We went from a searchable database that displays the the index as if it were in a long alphabetical list, like a super-long paper scroll.

 We got a series of replacement databases that pull up based upon the search criterion, not allowing you to "scroll" up and down through the whole index.  This was a mistake and prevents good researchers from accounting for reasonable permutations in names.  

 Point of fact:   the number of mistypes where a Recorder would  enter SMITH as $mith or some other weird letter is impossible to account for in a database the pulls up info for you.  Additionally, the variety of names indexed with a single bland space before the first letter is too numerous to count.