The Idea Gate

# The Idea Gate

Seventy-five out of a hundred is the bar. Below it, the idea doesn't get built — and the day you'd
have spent on it stays yours.

That's the whole point of the thing. Not a scoring toy: a gate that runs before a single word is
written, when rejecting an idea is still free.

The version of the story that made it exist: a topic got scripted, voiced, rendered, captioned and
written up, and it was already dead before the first word. Nothing in the pipeline was looking, so
nothing said so.

## The six weighted items

Thirty points of it are mechanical, seventy are judged — and every judged item needs one line of
evidence, in writing, or the score doesn't move.

1. Angle class — 30 points. Which proven shape is this, and what has that shape *actually*
pulled? This is the heaviest item because the class is the idea's genetics. A head-to-head, a
one-line prompt trick and a build-log diary do not have the same ceiling, and no amount of execution
closes that gap.

2. Universality — 20 points. Would someone who never opened a terminal feel this pain, in their
own words? This is the narrowness detector. A niche mechanic used as *proof* of a universal promise
scores well; a niche mechanic used as the *promise* does not.

3. Attention ceiling — 18 points. Is there a visible twist — a belief-flip you can *show* rather
than state — and a demonstrable on-screen proof: a number that animates, a before-and-after that
lands? "A feature, described" scores near the floor.

4. Compounding artifact — 14 points. Is the thing you're giving away named like a product, and
does it keep working after install? A file someone reads once and is done with has no compounding
sentence, and the funnel has nothing durable to hand over.

5. Freshness — 10 points. Is this roughly a re-take of something recent, yours or anyone's?
Platforms suppress near-duplicates, so the differentiating twist has to exist *before* the topic
locks, not be discovered in the edit.

6. Urgency source — 8 points. Is there a real expiring thing — a release, a price window, a
limited cohort? Fabricated scarcity scores zero and is banned outright, because inventing a deadline
burns the channel faster than a weak topic ever will.

An idea can clear 75.00 and still go out carrying a NARROW flag. When universality scores under
0.65, the gate ships the topic as a *stated risk* rather than rounding it into a confident number.

That refusal is the feature. A rubric that always produces a clean verdict is a rubric that has
learned to agree with you.

And the gate does fire. On this channel's own August slate it returned 61.94 — RETHINK, NARROW
on a topic that had scored 77.38 twelve days earlier on a forecast with no data behind it. That
video was not made. The slot it freed is the one this very video is filling.

• idea_score.py — the scorer. Standard library only: no ffmpeg, no network, no API key
• RUBRIC.md — the six items, their weights, and the justification for the 75.00 bar
• ideascore.template.json — the file the script drops on first run; fill six evidence lines

Install in about a minute: unzip it anywhere, then run python3 idea_score.py <your-idea-dir>. It
writes a template the first time and scores it the second.

One honest caveat, and the README says it too: the weights are calibrated against one channel's
corpus. You inherit the *shape* of the rubric for free; the numbers you should re-earn against your
own results.

Comment IDEA on the video and I'll send it over — though nothing is gated on that. The kit is at
sharpendaily.co/kit/idea either way, no email and no DM required.