Keywords

1 Introduction

During times of disasters, social media has become a dynamic center of activity. Across digital spaces, people work to connect each other with information about the event. Whether it is about missing persons, bombing locations, or memorials, people reach out to share and understand what they are experiencing across time and space. We are now awash in an array of social media systems, have increased access to technology, and are becoming a more digitally literate society.

Often, the experiences of trying to share, validate, and distribute information are disrupted by the systems we create. From locating the missing to validating content across sites such as reddit, YouTube, Twitter, Vine, and Facebook, people have an urgent need to connect and understand their situation and each other. In order to devise better solutions for these emergencies, we need to consider extreme use cases when designing user experiences within digital spaces.

This research builds upon findings from usability tests and ethnographic research from the past 10 years [1], including work examining the Indian Ocean earthquake and subsequent tsunamis [2], the London bombings [3], Hurricane Katrina [3], the Christchurch earthquake [4], and the Boston Marathon bombings [5]. In this paper specifically, we will compare the Mumbai attacks of 2008 to the Paris attacks of 2015, two terrorist events that share certain attributes across physical and digital spaces. Using a framework for understanding experience architectures based on actor-network theory and information studies, this paper examines these two events and presents research findings.

2 Brief Literature on Understanding Data, Information, and Knowledge in Relation to Actor Network Theory

Across social media, people are able to upload images, update statuses, tag content, check-in to locations, and use other features to connect with others. In doing so, they are able to declare associations, identify with different social groups, and assign status to themselves and their communities. They – people, places, things – are all actors within various networks that work together across these systems.

For this study of social media use during times of disaster, our team has used actor-network theory to understand the communication networks and theory from information science to examine how people locate data, verify information, and distribute knowledge across these networks. Coupled together, this framework creates a method for understanding how participatory networks and technologies can help people cooperate during times of disaster.

A Brief Look at Actor-Network Theory.

Actor-network theory (ANT) originated in the field of science and technologies studies as a way to talk about the interactions between and across people and technology. This theory began in the work of Latour [6], with later additions from Law [7], Callon [8], and Mol and Law [9]. ANT suggests that all participants, whether human or nonhuman, have equal agency to affect any given situation; my work here gives priority to human actors, while encouraging researchers and practitioners to look at how these moments are contextualized for people by space, time, and technology. Referred to as “actors,” these participants can be people or technologies. During an event, these actors come together to form temporary, creating assemblages of relations and forming a collective, referred to as an “actant” [6]. An actant is a network comprising any actors — mobile devices, tweets, pins, people, and so on — that have the ability to act and do act within the network. Examining these actors and their networks provides a broader understanding of the people, groups, governments, organizations, technologies, and places, — the nouns — to researchers and practitioners [1].

For the purposes of this paper, an “actor” refers to any active participant in the network. Actors may include people, organizations, events, and technologies (e.g. devices, websites, apps) – an array of nouns. Actors join together in a network to accomplish certain tasks. This network of interconnected participants is a device for information coordination and flow. The Internet exemplifies flexibility in a distributed network: it is without hierarchy or predetermined routes between nodes, and it has solidly symmetrical nodes [10]. Networks of this nature therefore have the flexibility to come together and disband later. The social web provides such networks with systems for the coordination and dissemination of information.

Data → Information → Knowledge.

Researchers and practitioners often hold strong opinions about what constitutes knowledge. In order to better understand knowledge creation, our team uses the concepts of data, information, and knowledge to see how people transform these ideas. These concepts are based in the work of industry practitioner Morville [11], systems analyst Kock [12], and internet studies scholar Weinberger [13]. In our application, we take a more pragmatic approach. Our work is to pinpoint how content moves through these three stages, while also understanding the context in which this activity is taking place.

As addressed in our earlier research work, these three stages are described as follows:

  1. 1.

    The initial form of content is data. Data can appear in networks as words, phrases, images, symbols, and so on. A simple example of data is a Twitter stream. Without any context, a Twitter stream is just data: links, text, usernames, and hashtags.

  2. 2.

    Information is the second phase of content. Information is validated data, and validation can come in several forms. For instance, participants can connect two pieces of data, such as an image and a name, to pinpoint a person’s identity. Participants validating data created richer, useful, and contextualized content.

  3. 3.

    Knowledge is the final stage of content. Knowledge is information that is shared within the network. It takes a form that allows for repurposing and distribution [1].

In considering these stages, it is important to see them as porous, rather than rigid. It is not always as simple as pinpointing each stage. In the case of international disasters, this inability to pinpoint these moments can be attributed to issues of translation and localization – in other words, understanding the cultures and contexts. In the examples below, we illustrate how participants push knowledge through these networks, moving through these three stages by leveraging their networks and deploying various social web tools.

3 Background on the Mumbai Attacks

From 26 to 29 November 2008, terrorist attacks in Mumbai, India resulted in 166 civilian deaths and at least 304 injuries [14]. At least 12 locations were affected by terrorism, including numerous transportation systems and tourism sites. Western journalists focused the majority of attention on the explosions and gunfire at the Taj Mahal Palace hotel and the Nariman House. Throughout this multiday event, many volunteers online and the mainstream media performed the challenging and daunting knowledge work of locating and validating information from eyewitnesses.

