The One-Bit Alarm
Some observations are common and add little. Others are rare enough, under a well-defined baseline, to change what a system should pay attention to. This paper examines when a small, bounded observation can carry a large amount of useful evidence.
The question
Under what conditions can a single observation justify a meaningful change in attention?
Rarity needs a reference
An event is surprising relative to an expectation. The baseline, observation process, and context determine whether an unusual event deserves attention or simply reflects an incomplete model.
Attention is a decision
A compact alarm can direct additional analysis without deciding the entire case. This makes the distinction between noticing something and taking consequential action central to the design.
Small signal, careful interpretation
The work considers specificity, repeatability, and alternative explanations. A strong alarm should have a clear meaning, a measurable cost of being wrong, and a defined next step.