My boys will grow up to a world where work is radically different than what I’ve experienced in my life. The way they’ll coordinate their jobs with others will be much looser and “networky” than in our current organization-centric economy. One of the causes of this shift are changes in the way we find people who share our interests, so this article looks at the technology infrastructure for doing that.
Mapping Our Interests
Thanks to Facebook, most of us already know what a “social graph”is. It’s basically a map of the people you know. Build an Internet service around that graph and you get a social network like Facebook or Google+.
An “interest graph” is also a map, but instead of connecting us to people, it connects us to ideas. For example, among other things, I happen to be interested in business, networks, myth, and music by Michael Franti and Dead Can Dance. We can also look at some of my interests by mapping them on a simple interest graph:
Those lines illustrate my interest in these ideas – not my real-world connections to actual objects and people. The line between me and Thomas Jefferson, for example, shows my interest in what he represents, but (obviously) not the personal relationship you’d see in a social graph. Also, that connection to The Matrix says I’m interested in the movie, but nothing about whether I own a physical copy of it (we’ll leave that to the Internet of Things).
Software engineers are developing lots of different ways to build an interest graph. The most obvious is to simply ask people by allowing them to “like” and “plus” things online. There are also ways to infer our interests that are just as powerful, if a bit more tricky. For example, Google might interpret my search for “Thomas Jefferson” as a sign of potential interest in him; and that signal would get a lot stronger the more frequently I do it.
All this commenting, liking, searching and foraging for information on the web leaves our own unique little “pheromone trails” - while capturing of our interest graphs in the software of today’s Internet giants.
Mapping Our Meaning
When it comes to the interest graph, we really are talking about an interest in concepts – representations of people and things, rather than actual people and things. It’s this distinction that will connect our interests to the emerging semantic web, the next big evolution of the web, aimed at infusing meaning into text and other objects as a way to more easily automate connections between ideas.
There are lots of ways that these connections between ideas are formed, but one of the most promising centers on the way we humans interact with search technologies. When I search for “Benjamin Franklin,” “Monticello” and “Thomas Jefferson” around the same time, it suggests a meaningful connection between these terms, and the more people search for these terms together, the stronger that signal becomes.
Google tracks these semantic connections between ideas in what it calls the “knowledge graph“ - and surfaces them through the “people also search for” section of that box you sometimes see in your search results (see picture). The company is essentially using our searches to connect ideas and build its knowledge graph. That’s not the only way they’re doing it, but it’s a great strategy for them because it will be quite difficult for others to replicate.
By connecting ideas together in this way, Google will soon be able to help us discover interests we didn’t even know we had. I might not know I’m interested in Thomas Paine’s writings, simply because I don’t know about them, but Google knows I’m interested in Jefferson and that he’s semantically connected to Paine.
From a more commercial perspective, how about a search engine that knows that when I search for a nearby park, one of the activities people generally do in parks is have picnics? Given that kind of automated understanding, the search engine might show me ads of nearby delis and bakeries with ads tempting me to buy my picnic supplies. Revolutionary marketing? Not really. Any semi-creative marketer would have thought of that, except that, this wasn’t a marketer. It was a machine, and we’re now headed into a world where more and more of that creative insight will be automated by the semantic web.
Where this all gets really interesting though is when it gets connected to our interest graph.
Mapping Our Shared Interests
Some technology observers stretch the term “interest graph” to include other people who share our interests. I think this muddies the picture though. Just mapping people to their interests is a rich enough problem in its own right. We’ve barely scratched the surface here.
So, to get at this connection with others who share our interests, I like the term “shared interest graph” because it clearly states what it is: a map of people who share your interests.
At the most basic level, you use a shared interest graph in two ways: 1) finding new interests; and 2) finding new people.
When it comes to using the shared interest graph to find new interests, Facebook’s new Graph Search is a slick example. I’m still assessing how relevant my friends’ tastes in music, shows, and books really are to my own tastes. But it’s worth noting that one of my favorite musicians is Michael Franti, and he is at the top of my friends’ collective music list on Facebook.
As for using the interest graph to find new people, Facebook Groups and Google+ Communities are good, concrete examples. I run a “Good Business” community on Google+, which has helped me meet lots of people who share my interest in “business as a force for good in the world.”
With a shared interest graph, you can also combine finding new people and finding new interests. Music sites like Pandora show you lists of strangers who share your tastes in music. You can then visit their profiles to find new music that you also might like.
Opening Up the Shared Interest Graph
If none of this sounds particularly new or eye-opening to you, it’s because we’ve been working with aspects of the shared interest graph for a decade or more, even if most of us haven’t known quite what to call it.
Online retailers like Amazon were actually the pioneers in shared interest graph technologies. These companies used the shared interest graph as a kind of crowd-sourced personal shopper for suggesting products we might like based on what they knew of our tastes, and this gave them a powerful edge over traditional retailers.
These earlier pushes into the interest graph and the shared interest graph were all based on proprietary data standards. Amazon, for example, has very detailed data schemas for describing all kinds of products, which they have painstakingly built over the course of many years. I know a little bit about the difficulties here, having years ago run a product team at Microsoft that annually standardized vehicle specifications for all automobile makes and models available in the United States. It was a massive and messy job, and a source of considerable competitive advantage for our car buying service.
But this world of proprietary data schema is changing, thanks to the rise of the semantic web. Take a look at Schema.org, the collaboration between Google, Microsoft and Yahoo to standardize semantic descriptions of things like products,local businesses and many other things. These are fierce competitors, but they collaborate on this problem because they know that the semantic web and standardized schemas will greatly strengthen search operators at the expense of proprietary data masters like Amazon. It’s usually not a good idea to bet against Amazon, but if the semantic web builds sufficient momentum, it may well be forced to open up its product databases to future semantic search engines – or perhaps drastically shift strategies in order to become the semantic search engine itself.
What this means is that our interest graphs could become much more portable. Today, the music I’ve liked on Facebook can’t easily move with me to Google+ or Amazon. If the semantic web unfolds the way many believe it will, that could well change. It’s hard to imagine all this data commoditizing in ways that would enable that kind of portability – especially when you consider the power these players have today. But the history of industry shows that over time, businesses do commoditize – and the semantic web is likely to be a powerful force for that.
Remember too that the semantic web won’t just free our interest graphs from the proprietary data standards of online retailers, it will also help us build links between our interests and other ideas. Today, much of the way we use the shared interest graph is to connect to new interests – usually in the form of finding new products and services to buy. That’s understandable; shopping is where the money is, and software development efforts have flowed accordingly.
But this same infrastructure – the technology of the shared interest graph – has the potential to help us be much more than just better consumers. It has the potential to connect us with other people who share an interest and a stake in what we care about. And I’m betting that from these same seeds will grow something bigger, something that will affect the very nature of the way we work together.