Countless social web participants worked to mobilize, locating data, verifying information, and distributing knowledge across the globe. From the use of Google Docs and Twitter to the numerous images uploaded to Flickr, Mumbai was a major tipping point for online participation in the wake of a disaster. This event began a new era for the social web, one where participation spanned multiple systems, where people were organized and knowledgeable in their own culture and systems of use.

Participants developed connections across systems in order to find and share information related to the attacks. At the same time, they encountered many obstacles and challenges as they attempted to exchange information about the attacks. They pushed information from blogs to spreadsheets, and tried to flush out misinformation on Twitter through hashtags. None of these systems were connected in ways that would have made their jobs of sharing information easier.

One major example of how this occurred is the use of Google tools. Specifically, Dina Mehta and others used Google Sheets to collect data, validate it as information, and share it as knowledge with their community. Gaining access to a faxed list of the injured and killed in hospital because of these attacks, they shared this data with their network through their blogs and on Twitter. They were able to recruit participants to hand type the content of the fax into a Google Sheet that was shared across the international network. People could then validate the content—who was killed, injured, or otherwise missing—and then repurpose this content as knowledge that they could share with the friends and relatives of those affected. In this way, the actors in this network—people, Google Sheets, Google Docs (now called Drive), Twitter, Blogs—were able to guide this content through these three stages of data, information, and knowledge through their participation in the network.

In describing this work and the group of participants who helped her, Mehta stated:

[T]he “we” I speak of is not an organization but a loosely joined community. We are bonded, and I truly believe that in the face of utter horror, wherever it might occur, we have a strong pillar in this emotional connection we feel as equal human beings and not in our narrow identities prescribed by nationality or religion or race or gender. This is an evolving revolution sparked by how people are using social tools on the Web [15].

Mumbai was a pivotal moment for online participation during times of disaster. Text-based systems such as Twitter and Google Docs paired with Flickr’s image uploading service to create a new era in the social web, one focused on linking and working across systems. Systems developers cannot ignore this new mode of use for the social web. In order to meet the needs of the new social knowledge workers, the industry and the field must think beyond their own single serving interfaces, systems, documents, and silos.

4 Background on the Paris Attacks

On the night of 13 November 2015, a series of terrorist attacks occurred across Paris, France. Resulting in 130 [16] deaths and hundreds of injuries, these planned attacks caused chaos throughout the city. From explosions outside the football stadium to gunfire at restaurants and bars, suicide bombers left the city in panic and caused major spikes in social media usage. The Bataclan concert hall was the site of shootings and explosions where many of the deaths and injuries occurred. Videos of these shootings were replayed across digital spaces, showing the band’s reaction as the shootings began and showing the injured escaping through alleyways.

Within digital spaces such as Twitter, reddit, Vine, YouTube, and Facebook, participants raced to locate, validate, and share knowledge. Traversing among these systems, participants had an easier time of sharing what they found, but much of this information sharing was lost in a deluge of messages imparting sadness, prayer, and support. Finding and validating knowledge during such chaos is a key user aim for digital participants, yet we have not developed systems to improve these user experiences in times of disaster.

There are many compelling examples of compelling social media experiences from this disaster, from trying to locate the missing on Twitter to looping Vines showing scenes of horror outside the Bataclan, to the many instances of using Instagram to share photos and videos of scenes around the city. Characteristic of this new genre, great confusion at the scene of the attacks results from people’s attempts to understand what (if anything) was happening, where it was happening, and who was affected.

Locating information across systems, participants began to piece together the event. From the original Vine that recorded explosions outside of the football stadium [17] that has more than 379,000,000 loops at this writing and was cross-posted to Twitter, to the eventual missing persons posts on Twitter [18], participants worked to understand what was happening. There were Vines and re-Vines of footage of both the start of the shootings as well as people escaping the Bataclan nightclub, where the majority of victims were killed. Finding the original Vine was difficult, as it was reposted relentlessly by other participants.

Across these systems, hashtag usage did not seem to normalize at first as is often the case in disasters [4], with #Paris #Parisattacks #Bataclan and others in use. During the early hours of the event, and among the chaos, the international participation across social media, the flood of reposted content, and the lack of stable hashtags, there was great difficulties in locating data and validating information.

5 Similar Issues Remain

In this section, we discuss the issues that remain over time. Social media participants have increased access to technology, along with increased literacy and skills for working across a multitude of apps and services. What this has led to is a lot of content – posts, tags, tweets, images, etc. Considering the growth of social media and the unrelenting constant that is terrorism and natural disasters, it is imperative that we create systems that allow data to flow, information to be validated, and knowledge to be shared.

Looking specifically at recent research from the Pew Research Center [19] in Fig. 1, it is obvious that social media is a growth area: Facebook reported 890 million active daily users in 2014, and Twitter reported 320 million monthly active users in 2015 [20, 21]. We need to consider these issues in the context of larger concerns for the future of digital technology and human computer interaction.

