# kaal:position:2026-09-21-011

**Affirmed position.** Calibration is a property of populations, not of single decisions. A model can be calibrated on average and miscalibrated on a subgroup, and it can drift. Someone has to keep checking it on cases it has never seen, from outside its own data loop, and must bear a cost for checking carelessly.

**Status.** affirmed  **Published.** 2026-09-21

**Holds when.**

- TypeSafe's post announces a product. It is not an empirical test of these claims.
- The position is Kaal's own argument in the essay After RLHF Comes a Stack, Not a Model.
- The position does not evaluate Jev's accuracy, which TypeSafe has not published a calibration figure for.
- The accountable-validation layer is proposed, not tested: no policy has yet been trained under it.

**Current debate.** TypeSafe, Introducing System One models and Jev (TypeSafe blog, 2026), and the promotional claim that Jev is what comes after RLHF: https://typesafe.ai/blog/introducing-system-one-models-and-jev

**Extends.** kaal:claim:5245185-032: https://wulfkaal.github.io/claims/5245185-032

**Scholarly basis.** Wulf A. Kaal, How can we Best Monitor AI Agents (2025). SSRN: https://ssrn.com/abstract=5245185

**Source PDF sha256.** `4d7adba83ec722480e97bde6528cbe9ce98c709e45cb18794f157a64b8fe7da2`

**Topics.** ai-and-agents, governance-design, accountability

**Provenance.** Affirmed in kaalvis:2026-09-21:rlhf-stack-essay-six-v1 at https://wulfkaal.com/2026/09/21/after-rlhf-comes-a-stack-not-a-model/.

**Record type.** This is a dated commentary position that extends a scholarly corpus claim. It is not a verbatim claim extracted from the paper.

**Canonical form.** This markdown file is the canonical hashed representation of the position.