Probability in human communication belongs to the sender's beliefs about the receiver, not to the frequency of messages, the author argues

Shannon's information theory measures information by how unexpected a message is: the less probable, the more informative, using frequentist probability (observe repeatedly, count the proportion). This paper asks whether a numerical probability can really be assigned to a message a person deliberately creates once, and if not, which concept of probability fits human communication.

A rationalist theoretical analysis with no empirical data: a review of the literature on information theory, on frequentist versus Bayesian probability (probability as degree of belief), and on Misesian praxeology, assembled into a proposed model.

• Frequentist probability applies only to classes of repeatable events. A single utterance is a singular event and the set of possible utterances is unbounded, so relative frequencies cannot be computed in the first place.
• Human action differs from natural events. We do not know why a coin lands heads, so we count frequencies; with people we infer intentions by analogy with our own minds, and a person can change their mind at any moment, so any prediction from past frequency can always be falsified.
• Bayesian probability fits better, because it is defined as the degree of belief of the person estimating: uncertainty in communication comes from our ignorance of other minds, not from randomness inside them.
• Receivers already interpret through context. The chain email with letters scrambled inside words remains readable because words are short, the sounds survive, first and last letters stay put and, above all, context makes the words predictable; a scrambled word on its own becomes hard at once.
• The central proposal is to move probability to the sender. The sender acts on a prior belief about what the receiver knows and whether they share a code, gathers evidence from feedback, and updates to a new belief. Each exchange is a “test” of the sender's hypothesis about the receiver, an entrepreneurial action whose outcome is uncertain.
• The author calls for communication theory to shift from a receiver-centred model of uncertainty reduction to a sender-oriented model built on intentions and beliefs.

For teaching communication theory this is a concrete alternative reading of the Shannon-Weaver model that places message design where practitioners already work: guessing what the audience knows and adjusting from feedback. For campaign designers the usable lesson is to write the prior belief about the audience down explicitly and to treat each release as a test that updates it.