Via AppleScript, tell Safari to fetch the HTML for every page of your Netflix ratings and output the collected ratings to a CSV file.

netflix_ratings_spider.rb

#!/usr/bin/env ruby
require 'iconv'

# This is a simple script to spider your Netflix paginated "What You've Rated" list.
# It requires an OS X based system with Ruby 1.9+, Safari, and AppleScript
#
# I could not find a way to back up my ratings (for all titles, not just my rental activity)
# without registering for a Netflix API key or handing my Netflix credentials over to someone
# who had an API key, so I decided to take a brute force approach and just parse the HTML for
# every page of my ratings history on Netflix's site.
#
# INSTRUCTIONS:
#  1) Launch Safari, visit Netflix.com, log in if necessary, and visit your "What You've Rated"
#     page. If the URL for page 1 differs from that of the STARTING_URL variable below, then 
#     update the variable's value. Leave the browser open on that page.
#  2) Set the OUTPUT_FILE variable below equal to the full path to the .csv file you'd like
#     for this script to write the ratings to.
#  3) Set the PAGE_LOAD_GRACE variable equal to the number of seconds that you would like to
#     give Safari to fully render each individual ratings history page before grabbing the
#     HTML source for the page.
#  4) Execute this script ($> ruby <scriptname>) and be careful not to interfere with Safari
#     while it visits each page in your ratings history.

# Config
STARTING_URL    = 'http://movies.netflix.com/MoviesYouveSeen'
PAGE_LOAD_GRACE = 5 # seconds of grace to allow for Safari to finish rendering a single page of ratings
OUTPUT_FILE     = '/tmp/netflix_ratings.csv'

# Character encoding converter instance used to force all HTML output into UTF-8 format
ICONV           = Iconv.new('UTF-8//IGNORE', 'UTF-8')

# For the given page's worth of Netflix ratings, glean out the title, Netflix URL,
# genre, and rating for each entry.
def glean_movie_info(html, ratings_array=[])
  next_url = nil
  html.gsub!("\n",'')
  html.scan(/<tr .*?class="agMovie".*?>(.*?)<\/tr>/).each do |movie_html|
    movie_html = movie_html.first
    if movie_html =~ /<td .*?class="cell-title".*?>.*?<a .*?href="(.*?)".*?>(.*?)<\/a>/
      netflix_url, title = $1, $2
      netflix_url = netflix_url.split('?').first
    end
    if movie_html =~ /<td .*?class="cell-starbar".*?>.*?You rated this movie: (\d+)/
      rating = $1
    else
      # skip "not interested" titles
      if title
        puts "Couldn't find rating for title '#{title}' (probably marked as \"Not Interested\"), skipping..."
      else
        puts "Couldn't get title or rating for this block:\n\n#{movie_html}\n\nskipping..."
      end
      next
    end
    if movie_html =~ /<td .*?class="cell-genre".*?>.*?<span.*?>(.*?)<\/span/
      genre = $1
    end
    ratings_array << {title: title, netflix_url: netflix_url, genre: genre, rating: rating}
  end
  if html =~ /<a .*?title="Go to the next page" .*?href="(.*?)"><span>next<\/span/
    next_url = $1
  end
  next_url
end

# Obtain the HTML source for the given URL
def fetch_html(url)
  applescript = <<-EOF
    tell application "Safari"
      activate
      set url of document 1 to "#{url}"
      delay #{PAGE_LOAD_GRACE}
      set htmlSource to source of document 1
      set the clipboard to htmlSource as text
    end tell
  EOF
  ICONV.iconv(`osascript -e '#{applescript}' && pbpaste` + ' ')[0..-2]
end

# Starting with the first page of ratings, keep gleaning ratings info
# and moving on to the next page until the last page (which will not
# have a "next" link at the bottom). Keep adding each page's worth of
# info to the ratings array, which contains a hash of info for each movie.
url_to_fetch = STARTING_URL
ratings = []
until url_to_fetch == nil
  url_to_fetch = glean_movie_info(fetch_html(url_to_fetch), ratings)
end

# Write the ratings to a .csv file
File.open(OUTPUT_FILE, 'w') do |f|
  f.puts "title,netflix_url,genre,rating"
  ratings.each do |rating|
    f.puts "\"#{rating[:title]}\",#{rating[:netflix_url]},#{rating[:genre]},#{rating[:rating]}"
  end
  f.puts
end

This of course no longer works, since netflix does not paginate but rather uses infinite scrolling.