Postgres sort alphanumeric only

Replace .order('title ASC') with this longer argument:

.order("
    CASE WHEN lower(title) BETWEEN 'a' AND 'zzzzz'
              OR title BETWEEN '0' AND '99999'       
         THEN lower(title)
         ELSE concat('zzzzz', lower(title))
    END")

This will sort case insensitive (lower); when values start with a digit or letter they are sorted normally, and all the other values will be sorted as if they were prefixed by 'zzzzz', forcing them to the end of the sort order.

Demo in SQL Fiddle

With Regular Expression

This solution combines the above idea with the idea of PJSCopeland (to use a regular expression). Again the strings starting with non-alphanumerical characters are sorted after those that start with alphanumerical characters:

.order("regexp_replace(lower(title), '([^[:alnum:] ])', 'zzz\1', 'gi')")

The \1 back-references the non-alphanumerical character that was matched, so all of them get prefixed with zzz.

Demo in SQL Fiddle

Disclaimer: I haven't tested this. It comes from the documentation for Postgres 9.1.

I have an inexact solution - the difference being that punctuation will simply be ignored, and your entries will end up like this:

.order("regexp_replace(title, '\W', '', 'gi')") # ASC is optional

=> ["Alley Oop podcast",
  "Always Listening: Podcast Reviews",
  "podCast411",
  "\"Success Living\" - Dr. Leigh-Davis",
  "talk2Cleo"
  "\"The Real Deal\" with Dr. Leigh-Davis",
  "Your Mom's House",
  "#WeThePeople_Live",
  "Zen Dude Fitness"]

• regexp_replace is the Postgres equivalent of Ruby [g]sub.
• \W means 'any non-word character' and will match anything other than A-Z, a-z, 0-9, and _.
  • If you want to ignore underscores as well, change \W to [\W_].
• '' is what you replace the matches with: an empty string.
• the g flag means 'all matches', and the i flag means 'case-insensitive' (dispensing with the need for lower()).