Tag a single comment on a talk page with the request-for-comment template, and within a day a bot delivers a link to it to a rotating sample of active editors — chosen for being around that week, not for having any opinion on the article already.
Most disagreements never reach this point. A line gets reverted, someone explains why on the talk page, somebody else proposes a middle version, and the back-and-forth a talk page is built for settles it. Requests for comment exist for the cases that don't resolve that way: two or three editors repeating the same two positions until the only thing left that could move the needle is a judgment from people who weren't in the room. A reliable-sources noticeboard pulls in editors who know sourcing specifically. An RfC pulls in whoever happens to be editing that week, which is the right tool once the stuck question is about the article itself rather than a specialist's call.
The format asks something uncomfortable of whoever files it: state the disagreement in a sentence or two, and leave your own position out of the sentence. Write the question as an argument for your side, and another editor usually rewrites it before anyone outside bothers to answer — which spends several of the thirty days re-litigating the framing instead of the substance.
The question put to the community should stay neutral and brief; arguing a position belongs in the discussion beneath it, not inside the question itself.
Thirty days is the default run. A bot removes the listing automatically at that mark whether or not anyone has reached a conclusion, though a request can close sooner once the pattern of comments is already obvious, or sit longer if no uninvolved editor has gotten to it yet. Closing takes judgment rather than a headcount: someone with no stake in the outcome reads the whole thread and writes a summary of where the arguments actually landed, weighing how well a position fits the relevant guideline against how many editors simply repeated an unsupported preference.
A consultant with a declared conflict of interest can ask for one of these the same way they can ask for any edit: openly, on the talk page. What the format does not survive is the consultant filing it themselves. An RfC's entire value comes from the fact that the people answering have no connection to the article — the moment the question itself reads like it was drafted by someone with a stake in the answer, that distance collapses, and editors who would otherwise weigh in start asking who wrote the question instead of answering it.
That is also why the format resists the thing most clients actually want from it: a shortcut to a predictable yes. Spending thirty days pulling in people who owe the subject nothing is exactly what makes the eventual answer worth having, and exactly the part nobody filing the request gets to control.
