidence on a Noul is undefined. The probability is the signal.
This glossary entry is for builders who already heard of Jev / System One and need a citable definition before they treat 0.5 as a “medium Score.” jev.pro is an independent unofficial field guide — not affiliated with TypeSafe AI. Confirm the primitive on docs.typesafe.ai. We do not sell or issue API keys.
How it shows up
Put type: "noul" on a key in questions for POST /v1/systemone. Optional criteria.true / criteria.false gloss the poles — keep them aligned with the instruction. A Noul whose true means “no” is a documented failure mode. All questions share the same state; question IDs are not used in inference.
Two complementary Nouls need not sum to 1. TypeSafe’s jaggedness note shows that P(refund) and 1 - P(not refund) need not match. Do not enforce arithmetic identities the model does not guarantee. TypeSafe publishes a self-consistency cookbook that routes uncertain nouls to review. Keep the raw noul visible.
Threshold noul or |noul − 0.5| in your code. TypeSafe examples of floors are illustrations, not defaults. Pin jev-1.13.0 once fitted; jev-latest can move. Calibration for Noul is about the noul itself under the RLCD vendor claim — group-level, not a per-row guarantee. Schema-safe ≠ factually correct.
Contrast
Noul vs Choice. Choice is relative among up to 255 unordered options and returns choice, probabilities, and confidence. Separate Nouls on each option ask whether each property is true in absolute terms — all can be low.
Noul vs Score. Score is an ordered rubric (2–10 levels) with a possibly fractional expectation plus confidence. 0.5 on a Noul is not a Score midpoint.
Noul vs confidence. Confidence exists only on Choice and Score. Do not invent a Noul confidence on the wire. You may define bands (auto / review / abstain) in code.
Noul vs a threshold. The noul is the measurement. The threshold τ is yours. Do not copy a Noul τ onto a Choice.
Common mistakes
- Treating 0.5 as “medium priority.”
- Expecting
confidenceor a two-key probabilities map. - Forcing complementary Nouls to sum to 1.
- Writing
true/falsecriteria that invert the instruction. - Asking Jev to apply “the 0.85 rule” instead of applying it in code.
See also
FAQ
Is 0.5 a medium score?
No. It means yes and no are equally likely. Do not treat it as a Score midpoint.
Self-consistency?
TypeSafe publishes a consistency cookbook that routes uncertain nouls to review. Keep the raw noul visible.
Can I add confidence myself?
You can define bands. Do not invent a field on the wire.
Many Nouls in one call?
Yes. Fan-out is the point. Docs say extra questions cost tokens and barely change latency — vendor description, not an SLA from this site.
Limits
This is a jev.pro working definition. Noul is a TypeSafe API question type. We do not invent search volume. Schema-safe answers are not automatically factually correct.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai. Never treat jev.pro as TypeSafe official documentation. We do not sell, issue, or proxy API keys.