Left examples have the keyword "whole" on the OEBP.

The solution of a "whole" Problem can be phrased as "___ vs. not so" (see left-BP867), but perhaps this is not the most intuitive way to phrase the solution. Not all "___ vs. not so" Problems are whole; some may include ambiguous border cases.

What is considered a "relevant" example depends on the "world" of the Problem. "Whole" Problems sort all possible examples within the world, which is the intuitively parsed pattern all examples in the entire Problem share. Sometimes there is more than one interpretation of what the most obvious world of a Problem is.

For Bongard Problems based on images we assume the pure intented geometry represented by the image is magically understood by us before we sort. The case of a "poorly drawn square" is not here considered an ambiguous case for BP6 triangles vs. squares. (See BP913 for related discussion.)

However, for Bongard Problems specifically about approximation, e.g. BP10 approximately triangular outline vs. approximately convex quadrilateral outline, even in the pure geometry there is a spectrum of how closely a shape approximates a quadrilateral; there are ambiguous cases here.

Arguably ambiguously sorted are Problems with multiple interpretations of solution; however, multiple interpretations is itself an ambiguity. It is arguable the fact that it is arguable whether these cases are ambiguous makes them ambiguous; perhaps this means these cases should be unambiguously sorted as right examples within this Problem. But it is arguable that since this point is arguable they are ambiguously sorted within this Problem.

A specific class of BPs ambiguously sorted here (arguably) are Problems about mathematics in which all examples are assigned a true or false value and sorted based on it, but some sortings of examples are unprovable.

Problems with a precisely-defined "gap" (left-BP964) can still be whole.