Cartographies of Disaster
The Japanese earthquake changed our relationship to place, and post-disaster social media changed it again
On Friday, March 11, 2011, at 2:46pm, a 9.0-magnitude earthquake struck the east coast of Japan. Its underwater epicenter lay approximately 43 miles from Tohoku prefecture’s Oshika Peninsula.
The w’s in that last Tweet indicate laughter in online communication in Japanese, from 笑い (laughter), transliterated — and input — in roman script as warai. At this point, the magnitude of the earthquake wasn’t clear. Earthquakes are common in Japan. This earthquake, however, was the beginning of the 2011 triple disaster in Japan — earthquake, tsunami, and nuclear meltdown, collectively known as 3.11 — which would cause more than 15,000 deaths.
Natural disasters are fundamentally experiences of place: The epicenter was here. It was this many miles from this other place. It affected here and here and here. Place is understood through position and relationship, through contact and distance.
Geography determines terrestrial points of contact. These change, but usually at a rate barely perceptible to the human eye. Politics and language anchor societal points of contact, through alliance, ideological similarity, and shared knowledge. These change more quickly than continents, but stay stable long enough to fill history textbooks. Communication technologies scaffold personal points of contact. These change quickly indeed.
Imagine a Tokyo resident on the morning of March 11, 2011, using her iPhone as she waits for the subway on the way to work. She navigates overlapping, intertwining spaces: the station of a specific rail line, a queue of passengers waiting to board, the language space of multilingual signs, the political space of a Tokyo ward, of Japan; plus the literal underground — but also spaces of the internet, of Twitter interactions, Mixi diaries, and SMS exchanges; of iOS and the network infrastructure of SoftBank. She probably isn’t making voice calls on her mobile because that’s considered a little rude in the communal space of the subway platform, but she could.
That evening, when she tries to return home, her places have been dramatically rearranged. Her subway station is closed. So are all of the others in easy walking distance. Though Tokyo is far enough from the earthquake’s epicenter that structural damage is relatively minimal, the world’s largest metropolitan rail system is out of commission. Her home in Chiba, normally a forty minute ride away, is now far, far away. Long queues of exhausted people wait for taxis and she doesn’t know how to walk home. Worse, ever since the earthquake she’s been trying to call her sister, her brother, her best friend. She can’t get through. She tries again to call emergency services to check on her parents. She can’t get through. She can, however, still access the internet. As a device, her phone is functioning perfectly, though the battery is a little low now, and parts of the mobile network remain solid. She can still use Mixi. She can still tweet. Like many of her friends, she has chosen not to link her Twitter account with her personal identity in order to preserve a degree of anonymity. This practice makes finding loved ones outside of her normal Twitter circle difficult. Still, she is glued to Twitter, searching for more information. After more than two hours in the unmoving queue, a stranger a few blocks away tweets that he is driving toward Chiba and has a seat left in his car. She arranges to meet him and heads home.
The triple disaster reconfigured geography and politics in Japan. Social media use during the disaster reconfigured communication technologies.
Subsequent new features have, quietly and without declaration, redesigned users’ relationship to place — and not just for users in Japan, but everywhere.
This is a history of Twitter use during 3.11 and 3.11’s effects on Twitter, in Japan and eventually the rest of the world. It’s a history of changing choreographies of place.
Earthquake and Infrastructure
At Twitter’s headquarters in San Francisco it was still Thursday evening, almost 9 pm, when the initial influx of earthquake-related Tweets hit Twitter’s servers — and then, since this wasn’t a standard earthquake, the Tweets just kept coming.
Twitter has long been popular in Japan. As a result, Japanese was the second language added to Twitter, with the Japanese-language version of Twitter launching in 2008. For many years, the number of Twitter users in Japan and Tweets in Japanese were second only to the US and English. According to one recent study, Japanese is even now the second most common primary language of Twitter users.
