The Coexistence of Sound and Bad Adaptations

From Pierre-P. Grasse's Evolution of Living Organisms:

Rabaud's study of cases like Tipula led him to confuse finality with utility Had he referred to paleontology he would have realized that the two concepts do not [always] coincide.  An end may be useful, useless, or indifferent.  The evolutionary trend may be good, bad, or  neutral for the succeeding line.  Evolution is indifferent to the survival or demise of the species.  Faulty evolution is littered with the bones of its victims.  The coexistence of a sound adaptation and a bad one, of useful, indifferent, or harmful characteristics, is one of the facts that demonstrate that the last word does not lie with selection in the world of living things.

Do Grasse's examples reveal something about the noise threshold of natural selection?   Or do they more often than not reveal that we are too ignorant to recognize what features selection is responding to?

If an apparently better adapted snail is outpopulated by another snail in its own environment, are we too ignorant to be making meaningful speculations about the selectionist history of a prehistoric taxon? or are we mostly in the dark about what each organism brings to the table (again, generally in the dark about the features at selection's disposal)?

Many adaptationists would say, "But of course, we must be temporarily ignorant about the adaptations upon which selection is acting!"  And most of those who would heroically defend science from "Darwin deniers" would nod their heads vigorously at this, as a reasonable (and useful!) defense against the ever-present dangers of provincial antiscience.  In other (i.e. non-political) forums, they might just as well argue that Grasse is citing meaningful evidence for the prevalence of neutral mutations -- which has its own usefulness for protecting Darwinism.

Or they might argue there is really nothing significant about it either way since it could be evidence for either neutrality or for our ignorance or anything in between, underdetermined as it is.  That is, the meaning of these examples depends entirely on which pro-selection argument it seems to be good for in a given context.