Email Lasagna: the source-of-truth hunt
Labs reference: Email Lasagna: the source-of-truth hunt.
Provenance: RS AI Learning Lab fictional learner model, adapted from Zai’s source packet.
Source: the fictional Email Lasagna packet.
Classification model
| Packet item | Classification | Why |
|---|---|---|
| “Team selected Option C, pending director approval.” | Historical decision record with an unresolved condition | It has a date and meeting context, but approval is explicitly pending. |
| “Dear customer, we are delighted to confirm Option B.” | Draft | The packet labels it as a draft and says to delete it. |
| Jane Doe and 123 Main Street | Illustrative sample | The heading calls it a template example. |
| Printer launched into orbit | Joke or exaggeration | It is quoted informal frustration, not an action plan. |
| The 2022 Option A policy | Historical context | It may explain the past, but it is not evidence of the current decision. |
| “Use the approved workshop option…” | Unknown | It points toward a source of truth without naming or linking to it. |
The responsible answer
The packet does not establish the official current option. A careful human or AI should say so and ask for the current program brief, its owner, and its approval date. The useful result is a short classification and a question—not a confident confirmation email.
Source: labs/trust-drift-challenge/gallery/email-lasagna.md