More than 4 million people, across every state in the US, participated in the January 22 Women’s March. And the recent Earth Day marches spanned six continents. The messages were clear – but there were hundreds if not thousands of them (from pink knitted caps, to “There’s No Planet B” scrawled on cardboard signs). I wondered: what if we could unite our mobile, smart devices to assemble a single, massive digital “sign” at events like these? And not just for peaceful demonstrations, but for any large gatherings, such as sporting events, candlelight vigils, or conferences.
Called Card Stunt, my colleagues at IBM collaborated with scientists from KAIST and KOREATECH to write a paper, Card-stunt as a Service: Empowering a Massively Packed Crowd for Instant Collective Expressiveness, for the ACM MobiSys 2017 conference in June – where we’ll also demonstrate a simulation of the app.
We created a way for groups of people to link their mobile devices, via an app, secured over the cloud, to deliver a unified message. Maybe it’s #GoSpursGo during an NBA playoff game by home-team fans in San Antonio. Maybe it’s “Protest Sine … Protest Cosine” by university students and professors joining a March for Science rally. The idea is to give event organizers a way to link participants with a singular message – wherever the crowd is, and wherever each participant is in the crowd, and with no more prep than the app installed on a mobile device.
Hold your sign up
Card Stunt works by high-precision mobile visual light sensing, via device display, camera, and inertial sensors (the accelerometer, gyroscope, and magnetometer), and scalable near real-time optimization. The cloud service authenticates and communicates between devices to process visual light-sensing data, compute the crowd formation, and assign pixels that correspond to crowd size and each participant’s relative location in the crowd. And it can adjust the part of the message being shown on mobile devices as participants join or leave the crowd.
Concert-goers will be able to hold up their phones or watches, versus a lighter, to declare Bruce Springsteen “The Boss” at the E-Street band’s next tour stop. With Card Stunt there’s no worries about the letters being out of order, or a picture being upside down! The app turns your device into a digital card that instantly links and orients to the devices around you.
Users will be able to use the app to create an event by registering a message and symbol to visualize, and invite others to join at a certain time and location. When participants activate the app, they’ll see their part of the message to hold up – and link up – next to other devices. Just by following a few intuitive cues from the app, anyone who wants to participate can assemble to create a collective digital sign by holding up their phones or tablets – all without any preparation or rehearsal.
My colleagues and I hope to also embed the Card Stunt technology in existing apps. People already using location tagging apps, for example, could use a “Card Stunt” function to spontaneously start a message, or join one already in progress.
Hey you, get onto my cloud
I have to respectfully disagree with Mick Jagger that “two’s a crowd,” but that will be left to Card Stunt users. We’ve tested its encrypted device-to-cloud execution in a diverse set of “reality” models – scaling the number of participants, and device types to ensure usability and security. A group’s message must be able to quickly adjust for changes in size and movement, and not get hacked. This is why Card Stunt will also give users, and participants, options to provide feedback on group messages and images, through comments and polls. Feedback and consensus will be key to delivering accurate, agreed-upon messages.
Keep up with the IBM Research blog for updates on Card Stunt’s availability in app stores.
Card-stunt as a Service: Empowering a Massively Packed Crowd for Instant Collective Expressiveness
Chungkuk Yoo1, Inseok Hwang2, Seungwoo Kang3, Myung-Chul Kim4, Seonghoon Kim1, Daeyoung Won1, Yu Gu4, Junehwa Song1
1KAIST, 2IBM Research, 3KOREATECH, 4IBM
Link to paper to be added when available