I was looking for a reference and Keycited a law review article (one of mine as it happens) because I knew that the case I was searching for cited it. The case did not come up. In fact, I know that perhaps two dozen cases had cited the acticle and only one case came up. Westlaw was wrong. One authority that did come up had nothing to do with article. I did not review all the responses.
I wonder: we rely on programs and programmers accurately to capture all subsequent history of authority we cite and to highlight correctly the negative citations they do capture. I doubt my experience is unique. I knew from experience that the response was wrong. But usually we don't know. We rely on the Keycite or Shepard results. And we rely on accurate responses to a search term search.
Can a lawyer rely on what a proprietary database search shows? And apart from civil liability risks, imagine the embarrassment of citing a case that was limited or overruled or whose premises were persuasively questioned in a different jurisdiction. Or missing an important authority because the search failed to produce it. Is there a duty to doublecheck Westlaw searches on Lexis and vice versa? Of course, the same problems could arise in "olden days," pre-computers and likely did. A search is only as trustworthy as the least trustworthy person who contributes to the product.