But the task the Twitter engineers faced that Thursday evening was monumental: 177 million Tweets were sent on March 11, 2011, a 26 percent increase from the daily average at that time. This included a 500 percent increase in Tweets from Japan, with multiple spikes of 5000 Tweets per second. At that time, the record for Tweets per second was 6,939, celebrating the arrival of 2011 in Japan. (The record that stands today also originated in Japan; a rate of 143,199 Tweets per second was recorded in conjunction with the airing of Castle in the Sky there on August 3, 2013.) Twitter engineers worked through the night to rebalance server loads so that user requests would be reliably answered.
As an internet service, Twitter depends on infrastructure and engineers outside its own system. In this case Twitter was accessed heavily via mobile phones. Efforts to keep the various parts of the mobile infrastructure running spanned the US and Japan — successfully. Although voice channels were overloaded, data channels continued to function, and Tweets twined across the world.
This is an amazing visualization. Using @reply and Retweet data it surfaces adjacent — and distant — Twitter places on a map drawn in political and geographic dimensions.
And it wasn’t just Tweets. Emails, Facebook posts, and Tweets interwove. In one case, a group of kindergarten students trapped on a roof amidst a sea of fire were rescued after one of their teachers emailed her son in London. He tweeted about her situation, asking for help and Retweets. Someone in Tokyo saw his Tweet and retweeted it — at the same time @mentioning the then-Vice Governor of Tokyo, Naoki Inose. Inose saw the Tweet and contacted the Director for Disaster Defense at Tokyo’s Fire Department, located in an office nearby.
Since telegraphy, writing, and before, communication technologies have changed distance and points of contact for physical places. But social media are not just connected and connecting channels, they also open spaces for interaction.
Tsunami and Community
The first waves of the tsunami began to hit Japan’s eastern coastal region approximately a half hour after the earthquake. Early estimates warned people to expect waves of 10 to 20 feet. Later measurements and reports saw waves of up to 124 feet. In some locations waves remained as high as 26 feet even a half-mile inland. An area of approximately 217 square miles was flooded.
Images and videos of the tsunami’s devastating power and the destruction it left in its wake began to circulate. They would continue to circulate for days, shared on social media and dominating television coverage. Twitter served as an important hub for sharing links to visual media.
Today, the thick digital mantle of a natural disaster seems normal. In March of 2011 it was unprecedented. Twitter has been used to share emergency information as far back as the California wildfires of 2007, and was drawn on during the 2010 Haiti earthquake to coordinate relief efforts. But these were small in comparison with the digital recording, sharing, witnessing, and experiencing of 3.11, on Twitter and elsewhere.
Andrew Gordon, Director of the Reischauer Institute of Japanese Studies at Harvard, which designed and maintains the Digital Archive of Japan’s 2011 Disasters, describes the resultant “blogs, tweets, audio recordings, non-governmental and relief organization communications, photographs, videos, news articles, disaster-related government websites, and other digital documentation” as a “global barrage.”
In Japan, as disaster coverage continued to dominate television, concerns about the effects of this coverage on children led to the recording and sharing of special children’s videos as well. These were often announced on Twitter.
Whether you understand phenomena like natural disaster–related Tweets as artifacts or traces, analogs or derivatives, representations or manifestations, the events of physical places took over the spaces of communication technologies.
Locating critical cries for help in this barrage was a challenge. Roughly an hour after the earthquake, with the devastation of the tsunami still ongoing, a Twitter user started the #j_j_helpme hashtag.
The j’s here stand for Japan and jishin (earthquake). At this point, Twitter hashtags weren’t enabled for kanji or kana, the main writing systems of Japanese. Thus, for link functionality to work, hashtags had to be in romaji (roman script). This is perhaps not as much of a barrier as the reverse would be for the average North American. Remember the ‘w’s for laughter in that early Tweet? One of the main systems for typing in Japanese uses romaji as its input script. Still, it is a barrier. Hashtags weren’t in wide use.
