"Chad Thompson" <chad(at)weblinkservices(dot)com> writes:
> Heres an oddity. Why would it take more time to not find an answer than it
> would to find one?
Because the successful query stops as soon as it's exhausted the LIMIT
(ie, after it's found the first matching combination of rows). The
failing query has to run through the whole tables looking in vain for
a match. Note the difference in number of rows scanned in the lower
levels of your query.
regards, tom lane