When we first post a blog post, Facebook is often (but not always) confused about what a page is. In particular, if you try to use the "Like" button on a blog page or if you try to share a link via a link in your profile, Facebook will see the root page of the blog, not the actual page of the message.
For example, we recently posted: http://thisorthat.com/blog/2010-song-of-the-year-round-1-results
If the user "loves", he returns the title for: " http://thisorthat.com/blog "
When you run the message url through FB Linter, you can see the problem. " http://developers.facebook.com/tools/linter?url=http://thisorthat.com/blog/2010-song-of-the-year-round-1-results "
In the Debug section, an additional og: title, og: url, description and og: image header is displayed (last 4 lines of the Debug section). This is the metadata from the root directory. We cannot understand why. Why does Facebook see the correct metadata on the page, and then also pulls the metadata from the root directory (and uses it incorrectly to populate the data for the Initiative button and the Sharing tool)?
Another oddity. FB "figure it out" in a few days. Of course, by then relatively few people were viewing the message.
UPDATE. I want to thank Peter Bailey again for his answer, but we also found that there was another problem that we had to solve before the Like button worked.
The problem was that we were displaying the Facebook Like button, which had not yet been published. The problem is that the FB then cannot resolve the url and “guesses” about the correct url. In our case, he always guesses thisorthat.com/blog. Unfortunately, it then caches this assumption for several days and this cache cannot be cleared by Linter. So the final solution was to fix the og type: as suggested by Peter, but also remove the “Like” button from the blog preview. It is very important that you do not show the FB page before publishing it or, if you want, change the URL.
Jon kelly
source share