The following day the Twitter Japan office published a blog post explaining how to use hashtags. The post highlights some already in circulation, like #j_j_helpme, and suggests additional ones. After listing 26 such hashtags, it offers a template for generating new hashtags organized by prefecture. Fourteen of the listed hashtags follow the model of #save_[prefecture name].
More than three years later, this list of hashtags seems optimistic in its specificity. It’s hard to imagine users adopting such a lengthy system in a short timeframe. And yet, in the week following the disaster, many of the hashtags listed were used tens of thousands of times. Some, like #jishin, #j_j_helpme, and #anpi (safety), as well as #save_miyagi and #save_ibaraki (two of the affected prefectures) were used hundreds of thousands of times. A number of hashtags of solidarity were also used for international support, including #helpjapan and #hope4japan, often in combination with links to donation sites, such as that run by the American Red Cross.
Within Twitter, hashtags are a building material of place. They lay the foundation for what is simultaneously a channel that can be accessed and added to, and a curation of information and conversation.
At roughly the same time that @ntm_u_kyo was tweeting about #j_j_helpme, others started tweeting about the explosion of the Cosmo Oil refinery in Ichihara, Chiba.
Yet another extended disaster seemed to be unfolding. Several hours later, Twitter users began to circulate warnings of toxic rain in conjunction with the explosion.
However, by the subsequent day, bloggers and officials were taking to different forms of social media to debunk these rumors.
Similar rumor patterns, with swells and debunkings, were later seen on Twitter during the London Riots and Hurricane Sandy.
Social media rumors are a contemporary form of folklore. From traditional tales like why rabbit eyes are red, to superstitions about setting your keitai’s wallpaper to a particular image for good luck, folklore has long been defined through multiple existence and variation. That is, to be considered folklore, something must exist in multiple locations and have distinct forms that are nonetheless recognizably the same. For many contemporary rumors, like these of toxic rain, multiple existence and variation occur in the assorted spaces of social media platforms.
So far, most studies of Twitter use in crises have focused on such large-scale dissemination patterns, whether of rumors like these or of confirmed facts. This focus tends to obscure smaller-scale, very personal interactions — interactions of intimacy, of connection and contact.
In a paper examining social media use and activism in Japan in conjunction with 3.11, social scientists David H. Slater, Keiko Nishimura, and Love Kindstrand suggest that an important use of Twitter during the crisis was to share emotional support.
The researchers highlight the use of textual versions of physical comfort, such as hugging (gyu gyu) and patting someone on the head (nade nade), suggesting that “For those who lived alone and spent that first anxious night tossing and turning in their beds — literally, given the force of the aftershocks — social media provided a community of similarly frightened people who cheered one another up.”
Platforms like Twitter were places for people to express and experience closeness. Geographically they may have been far apart, but on Twitter they were very much together, enacting virtual gestures of physical comfort.
Meltdown and Evacuation Zones
Such comfort was much needed. The disasters of March 11 didn’t end with the devastation wrought by the tsunami. The struggle with the reactors at the Fukushima power plant was worsening. At 9 pm, residents within a 3 km (1.9 mi) radius of Fukushima Daiichi were advised to evacuate. What was considered a safe distance would change dramatically over the next 24 hours.
The following morning, radioactive material was vented into the atmosphere in an attempt to avert a catastrophic meltdown. That afternoon the evacuation radius was extended to 10 km.
Later that day the evacuation zone would be extended to 20 km.
This is an obvious matter of place — of geographic, terrestrial place. Or is it? 10 km and 20 km are suspiciously round numbers. They are social, political numbers. They are numbers informed by the measurements and analysis available at the time (both of which are grounded in particular people and particular theories of modeling the world). And they are numbers informed by concerns about panic and safety, reputation and relationship.
Language also shapes place, and translation can be a critical part of emergency response. What information, on what channels, gets translated can determine lives. In some locations, like Tsukuba, during the days following the earthquake volunteers gathered to translate municipal Tweets from Japanese into English, Chinese, and Korean.
