When David and I were still working for MIT, one day he showed me an idea that he had while working on faceted browsing. He called it “link sliding” and when he explained it to me, in abstract terms, it didn’t make sense right away.

The main problem was browsing multi-typed data in a faceted browser. You could have Movies and Actors, say, but showing them on the same page and mixing the properties of the two types was confusing… but showing them on different pages prevented advanced interaction.

David’s idea was to organically combine the ability of faceted browsing to drill down on a set of given items, but then to use the faceted values as the new set of items, thus ’sliding’ the faceted browsing window onto the selected set and make that the new point of view. This would create a way to browse “sideways” from a particular set of items, following items of different type that are connected to the currently browsed ones.

I told you it didn’t make sense right away

Because he sensed that the beauty of his idea emerged only during interaction and was not evident on paper or on a whiteboard, David did implement a small working prototype to show his vision… but the problem was to find a database infrastructure and API that was fast enough and flexible enough to support his vision. We had a hard time finding that and the prototype was never released or published, with much sadness from both of us.

This was the state of ‘link sliding’ when he went to work at Metaweb: an idea that we both thought really powerful but could never get it to work on large enough datasets to make its usefulness come alive and its appeal obvious.

But last month David managed (yet again) to blow me away (and believe me, it’s getting harder and harder) by showing me an internal prototype of Parallax, his implementation of link sliding on top of Metaweb’s Freebase.

Thanks to the power, flexibility and speed of the Freebase API and its massive amount of multi-typed data, Parallax not only shows the wonderfully innovative UI paradigm that David invented, but it rivals, if you ask me, Freebase’s own user interface for usefulness.

But let’s use it for something useful: have you ever wondered if there is a relationship between the economical indicators of corporate campaign contributors to Barack Obama and their contributions? Try asking that question to Google, or Wikipedia!

Parallax and Freebase, working together, not only allow you to ask that question relatively easily, but also allow you to embed the resulting (web friendly! look ma, no flash!) plot directly into your blog!

So, first, let’s see if there is a simple relationship between contributions and revenue, as in corporations that make more money are more or less likely to contribute, or this is not an influencing factor. A scatterplot will help us with that:

Hmm, nothing catches the eye, but maybe there is one between contributions and capitalization?

Unfortunately, not a lot of data points (Freebase does not contain economical indicators for all the companies that contributed at this time), but we can wonder if the two clusters of companies that the above scatterplots seem to highlight are meaningful or they are just artifacts of the current visualization and lack of more data points.

Software that gives you questions, not only answers, is the best kind of software.

The above scatterplots were built from this parallax page, but I suggest you to start your parallax exploration from Parallax’s main page and discover the real potential of a web of data.

Oh, and make sure to watch the screencast, it will blow you away!

Gotta say, can’t wait to start my new job at Metaweb!