A research report on the dynamics of crowd-sourcing music at the frog SXSW Interactive Opening Party, by Bonnie Reese, Mike Herdzina and Shaina Donovan—see Parts 1 & 2 for more information.

Crowdsourcing Gives Everyone a Voice

The overall response to the Crowd Sourced DJ was overwhelmingly positive. It was described as "novel" and people loved that it allowed them to participate in the music selection. When we probed further about what made it attractive, party-goers were enthusiastic about the philosophy of crowdsourcing, noting that it is a vehicle that gives everyone a voice. We were struck by the passionate tone and language of the participants. Some of the comments included:

- "It's for the community by the community," - "It's democratic," - "You have a chance to have a say"

However, while people embraced the spirit of crowdsourcing, many openly acknowledged that it compromised the quality of the output. We heard comments like:

- "The masses have bad taste" - "I don't trust the public"

And yet those interviewed did not perceive the conflict between crowd selections and individual taste as an inherent negative. One music-savvy partygoer intimated that while the music being played did not align with his preferences, it still "fell within his range of acceptability." In the context of a large social experience, like a party, the spirit of empowering the crowd reigns supreme. Many people acknowledged the party context and noted that there's a time and place for everything. So while crowdsourcing is okay in one moment and social environment, it may not be appropriate for every situation.

Crowdsourcing Makes Everyone Think Like a DJ

So what were the biggest influencers when individuals had to make a music selection? First of all, people took into account the audience for the music as well as the party context. "Party music" was referenced almost as its own genre (although based on the range of musical styles we heard, we doubt that everyone would agree what the key characteristics of "party music" are). While many used the word "upbeat" to define "party music", we didn't see further alignment in the music played. One partygoer commented "What am I in the mood for? Something that will create the right environment. What will make the right environment? Upbeat music."

Overall, the music that was selected could be described as more "upbeat" on the music scale. (Click to view full-size infographic)

People also mentioned that they overrode their own preferences to choose music for the crowd, noting music as a "shared experience." One party-goer said, "I avoided a few songs that I wanted to play because I wanted to choose for the crowd," while another person noted "I think it is more important that the crowd has fun." People mentioned avoiding songs they liked that might bring "down" the mood. They continually referenced music's role in creating the right social environment: "The music affects the vibe—it drives conversations and relationships." These comments acknowledge the importance of music's role in shaping the experience for individuals and the crowd as whole. They also point to a party environment as a scenario within which the greater good should come before personal preference.

It was also interesting that when asked to state their opinion about the quality of the music (on our scale from "it sucks" to "awesome"), a number of people said that "it's like voting for myself," showing an identification with the crowd. This leap from crowdsourcing to an identity alignment with the crowd was very interesting. Crowdsourcing went beyond simply being a method to create a shared playlist to creating a dynamic of communal identification. One person even said, "It's democratic, so people will like the music," voicing an assumption that music chosen by the crowd will automatically be embraced by the people within that crowd. It's worth noting that there were no overall trends in perceptions about quality of the music from the data we gathered at our Music Perceptions Wall; where people placed their dots on the scale truly ran the gamut (and some partygoers just wanted to be clever about where they put their dot regardless of what they thought of the music).

Music Choices Are Still Personal

Despite the identification with the crowd, people voted for the music they liked. This personal connection to music often led people to select their favorite artists and songs, even knowing they would never get played. Many people said that they made a selection knowing this. One woman noted that when she "came of age," her music preferences represented her spirit of rebellion, and she felt it important to honor this rebellious spirit by choosing songs no one else would choose. And while many partygoers took into account the party context when selecting music, no one that we spoke to chose songs they disliked. So all selections, despite being chosen for the crowd, still reflected an inherent musical preference on the part of individuals.

There's Joy in Discovering New Music at a Party

Many people placed great value on the discovery of new music and had a hope that would occur in a social setting. Some saw crowd-sourcing as an effective way to provide this introduction to new music, while others were skeptical that the crowd would actually select the unexpected. They perceived the crowd's "winning" music selections as inherently mainstream, and expressed a desire to expand their musical horizons as part of the social music experience. Regardless, the encounter with new music be it curated by a DJ, chosen by a friend, or randomly selected was seen as a valuable part of social music listening.

Visualized Data Tells a Story

Throughout the evening, we gathered data about each vote from the TouchTunes devices and from the mobile app. The resulting information answers questions such as: What songs were voted for? What time were the votes cast, and by which device? How many times over the course of the evening was a particular song voted for, and was it eventually played? This information as it exists in its raw form is interesting but offers little meaning on its own.

So, how do we shape the information into something digestible and meaningful? frog's SXSW music selection visualization combines numerous data sets into an interactive timeline. Recorded audio spectral analysis, time-lapse video, and voting activities captured in real-time at the event are rendered using D3.js, a javascript framework that builds on web standards like SVG, HTML 5 and CSS 3 to create engaging, interactive graphs. Fast-forwarding through the timeline allows you to feel the buzz of the party. Noise levels rise and fall, the clock indicates time moving forward, and voting activity is visible, measured in icons depicting whether the vote came from a TouchTunes device or from the mobile app. Some timespans are less active, while others suggest minor conspiracies to vote songs to the top. Votes that led to a song being played are celebrated with a crown. The information is visualized so that each data stream provides a different perspective on the evening, serving as a reminder that successfully articulated data tells a story that one person and one perspective alone cannot.

Our Conclusion

In looking back at the evening, we were surprised by the lack of a cohesive musical narrative. There were no discernible music trends or patterns in the music itself or the perceptions of the music over the course of the evening. No dancing climax or low point that everyone pointed to unilaterally. But that really shouldn't be a surprise given that narrative is carefully constructed to support a singular vision. The music narrative for this party was not about a singular vision, but instead a collective voice defining its own soundtrack. What we learned from our fellow partygoers was that the spirit of the crowdsourced vision for the music was more powerful than the music itself.

We have no illusions about the unique nature of frog's SXSW party and how this might have shaped our findings. It's a combination of pre-conference revelry and enthusiasm about cutting-edge technology. It's an environment where people are open to new ideas and want to participate. We were gratified by the crowd's enthusiasm about our research, and we enjoyed talking to a range of people about the music and their experience. We only wish that it were always so much fun to collect data.

frog at SXSW 2013: The Crowd As DJ:
» Part 1: The Concept
» Part 2: The Methodology
» Part 3: The Results