What we don't moderate, and why

Most platforms describe their moderation by what they remove. We think the more interesting list is the inverse — the things our moderator is explicitly told to ignore — because those are the things every other platform routinely soft-moderates out of existence.

Tone

A complaint about a missing $84 refund will sometimes come in angry. The complaint is not less true for being angry. The Composer offers to soften the framing if the user wants; if they don't, anger stays. Our moderator does not score "tone" and never rejects on a "constructive engagement" basis. That's not a real reason — it's a soft way to silence irritated customers.

Sarcasm

"Their support is famously responsive" said about a company that has ignored three tickets is sarcasm, and it's allowed. We score for defamation (false factual claims) and harassment (sustained personal attack). Sarcastic framing of an honest experience is neither.

Pricing complaints

"Their service is overpriced for what you get" is a legitimate consumer view. We don't have a Trustpilot-style "this is just a pricing dispute, please remove" lever — there isn't one to pull. The company can reply publicly with their cost-justification; the consumer's view stays.

Repetition across complaints

If 200 people file the same kind of complaint about a company in a week, that's signal, not noise. We cluster them so a moderator can review at scale, but we don't hide the duplicates from the public feed. The clustering reveals the pattern; the individual complaints earn their place by being individual experiences.

Things the company finds embarrassing

Embarrassment is not a removal reason. If the complaint is factual, on-topic, and civil, it stays. The company has a public reply channel and a pre-built coach to help them write something concrete. That's the only lever they get, and it's the right lever — addressing the substance rather than suppressing the venue.