I think there is a distinction between doing "title searches" and "abstracting."  The typical title search is guided by underwriting guidelines, and is solely for the benefit of title insurance producers.  Abstracting, strictly speaking, is about creating a permanent document that comprehensively represents all matters of public record affecting title (and those that don't, if they look like they might).  Here in Minnesota, an abstract is a certified document reflecting this.  A proper abstract does not delete an item just because there is little risk of it being a problem.  I agree with Robert that the practice of law encompasses abstracting, but our commerce department non-concurs.

Title searches, on the other hand, are much more limited by economic considerations.  Underwriter guidelines (which I have seen revised on the spot upon underwriting counsel being made aware of the time-scope of computerized indexes), are guided by cost and risk calculations.  For example, if the computerized records only go back to 1985, and the property has several transfers of the fee since then, then make sure there are no open mortgages, and no obvious deed errors.  Easier said than done, but IMO, the underwriters believe "artful" searchers know the ins-and-outs of doing this.  I say "artful" because I don't know of any hard/fast rule when a longer search needs to be done, and when a shorter one is fine.  Experienced searchers look at the situation and make a judgment call.  To some extent, this judgment call overlaps with title examination, but title exams are still relevant after the search person submits the findings.