The indexing parser (and query parser) treat the '@' as a part of a term. This is more because of the parser's antecedents than something we want today. Making the '@' a term separator will allow the parsers to split, for example, the email address "john@smith.com" into the terms "john", "smith" and "com" rather than "john@smith" and "com". This makes searching for items by domain name, eg "smith.com", possible.
The ramification of doing this is that the whole index needs to be re-created.