Blog Post

Prismatic wants to build a social network around all of your interests, which is actually a lot harder than it sounds

There are social networks all around us these days, from Facebook (s fb) and Twitter (s twtr) to more targeted networks like Instagram and Path — and new ones seem to be launching every week. Do we really need another one? Bradford Cross, CEO of San Francisco-based startup Prismatic, thinks that we do, and he is doing his best to create one using the news recommendation algorithms that he and co-founder Aria Haghighi have been fine-tuning since 2011.

The next iteration of that network launches today, with a comprehensive redesign and a host of new features that are intended to stress the social side rather than the pure news-consumption side. While the older version of Prismatic allowed users to “follow” or connect with users whose interests they shared, those features are much more prominent now — and so are other aspects, such as the ability to comment and tag other users.

A network based on your real interests

When it comes to comments, Prismatic has taken an interesting tack, similar to the one that Tumblr took when it wanted to add comments: instead of just allowing people to post a random comment on someone else’s post, they have to “reblog” the original and attach their comment to that, and it appears in their own stream. Doing that, Cross says, makes it less likely that trolls will emerge, and also encourages more authentic interaction.

Prismatic activity

The new Prismatic is designed to make it much easier for users to build a network around the content that interests them, turning the “customized newspaper” model that the service started with into the core of an interest graph. While plenty of others have their sights set on the same goal, Cross believes that Prismatic is better equipped to actually do it.

The Prismatic CEO’s take on social networks like Facebook, Twitter, LinkedIn (s lnkd) and Google+ (s goog) will resonate with anyone who has become frustrated with the way that those services implement — or reflect — our social graph. In a nutshell, he believes that neither Facebook nor Twitter (nor anyone else, for that matter) does a particularly good job of reflecting your actual interests, or at least not the broad range of interests most people have.

Prismatic profile

Existing networks box you in

For most people at least, Facebook represents the part of their social graph that is made up of family and friends, the people they have ties to in the so-called “real world.” But they may have very little in common with those connections when it comes to interests such as movies, books, politics, hobbies, etc. Facebook has tried to make it easier to organize a graph around your real interests with lists and other tools, but none of them have been that effective.

Twitter, meanwhile, is also supposed to be an interest graph — because you can follow people based on what they tweet about and/or what they do. But as Cross points out, that winds up slotting most users into a fairly narrow range of interests: people follow you because you tweet about technology or politics or sports, and so you might feel a little odd deviating from that to tweet about the movie you watched or a great meal you had recently.

“Because you have to be ‘on brand,’ you wind up getting boxed in. The more you work on that, the more you can’t share stuff in other areas because it’s confusing for people. And now you’re stuck, because you have Twitter or Facebook and other than that you just have a few networks for specific things like photo sharing — there’s nowhere to share your other interests.”

Prismatic fullview

It’s about relevance, and that’s a data problem

Prismatic’s solution is to use the algorithms that Cross and Haghighi have been putting together based on people’s interests since they started what would become Prismatic with a few computers and some mattresses in a condo in SoMa. As a data scientist who used to run a hedge fund and then built a data service for a startup called Flightcaster, Cross has no background in media — but plenty in understanding large data sets and how to make use of them.

After connecting the service to either their RSS feeds or a source like Facebook or Twitter, users browse through content that they have subscribed to, but also get shown things Prismatic thinks they might be interested in. As they share and/or comment on those articles, the system learns more and more about what they like and can make better recommendations — and the connections between users with shared interests gives it more to go on.

“All this stuff is basically a relevance problem — the problem of routing content to the right people. And our hope is that with the new Prismatic, you can actually start to build a sense of community around your actual interests, because the existing models aren’t working.”

Ideally, Cross says, those interests will be everything from food or architecture to music and religion — all the broad categories that users might be reluctant to dive into or share on other networks out of a concern that they might be “polluting” those specific streams with non-relevant content. Whether Prismatic actually accomplishes that will determine whether it survives, or joins the growing pile of failed social-networking experiments.

Prismatic raised $15 million in a Series A round of financing earlier this year, from a group of investors that included Accel Partners and Russian investor Yuri Milner.

Post and thumbnail photos courtesy of Shutterstock / noporn

10 Responses to “Prismatic wants to build a social network around all of your interests, which is actually a lot harder than it sounds”

  1. The website does not really makes me think “social network”, it resembles more an article currator, and there are many of them available. Even wordpress and tumbl.r does a great job of content agregation. I suppose it would depend on your personal definition of “social network” whether they would qualify as such or not.

    What does interest me is the way they handle comments, where you have to “reblog” to add your comment. And the interesting thing on this is how it will influence SEO and will Google be forced into changing their algorithms again. Will follow them just to see how this will pan out.

  2. Marc Rougier

    We at (I’m one of the co-founder) also believe there is a need, in addition to existing social media, to connect people via their interests. Bradford Cross is right. And you’re totally right too, Mathew, this is no easy undertaking!

    We’ve been building an interest graph since 2011 from a slightly different angle: while we believe in connecting people through their interests, we also believe that readers’ interest might be a bit loose (it’s somewhat difficult to define your expertise or passion if you have 162 interests). So, instead, we are primarily defining people interests based on what they publish (because publishing more demanding, more engaging than to read, and as a result, more defining to publish than to read). With all due respect to the concept of “readers interest”.

    Our publishing by curation platform helps professionals and businesses find (thanks to a recommendation engine), enrich (to make it personal and add value) and then publish content relevant to their interest. Interest graph is then inferred from who publishes what (and to a lesser extent, indeed, who reads / follows what). Users demonstrate interests and thought leadership via what they publish and then effectively connect with each other via the Interest graph.

    Sure enough Mathew, still a long way before Interest graphs organize the web as effectively as search engines and social graphs have been doing it; but it’s happening and I believe it’s valuable ( alone has attracted 800.000 publishers and 100.000.000 visitors so far – work in progress!). Praise to Prismatic to their take on it!

  3. One can say that they are trying to repackage (or evolve ) the forum.
    The problem is that it doesn’t really bring anything new to the table so it’s all about how it’s repackaged not about the information.

  4. Steve Ardire

    Avid fan and user of Prismatic and like bar being raised to fuse social graph with interest graph so look forward to this but to hit the trifecta you also need to incorporate knowledge graph with sophisticated machine learning and cognitive reasoning ;-)