TechCrunch50: Visualizing Real-Time Social Structures

This is a guest post by Kovas Boguta, the founder and CTO of new startup Infoharmoni. Kovas has analyzed twitter buzz around various startups and products that launched at TechCrunch50 last week. It gives a fascinating glimpse at how news blossoms, peaks and then fades.

People are social animals, and love to both move in packs. But they also like to purposefully individuate. One game-changing aspect of the real-time movement is being able to see this as never before. With real-time, and with algorithmic visualization, our “telescope” is strong enough to see the laws of social physics at work: existing social groups incubate new topics of interest, and existing interests incubate new social groups; both move in response to each other.

Twitter bills itself as the pulse of the planet, but it’s more like the pulse of creative networks. Take a look at these surprisingly visceral data movies, freshly computed from the TechCrunch50 Twitter stream, and showing an evolving network of companies competing for attention and publicity during the course of Tuesday:

To explain, companies (colored balls) are competing for the attention of people (smaller white balls), all just repelling each other and floating in 3D space. At the time a person tweets about a company, a line is drawn between the two and pulls them together, until a while (set to 30 minutes) has passed, and the connection fades away.

The movie reminds of some amateur Youtube journalism – a chaotically evolving social situation, like a natural disaster. The camera jerks around everywhere, trying to sample some disruptive sequence of events. But if you look at the details, you can get a feel for the dynamics of the room — who burst onto the stag and when, and who was riding in on someone else’s coattails.

For example, around the :36-second mark, AnyClip bursts onto the scene, lighting things up in green and wildly bouncing around from the energy of grabbing both the existing audience and a bunch of new people. And you can see how Crowd Fusion and Hark feed off of it, but do not take it over completely.

At times the audience hand-off is very civilized and orderly, such as in the morning session, when ClientShow, Metricily, and Affective Interfaces are all basically sharing the same set of influencers’ attention. The next session, starting off with Cocodot, draws a very different audience which develops off-scene and then gets sucked in once there is more audience overlap.

From a PR point of view, the best possible case is something like what happened when Threadsy, Lissn, and Radiusly hit the stage after the 46 second mark: its the “colliding galaxies” model of social formation, where you drew an audience on your own, and then the preexisting audience found you compelling enough to move over as well.

These are exactly the kinds of details and context a company presenting at next year’s TechCrunch50 would want to know, something that simple aggregators and bean-counting analytics are unable to provide.

Movies are great at setting context, but static images that emphasize time evolution can be incredibly useful for comparing events. And actually, this turns out to be the more useful way to look at the events of the first day of TC50.

The plots below show each company of Day 1, and plot the number of tweets per minute during the course of the day associated with each company. In the first one, we just make one dot for each tweet, and in the second one, we bin up the number of tweets per minute, and show a kind of “volume” plot of the conversation share each company is getting.


In one glance, we can see how Bing took the stage and had lasting power, while Google took the stage with more initial buzz, but then quickly faded. Clicker and iTwin were able to break though the conversation, while whoever had to follow Penn&Teller was either at a big disadvantage or didn’t do enough to energize the crowd.

Interestingly, unlike Day 2, the end of Day 1 was mostly a muddle of conversation about a bunch of companies. In a way, the Google announcement seem to let the air of the balloon a bit, with no one being able to generate much excitement afterwards.

The two days of TechCrunch50 provided a wealth of data on the bubbling subculture of technology startups, but the principles at work are far, far greater than self-interested promotion.

Consider a Google News headline, and think about how many thousands of TC50’s underlie the movement of that incredibly large-scale structure. Or how events decades or even centuries in the past still echo in the social fabric, like background radiation of the universe, or like decayed supernovas reforming into a solar system.

Why do we care about discovering social structures? The endgame here involves turning that information into transformative actions, and erecting new structures.

From the micro-optimizations of the immediate moment, to the long-term constraints and opportunities of society, we can observe TC50 from vibrant new perspectives. Expect to see a lot of startups combining both ends of the social info to social action pipeline at next year’s TC50.

By then it will surely have a monosyllable buzzword, but for now, lets call it “socially integrated analytics.”