The earliest single user implementation would be how people organized their bookmark folders into subfolders. After that, you could argue that RSS readers enabled an individual's IG. By subscribing to sites with niche topics, you were basically hitting nodes on the interest graph.

Things get interesting when you start to talk about the modern multi-user interest graph.

Delicious was probably the first web-based property to expose the interest graph via it's tags system.

Currently Reddit and Pinterest hold the interest graph crown with subreddits and boards respectively.

Quora tried, but they've ended up focusing too much on personalities/celebrities and away from the IG. I can follow BobbyUser on Quora, but I can't follow just BobbyUser's posts on Gardening.

StackOverflow adds hierarchy to the interest graph, with the notion of high level sites and lower level site specific tags. They also limit themselves to the QA space and not so much the discovery area. They are most certainly QA first and only use the IG as a mechanism of curation/organization.

You make several good points. Here are some more notes:

Demonstrating interests via tags co-evolved between Delicious and Flickr; WordPress and Tumblr are the main beneficiaries of their work. Tag clouds were a nice hack for visualization.

The five biggest Interest Graph players -- Google, Facebook, Amazon, Microsoft, and Apple -- do it so seamlessly it's easy to forget that they are all collecting Big Data based on our gestures.

Since the Big Five are all private Interest Graph collections, I get more excited by the companies that allow public self-expression and gestures: Twitter, Tumblr, WordPress, Pinterest, Reddit, Flickr, Delicious, Quora, and Stack Overflow.

No one company can own the complete Interest Graph but these public services suggest there could be an aggregator / indexer / search engine for the public Interest Graph that is analogous to what Google and Bing do for the public Web.

I feel like twitter stumbled into the interest graph by accident with hash tags. After that, between search and the awkward lists feature, is attempting more curation... It's still so loosey-goosey over there in twitter land though. More of an "implicit" IG.

In comparison, reddit is the most "explicit" IG on the web today. That's not to say they are the best possible implementation, but I feel like on the spectrum of implicit <-> explicit, you should be more explicit in general. I don't know that data mining tools are accurate enough to get explicit IG nodes out of an implicit graph.

That is an excellent point.

Twitter benefits from having the broadest spectrum of interests (compared with the rest of that group) but to this day has not capitalized on it.

By comparison, Reddit and StackOverflow have a much more narrow band of interests but have a very rich gesture set compared with Twitter.

Perhaps that tradeoff is inherent to the nature of developing a service.

Which makes Facebook very interesting even though it is mostly private because it is both broad in terms of interests, and explicit in terms of the graph.

See also: What is the Interest Graph?

You never mentioned Tumblr or WordPress in your initial discussion.

I wonder why. Their tags are very good.