Few pieces of dietary advice have been reversed as often as the one about breakfast. The instability is informative, because it shows how a particular kind of evidence behaves when it is asked to answer a causal question.
The original claim came from observation
Early support for breakfast came from studies noting that people who ate it tended to have different health and weight outcomes from people who did not.
An association of that kind establishes that the two groups differ. It does not establish that the meal is what makes the difference.
The step from the first statement to the second is where most of the subsequent confusion originated.
Breakfast eaters differ in many other ways
Regular breakfast tends to accompany a settled schedule, which in turn accompanies sleep patterns, working hours, smoking status, income and physical activity.
Statistical adjustment can account for the factors that were measured, but it cannot account for the ones that were not, and household routine is difficult to capture in a questionnaire.
So an apparent effect of breakfast may be partly an effect of the kind of life in which breakfast happens routinely.
Controlled trials produced smaller effects
When people have been randomly assigned to eat or skip breakfast, the differences observed have generally been more modest than the observational picture suggested.
Randomisation breaks the link between the meal and the surrounding lifestyle, which is precisely why the results diverge.
Those trials are also short and involve people who volunteered to change their eating pattern, so they answer a narrower question than the headlines usually imply.
The mechanisms point in several directions at once
Eating in the morning interacts with circadian rhythm, insulin response, appetite later in the day and the timing of physical activity, and these do not all favour the same conclusion.
Someone who skips breakfast may compensate later, or may not, and whether they compensate depends on individual factors rather than on a general rule.
A question with that many interacting parts is unlikely to have a single answer that applies across a whole population, which is what advice tends to demand.
Why the reversals keep happening publicly
Each new study is reported against the previous headline rather than against the accumulated evidence, which makes normal scientific refinement look like contradiction.
The stable part of the picture is unremarkable and therefore rarely reported: what is eaten across a whole day matters more than which hour it is eaten in, for most people.
Anyone with diabetes, a medication schedule tied to meals or a diagnosed eating disorder is in a different situation and should follow clinical guidance rather than general advice.