so, i'm not absolutely clear as to whether or not the research work was actually done.  a couple of scenarios come to mind: 

1.  Given that BofA has been losing entire funding packages that have courier tracking from title companies to their processing facilities and the title companies scramble to reassemble documentation at the last second, I suspect that a significant percentage of these non-searched properties have been searched and lost by the bank.

2.  With poor internal quality control, searches being done in India by a guy punching in a 9-digit APN number and having the data auto-populate into a title report tends to deliver a product that relies solely on the accuracy of flawed title plant databases that go back to the 1970's or thereabouts, hence missing decades or more of public records.  With a good coding crew, a few of the added references on the database to older documents (cross-references from resolutions, amendments, modifications, agreements, subordinations, substitutions, etc...) might be included for a few years more data, and in really lucky counties the data goes back to the 1950's, along with a legal description mentioning the subdivision map recording.  On these half-assed projects, general exclusion provisions are inserted and are thought to provide sufficient coverage alongside the due diligence of "the computer said so".... interesting paradigm.  Can I now issue high cost reports that simply clip the online Grantor-Grantee database with a few disclaimers?  ROFL.  What a joke.

I think that BofA needs an independent auditor to review their internal procedures to ensure that they really know what is happening.