A trial balloon is a policy idea handed to reporters on purpose, with nobody's name on it, so that the people who handed it over can watch you react before deciding whether to own it. It arrives looking like a leak. It reads like a scoop. Very often it is neither. It is market research, and you are the test market. * The phrase is a translation of the French ballon d'essai — a small unmanned balloon sent up to read the wind before a pilot risked the ascent. English picked up the figurative sense in 1826, according to Merriam-Webster, and the logic never changed: send something cheap up first, and if it gets shredded at altitude, you have not lost a pilot. * Here is the mechanism in plain terms. Someone inside a government, a company or a party wants to know how an unpopular idea will land. Announcing it costs them something — a named proposal is a commitment, and commitments have to be defended or withdrawn in public. So instead the idea goes to a reporter through an unofficial channel. The story runs sourced to "senior officials," "people familiar with the deliberations," or "sources close to the process," usually with a line saying no final decision has been made. If the reaction is fury, nobody has to retreat, because officially nobody advanced. If the reaction is a shrug, the idea reappears three weeks later as a plan. * That deniability is not a side effect of the anonymity. It is the entire product being purchased. * The clearest recent case study is Greenland. On August 15, 2019, The Wall Street Journal reported — sourced to people familiar with the deliberations — that President Trump had repeatedly raised the idea of the United States buying Greenland from Denmark, "in meetings, at dinners and in passing conversations," and had asked the White House counsel's office to look into it. Not one name on the record. The reaction phase ran for days: Danish politicians called it a terrible idea, a former prime minister asked whether it was a joke, and Prime Minister Mette Frederiksen called it "an absurd discussion" and said Greenland was not for sale. * Then came the part that told you the balloon had been real all along. On August 20, 2019, Trump postponed his state visit to Denmark, writing that Frederiksen had shown "no interest in discussing the purchase of Greenland," and the next day he called her comment "nasty" (NPR, ABC News and The Washington Post all covered the sequence on August 21, 2019). An anonymous "reportedly" story from five days earlier had just been confirmed by the reaction to the reaction. Note what got cancelled, though: the trip, not the idea. And it came back — by January 5, 2026, Euronews was reporting Greenland's prime minister saying "enough" to the latest Trump threat over the island. A balloon that pops is not always dead. Sometimes it is just refilled. * The tells are learnable, and once you see them you cannot unsee them. Watch for a story where every source is unnamed and every named official declines to comment. Watch for verbs of contemplation rather than verbs of action — considering, weighing, exploring, has not ruled out, is said to favor. Watch for the escape hatch sentence: no final decision has been made. Watch for the idea appearing in one outlet first, in a form flattering to whoever leaked it. And watch for the follow-up that never comes, which usually means the wind was bad. * This is not only a White House trick. A supermarket chain floats a price rise to a trade publication before it hits the shelf. A football club lets a "club source" tell a local paper what season tickets might cost next year. A council leaks a proposed route closure to the regional press. An employer briefs a business reporter that the office is "reviewing hybrid policy." In each case somebody is measuring how much anger a decision will cost them, before paying it. The nasty asymmetry is that they only run the test when they suspect you will hate the answer. * Which is why the way you read these stories matters more than it looks. A trial balloon is one of the very few moments when ordinary reaction has leverage, because the reaction is literally the data being collected. Silence is not neutral in that system. Silence is scored as consent, and it is scored quickly. * Reporters are not the villains of this, though they are not innocent bystanders either. The story is genuinely true — officials really are discussing the thing — and the alternative is watching a rival outlet publish it instead. That is the trade that makes the whole machine run: access in exchange for carrying an idea in the shape its owner chose. The better newsrooms push back by demanding the source explain why anonymity is needed, and printing that reason. When you see a line like "spoke on condition of anonymity because they were not authorized to discuss internal deliberations," you are at least being told the price of the story. * So four questions to run on any "officials are considering" piece. First: who benefits from this being unattributable — is it the public's safety, or the sender's plausible deniability? Second: is there an actual decision here, or only a consideration wearing the clothes of one? Third: does anyone denying it do so on the record, and did the denial contain the word "never" or just the word "currently"? Fourth: does the outlet tell you why the source was granted anonymity, or does it simply grant it? * Answer those and you have converted the story back into what it actually is — a question being asked of you, in public, by someone who does not want to be seen asking. Treat it that way. Do not file it as settled news, and do not file it as nothing. File it as a vote you are being invited to cast by reacting, on a ballot nobody told you was open.