Emerging trends in social multimedia.

ABSTRACTThe past two decades of research in the multimedia community might be broadly described as an investigation into methods for automatically organizing collections of media or otherwise making them searchable and accessible. The un- derlying assumption is that people have the need to archive and access their growing collections of media, but that they are unmotivated to invest personal time and effort into manually organizing these collections; therefore, they require fully-automated or machine-assisted methods for managing these collections. A number of recent trends have emerged that seem to contradict these research goals. We will discuss some of them in greater depth in this presentation. There has been a growing interest in studying social media. This has been an \area" in ACM Multimedia in recent years, after being a workshop for several years. This research area has been somewhat reactive to the realities of the marketplace. Specifically, services like Flickr1 and YouTube2 made it possible for users to quickly share their photos and videos and provided easy tools for adding titles, descriptions, and lightweight labels (or \tags") to the pieces of media and organizing them into sets and lists to be shared with interested communities. What is interesting is that these actions were all driven by users manually labeling their media collections, not by automatic tools provided by the services. The social rewards of having media be viewed and appreciated motivated users to enrich their photos and videos with additional metadata. Media is not newly social, however. Before the proliferation of digital cameras, people still made slideshows of their vacations and pasted pictures into photo albums. The motivation was still largely to have friends and family over and to have a social experience around the media. The limiting factor; however, was that the audience was limited to those few people that could be physically co-present. The rise of online social media increased the potential audience, which changed the motivations for self-curation for many users. Similarly, common wisdom within the multimedia research community suggests that there is a pervasive desire among everyday users for systems to archive, organize, and retrieve images and videos at a later point in time. Increasingly, however, people are capturing photos and videos with the intent to only share and consume them in the extremely near future, without any intention of archiving or revisiting them in the future. On Instagram3, users can see a reverse-chronological feed of photographs from people that they follow, but have relatively few tools for curating and managing their back archive of photographs. The experience is about sharing in the present. More extreme services such as Snapchat4 or Facebook's Poke App actually erase the digital copies of the media after a short time, thereby forcing the media sharing and consumption experience to be extremely ephemeral. We propose that this behavior around ephemerally sharing media objects is motivated by the changes in the cost of acquisition and transmission of images. Where this mode of immediate destruction of images would be un-thinkable in the era of film and somewhat impractical in the era of slow internet connectivity, this can be a social gesture in the age of pervasive broadband and connected devices. We can also observe that much of media consumption prior to these services has also been ephemeral, historically. For example, the early era of television broadcasts were meant to be consumed live and offered little in terms of archival and later viewing. Much of the media content that we experience in a given day, in magazines, online, and on television is consumed in the moment and discarded. In this presentation, we will examine how these trends have shaped media consumption patterns in recent times as well as back historically to the beginnings of photography and video production as we know them now.