Gravatar has some not-so secure issues relating to privacy that allow reverse lookups which isn’t good and could potentially leak information people don’t necessarily want to release.
My favorite solution to this problem and a few related others (like updating my bio and where you can find me on social media) is the meta data route using something like Microformats. Since I provide an h-card on my website’s homepage, it should be relatively easy for any service to take my URL as my identity (rather than one of my thousands of email addresses) parse my page and find my name, photo, bio, etc. and display them.
Nearly every social silo on the planet wants all of these details, so why should I need to incessantly have to input them manually much less keep them up to date? And I’ve yet to see a social service in the wild that hasn’t asked for my URL, so it’s obviously pretty universal.
Jeremy Keith‘s Huffduffer is a great example of something that already uses this data nicely. It doesn’t pull in my photo (though I think at one time he did have a set up that would poll Flickr avatars?) or my bio, but the “Elsewhere” section of my Huffduffer account lists where you can find me on dozens of social media accounts as well as my own websites. Huffduffer can do this because I gave it my domain name and the service parses my page looking for the rel="me" tags on my homepage. It could easily pull in my other provided data.
Incidentally Kevin Marks has also proposed a distributed verification system (remember the problem that Twitter had of attempting this?) that uses the rel="me" idea.
I’ll note that my own website will parse yours to pull in the author name, URL, and avatar to display a reply context for this response on my website! So hooray for microformats! (Though I’ll note that I did modify them a tad for my own idiosyncrasies.) My site does this with David Shanske‘s excellent Post Kinds plugin uses Parse This, which parses for microformats, JSON-LD, and then, if nothing is found it falls back to Open Graph Protocol. He’s been extending it lately to cover a handful of the bigger snowflake services like YouTube, IMDb, etc. to cover some additional edge cases that don’t have good mark up. Incidentally Aaron Parecki has a version of something like this called X-ray, which he uses for various things including microsub readers, not to mention the variety of other parsers available.
I’m sure there may be other versions of this in the wild, but it would be cool to see more social services provide functionality like this.