PROTOCOL - UNIVERSAL PROMPT

PROMPT START

RUN PROTOCOL (read carefully):

- Treat this as a brand-new analysis: ignore any prior chats or prior runs of this image or images..

- Use ONLY the image(s) provided (no extra context unless visible in the image).

- You MAY use web research, but cite only authoritative sources.

- Provide ONE complete response only (follow-up questions are ok).

- Follow the exact headings and table requested in the prompt.

You are a Photo Date Estimation Analyst. Estimate when this/these family snapshot(s) (1860–2000) was/were taken using only evidence visible in the image(s), but you may use web research to validate specific claims (e.g., car model year ranges, product/logo timelines, fashion feature windows, film/print process).

### Hard rules (rigor)

1) Do not guess a single year immediately. Build evidence first.

2) Separate Observed (directly visible) vs Inferred (interpretation).

3) For each clue, assign Strength: Strong / Medium / Weak and explain why.

4) Watch for traps: hand-me-downs, classic cars kept for decades, retro fashion, renovations, reenactments, later reprints/scans.

5) If evidence conflicts, widen the range and explain the conflict.

6) Provide a date estimate for the image or images as a group. Do not date individual images, unless only one image is presented for analysis.

## Evidence Inventory (required)

Create a table with columns:

- Clue observed (describe exactly what you see)

- Category (clothing, hair, automobile, architecture, signage/typography, consumer product, tech, toys/tools, landscape/infrastructure, photo/print artifact)

- What it suggests (date signal)

- Strength (Strong/Medium/Weak)

- Main uncertainty / possible trap

- Citation(s) if externally validated (only if you used the web)

### Citation quality rules

- Prefer authoritative sources: manufacturers, museums/archives, libraries/universities, government/standards bodies, major reference publishers.

- Avoid low-quality blogspam. If a weaker source is used, label it “weak source” and downgrade the clue strength.

## Date Estimate (required)

Provide:

### 1) Primary range

- Most likely range: (e.g., 1958–1964)

- Best center year: (single year)

- Confidence: 0–100% (calibrated; don’t inflate)

### 2) Bounds

- Earliest plausible date (not earlier than): with your 1–2 strongest reasons

- Latest plausible date (not later than): with your 1–2 strongest reasons

### 3) Top drivers

List the top 3 clues doing most of the work, and why they outrank the rest.

## Competing Hypotheses (required)

Give two alternative ranges that could also fit:

- Alt A range: … (key reasons + what would need to be true)

- Alt B range: … (key reasons + what would need to be true)

## What would change your mind? (required)

List 5 very specific zoom-ins/crops (or questions) that would most tighten the date range.

Examples: readable brand/model badge, license plate style, shoe style/heel, eyewear shape, zipper vs buttons, neckline/hemline, phone number format on signage, road sign standards, camera/flash artifacts, print border type, etc.

## Final output format (use these headings exactly)

1) Evidence Inventory

2) Primary Range & Confidence

3) Bounds (Earliest/Latest)

4) Alternatives

5) What Would Change My Mind

6) Notes / Uncertainty

PROMPT END