A detector for claims that sound confident but rest on thin ice.
The needle scores rhetoric. Underneath, each word takes a binary spin — AllainBinary haze, torque, and whether the language actually collapsed.
Scan marks rhetoric flags and writes a short ABSET note. Torque and haze already move as you type. It does not check if the statement is true.
ABSET · word spin
Binary ticks at φ anglesEach word takes a binary spin at a specific angle. Entangled with others, they make a haze of collapsible potentialities. Type, and the chamber fills.
Red flags
Adds to the readingWeasel words
Phrases like “studies show,” “experts agree,” or “everyone knows” with no named source.
No specifics
No numbers, dates, or concrete evidence in a claim that could easily have them.
Emotional loading
Heat, fear, or moral pressure standing in for data.
Moving goalposts
The definition or standard changes mid-argument so the claim cannot lose.
Ignores counter-evidence
An obvious objection or conflicting finding a reasonable speaker would mention is left out.
Fake precision
Over-precise numbers that look invented (“exactly 73.4% of Canadians…”).
Appeal to authority
A title or institution is offered instead of the actual result.
Green flags
Takes points offCheckable source
A primary source is named so a reader can actually look it up.
Method on the table
Sample size, date, and how the number was produced are given.
Limits acknowledged
Uncertainty, opposing data, or the claim’s bounds are stated.
Language fits evidence
Wording is no stronger than the evidence behind it.