The Data

We wanted the best of both worlds—the raw information about each tweet, and also to know which were the most-retweeted tweets around keywords. To do this meant combining two APIs: Twitter’s public rest API and the Search API provided by data reseller GNIP (the company actually became part of Twitter during the process of creating Twitter Reverb), which has a smart infrastructure around the way users can interrogate tweets, and covers a rolling 30 days of data.

Using both APIs served as a check on the data, says Earl Swigert, Periscopic’s technical director. This is how he explains it: “We use the Twitter service GNIP to generate a bulk dataset of historical information, and also determine the most popular tweets during the specific search criteria. We then pass this entire dataset through the Twitter API to ’rehydrate’ the Tweets and update their specific datapoints. The double filtering of the Tweets makes sure the information is as accurate as possible for each search."

The Design

We wanted something that would work on mobile and on television, both of which bring their own challenges. And the data in itself is not straightforward —the temptation being to show the world how big it all is by producing something over-complicated. Periscopic co-founder and design director Dino Citraro describes it like this: “Twitter data is deceptively complex. There are many data points that exist for a single tweet, and multiple layers of complexity when you look to extend relationships between followers, favorites, and retweets. The largest design challenge our team faced was determining how much information was too much.”

It wasn’t about creating a dataviz but about telling a story, says James Buckhouse at Twitter: ““Reverb is designed to scale from mobile to broadcast and to tell the story of an event. Our focus was squarely on the story that a journalist might want to tell, not just trying to figure out a new way to present data.”

So, the design changed a lot over a few weeks before we got to something we were happy with. These are some of the stages:

This was an early look at mobile integration screens, the focus being on distributed retweet network and impact:

That design evolved to provide context for individual retweets (shown as smaller circles). We explored various layout placements and text, too.

But in the end we wanted something that was easy to understand but also told the story via tweets, which is where this design came in:

Clicking on the circles at the bottom would take you through to the tweets themselves:

The spinning circle in the middle represents the numbers of retweets and the major influencers. This is what happens when that overlay is swept back:

We’re planning to develop Twitter Reverb more in future. At the moment, we’re limiting distribution to a handful of partners, based on the most interesting news stories of the time.

It’s not the be-all and end-all of visualizations, but that’s not the point: we wanted to create something to help highlight the way a story explodes on Twitter. Something non-coders could use to interrogate an API and create a visual—Users like me.