I got the following ideas after reading pertinent passages of "The Theory of the Photographic Process", third edition, Mees and James, editors.
In the pH range where ascorbic acid is not actively a developer, it behaves much like a sulfite. It has been shown (p. 367 of the reference) to be not significantly different from sulfite in affecting the developing activity of metol at pH 8.7 when the molar concentrations of the sulfite and ascorbic acid are the same. The mechanisms by which they work are different, the sulfite by forming the sulfonate of metol and the ascorbic acid by reducing oxidized metol to its original state. It would seem, then, that either sulfite or ascorbate ought to serve the same theoretical purpose in a superadditive mixture of, say, metol and hydroquinone.
Another property shared by sulfite and ascorbate is the abiliy to reduce the latency period of hydroquinone. It would seem that by most theories of the superadditivity phenomenon, either sulfite or ascorbate should be able to do the job. The required amount of either is quite small compared to the amount of sulfite commonly used. Thus, the combination of metol, ascorbate and hydroquinone should not be considered as a 3-agent developer when the minimum amount of ascorbate to achieve superadditivity is used, any more than the combination of metol, sulfite and hydroquinone is. There is usually little need to remove the sulfite, but when formulating a developer for staining, the less sullfite, the more intense the stain.
When we substitute ascorbate for hydroquinone in a PQ or MQ developer, we can leave out the sulfite and still have superadditivity because part of the ascorbate takes its place in restoring the phenidone or the metol. In this case, the developer does not stain the image.
Pp 289 and 304 of the reference will also be of interest.