Our company (EDIT: not related in any form to readable or readability) does nonrss based crawling (blogs, news, message boards), so we spent quite some time on automatically identifying article, author and date and comments in articles.

If someone is interested in creating something similar to readability (e.g with us doing the article extraction for you) or does a need a website article extraction, you can contact me t.britz@trendiction.com.

PS. We have so many other ideas on our own and that's the reason we are not doing it ourselves.