I begin with the ‘grounds clause’ and then expand on whether the two clauses satisfy a proper definition of knowledge in light of the justificationist program.
The ‘grounds clause’ is necessary as well, but not sufficient, for it is intuitively wrong to say that Susan knows that x is true solely by adopting the grounds clause. Although Susan has grounds that justify the claim that x is true, she does not justifiably believe this. She may believe that by guessing, or flipping a coin, but not through any epistemically responsible manner, or she may not even believe that x at all.
Since the reliability and grounds clauses are both necessary, S knows that x iff
S reliably came to believe that x on grounds that establish the truth of x.
Susan knows that x if Susan adopts x in an epistemically responsible manner and only when x is the case. Take the counterfactual: Susan knows that x because Susan adopts x in an epistemically responsible manner and Susan wouldn’t have adopted x in the nearest possible world where x wasn’t the case. For example, Susan is good at knowing when there are elephants near her, as she only believes there are elephants when there are elephants, and she rarely fails to notice one when there is an elephant. If there were no elephants, Susan wouldn’t have adopted the belief that there were elephants. Problem solved!
And yet, the nagging question remains: can it be said that anyone can satisfy both clauses and be cognizant that they have satisfied both clauses? If Susan is unaware that she satisfies both clauses, what sets her apart from the occasions when Susan comes to the truth through epistemic luck? While Susan may pick out the real elephant from a lineup of paper-mâché elephants, is Susan guessing or knowing? Justification, then, is superfluous. Susan may have knowledge in some cases, but it does her no good if she cannot demonstrate, at least in principle, that there is a proper grounding.
One option is to hold Susan to a stricter standard of epistemic responsibility. Susan is an ideal evidence-gatherer. This demand for complete relevant evidence will then remove the threat of Gettier situations, meaning that Susan will not come to the truth through epistemic luck. This way, Susan need not be aware that she satisfies both clauses, making the initial problem moot: she will necessarily have a proper grounding. However, we must give up most of our ordinary claims to knowledge, since this solution is equivalent of skepticism. We are not in any way ideal evidence-gatherers. In the process of safeguarding empirical knowledge claims against the threats posed by Gettier situations, we concede that we do not really know a great many of the things that we commonly think we do know.
Since the option of assuming Susan is an ideal gatherer of evidence contradicts the fact that we are not ideal gatherers of evidence, another option is to claim that there is a fourth clause: Susan forms belief that x with no Gettier-like problems facing Susan. This is the indefeasibility clause, which runs into its own problems: more and more qualifications were in the past ten years to rescue this fourth clause, and these qualifications themselves require that we assume something that clearly is not the case: denying Gettier-like problems amounts to denying significant problems that face us every day. Furthermore, even if assuming that the indefeasibility clause is a proper ad hoc patch to the swelling dike, how is Susan to know that any single case is not a Gettier-like problem?
“We can give an intuitive characterization of what is wrong when a man’s justification is defeasible by saying that he is less than ‘ideally situated’ with respect to the evidence bearing upon h. The notion of being ideally situated is, of course, a pipe dream. In any full sense of the term, anyone who falls short of omniscience is thereby less than ideally situated” (Marshall Swain, Epistemic Defeasibility, American Philosophical Quarterly, Vol. 11, No. 1, 1974, p. 19).