ems-score

name	ems-score
description	Use when scoring a signal from weirdness_ledger.md or source_map.csv using the 0-16 demand quality rubric, or when the user asks how strong a signal is.

EMS: Score Signal

Overview

Scores a named signal using the 0-16 rubric. Bases the score only on evidence already captured in source_map.csv and weirdness_ledger.md. Updates signal_scorecard.csv with the result and reasoning.

Inputs

• Signal ID or description (required): e.g., S001 or "developers maintaining memory files around coding agents"

Scoring Rubric

Score each criterion 0, 1, or 2:

• 0: absent
• 1: weak or occasional
• 2: repeated and strong

Criterion	What to Look For
Rough UX tolerated	Users stay despite bad product quality
User tutorials	Unprompted explainers, guides, scripts
Incumbent replacement	Old tool or process being abandoned
API/integration requests	Users asking for connections to other tools
Hacked use cases	DIY workflows, forks, templates, scripts
Vocabulary forming	New terms appearing before category names
Copycats appearing	Multiple tools claiming to solve the same thing
Outside curiosity growing	Non-users asking how the behavior works

Interpretation

Total	Meaning	Next Action
0-4	Probably noise	ignore
5-8	Watchlist	watch
9-12	Early market candidate	interview users
13-16	Serious founder attention	interview / test concierge offer

Process

1. Read the signal's entries in weirdness_ledger.md and source_map.csv
2. Score each criterion based only on captured evidence — do not infer
3. Note the basis for each score
4. Add or update the row in signal_scorecard.csv
5. Flag any assumptions or evidence gaps in the notes field

Output Format

Signal: [ID and name]
Score: [X/16] — [interpretation]

Criterion scores:
- Rough UX tolerated: [0/1/2] — [reason]
- User tutorials: [0/1/2] — [reason]
- Incumbent replacement: [0/1/2] — [reason]
- API/integration requests: [0/1/2] — [reason]
- Hacked use cases: [0/1/2] — [reason]
- Vocabulary forming: [0/1/2] — [reason]
- Copycats appearing: [0/1/2] — [reason]
- Outside curiosity growing: [0/1/2] — [reason]

Assumptions or gaps: [list]
Next action: [ignore / watch / interview users / test concierge offer]

Common Mistakes

• Scoring based on potential, not evidence: only score what is captured
• Rounding up weak signals: 1 means occasional, not "kind of present"
• Skipping the notes field: weak assumptions must be documented explicitly
• Scoring a signal without reading the source files first