Fig. 1.
figure 1

Social networking site use as percentage among all American adults by age group, 2005–2015 (no data available for 2007) [19] (Color figure online)

Full size image

5.1 Problems Locating Relevant Data

Locating information may have been easier during the time of the Mumbai Attacks because the pool of users was smaller in 2008 than they were during the time of the Paris Attacks. In January 2009, there were an average of 2 million tweets per day, a number that grew to 200 million tweets per day by June 2011 [22]. The genre of how to respond to terrorist events has also evolved. We can expect to see a show of concern, in this instance using hashtags such as #prayforparis. There are also retweets, reposts, and a general recycling of content. These movements can clog the network, creating an echo chamber in which original content (data) cannot be located or verified (information) because the reposted content, and thus original, validated material cannot circulate the network as knowledge.

An example of such use of Twitter during the Paris Attacks can be shown in Fig. 2. With millions of tweets being sent in the days after the attacks, locating relevant data can prove extremely problematic. How can we find useful, original content?

Fig. 2.
figure 2

Top languages used in the 4 million tweets sent following the Paris attacks, using data from NYU Social Media and Political Participation (SMaPP) [23]

Full size image

5.2 Issues Validating Information

After locating this information, it can be extremely difficult to validate it. Pinpointing original content often comes down to trusting the actors involved in the network. There are many questions we may consider as we examine the content. Is this account established? Is this content in line with other posts made by this participant? Does the poster make any claims to the data itself – do they take ownership of it or explain the originality of the content?

In an age where there term “Photoshopping” has become a verb, trusting content can become even more arduous. Being able to validate data through multiple sources of information is important. During extreme events such as natural disasters and terrorist attacks, we can use data such as geolocation metadata from posts, but only if people have turned on this information. Considering how often people opt out of these geolocation features, we cannot count on this data to be accurate or available.

5.3 Roadblocks for Circulating Knowledge

Major roadblocks for circulating knowledge are the systems themselves. Due to growth in social media use, increases in reposting original content, and posts in the #prayfor style, much of the original content can be lost. Recirculating validated content as knowledge, rather than simply reposting, can be a challenge.

From the Mumbai Attacks and the Christ Church earthquake, we have evidence showing that hashtags can be used advantageously to share information [4]. However, the participants during these events were in a much smaller community of people than those participating during the Paris Attacks, which drew a large, international set of users on Twitter [23].

One answer is in the ways that Twitter’s new Moments feature operates, collecting and sharing posts that they deem relevant and useful. Their curated missing persons moment from the Paris Attacks is still posted as of this writing [24]. Such historical knowledge repository work is important to future generations of users of these systems and the histories, the stories, we leave behind.

And of course, as academic researchers—we need to be able to have archives for this knowledge. Many, many of the tweets, YouTube videos, Vines, and Instagram content are gone. They have been taken down by their content owners or by the systems themselves. How might we preserve this content for future generations, historians, and ourselves?

6 New Opportunities for Engagement and Knowledge Sharing

As we can tell from the issues discussed above, there are several opportunities available to us as researchers and practitioners. In this section, we discuss the ways in which we ought to improve the user experience of digital spaces and areas that need our research attention.

There has been a dramatic shift between “use” and “participation” in the popular use of social web tools, such as Twitter, Instagram, Facebook, Pinterest, Flickr, Snapchat, Vine, and Wikipedia. “Use” is scaffolded by robust user-centered design theories and methods, while “participation” is pushing forward new ideas, theories, and techniques for best practices in researching, building, and supporting sociotechnical systems. In the past, experiences confined users to a set of interfaces contained within systems; now, social web participants consider a wide array of communication options across multiple networks. Researchers must explore these spaces as participants, open to re-evaluating preconceived notions of “social”. We can learn how to build more flexible systems, methods, and policies through these activities.

6.1 Considerations for Practitioners

Above all else, it is important for practitioners to recognize that their content will be of the most benefit to all if they allow it to flow, unhindered, across other systems. Instagram’s seamless posting to both Facebook and Twitter is a useful example, just as the difficulty in sharing search strings in these same digital spaces points to issues.

Considering the use cases in this paper, it can be argued that looking at such extreme examples can help in the testing of systems. How well can they scale to hundreds, thousands, or even hundreds of thousands of users? Are there spaces for participation? How do we let content move across systems?

More importantly, this kind of research can help us make important arguments about use, access, culture, and context. This research asks that we look at the networks in which our participants experience our systems – that we consider participating with them, that we use our own technologies to do this work, understanding the affordances and the limitations. How might we make more responsive, useful social media spaces? We can work within them and use them alongside our participants, engaging with them and experiencing it with them.

6.2 Areas of Future Research

We cannot ignore the possibilities and the participation of the social web. Each day, new systems are capturing our attention, our posts, and our pins. How these systems might be used during extreme cases of disaster point to the ways in which we can develop our networks and technologies.

While this longitudinal study has been an arduous undertaking given the severity of the topic and the reality of these situations, it has also been highly rewarding. Further areas of research should include eventual disasters, such that we can continue to understand these communication practices. Such contextualized data will help us create systems that make space for participation in times of stability as well. It is our responsibility to build systems that support human interaction and productive communication during times of disaster.