But it’s not as simple as language creating rigid or inevitable place boundaries. Or even distinct places. A study by Son Doan, Bao-Khanh Ho Vo, and Nigel Collier, researchers at Japan’s National Institute of Informatics, compared Japanese and English Twitter use in the Tokyo area during the disaster. They found that though Tweets about the earthquake occurred almost simultaneously in the two languages, concerns about radiation peaked earlier in Japanese than English. Language context shaped situation awareness.
Not everyone is in the same place on Twitter, and not all places on Twitter are next to each other. Twitter users who are geographically in the same location may be in very different Twitter places.
Lifeline and the Japanese Postal System
Disasters like 3.11 surface complexities of place: How different types of place connect, constrain, and overlap. How places of geography, politics, and technology mix, fuse, and transform. In turn, responses to disaster also have to manage place complexity.
In September 2012, a year and a half after 3.11, Twitter launched the Lifeline feature in Japan. Lifeline allows users to search Twitter by postal code for government, utility, and transportation accounts. The assumption is that such accounts will provide accurate, relevant information in crises.
Unlike Facebook’s earthquake feature or the early earthquake warning system built into iOS 5 and beyond — both Japan-specific features — Lifeline isn’t a form of geocasting. Geocasting is when a producer targets publishing to a particular region or location. With Lifeline, users choose the locations of focus, users direct information delivery. This is very much in line with Twitter’s history of backing its users, from the creation of hashtags to the challenging of secret subpoenas.
There are now more than 500 Lifeline accounts. Oddly, the feature can only be accessed via the web. This is a surprising omission given that during crises so many people use Twitter through mobile apps. Further, the feature is available only in Japan. Except that’s not really true. Lifeline appears as an option if your location is set to Japan. But you can set your location to Japan from anywhere.
What the feature actually does is use the Japanese postal code system to return Japanese accounts. You can explore the Lifeline feature for yourself by setting your country content to Japan via Twitter’s web interface, then clicking on the Discover tab. The Lifeline feature will appear as an option in the menu on the left-hand side. A video walkthrough in Japanese is here.
Think for a moment on the satisfying elegance of this: Twitter — Twitter — has tied a feature to the great grandparent of social media, the postal system. In using the postal system as a key point of reference, Twitter implicitly adopts the system’s model for arranging places. This arrangement, in turn, emerges from a history of political decisions from the Tokugawa period to the modern day.
I’ve mentioned Lifeline to a number of global activists and emergency management specialists based in the US. None of them had heard of it. Even just a few clicks can amount to a long distance between two places.
Meanwhile, Twitter was still working on how to increase its usefulness during emergencies. The next — and most recent — development laid out a new, transformative understanding of place.
Twitter Alerts and Device Places
Today in Japan — or rather, on the Twitter platform experienced when your account location is set to Japan — the Lifeline feature coexists with another technological response to natural disasters, Twitter Alerts. But the Twitter Alerts feature exists not just in Japan/through the Japan location setting, but across the Twitter system.
In September 2013, Twitter launched Twitter Alerts in the US, Japan, and Korea. That is, emergency services located in those nations could now have the feature enabled for their Twitter accounts. Since then, the feature has been enabled for organizations in Australia, Brazil, Ireland, Spain, and the UK, as well as for global agencies and nonprofits like Ushahidi, Water.org, and the World Health Organization. There are presently 270 organizations with access to the feature, and that number continues to grow.
As with Lifeline, Twitter Alerts uses institutional reputation as a marker for the likelihood of providing accurate information. However, unlike Lifeline accounts, which are organized around search, Alerts are organized around visibility. The Alerts feature helps you see the important messages amidst an information barrage. It’s a worthy goal: In a state of distress, trying to distinguish relevant information from dross can easily become overwhelming.
To this end, Twitter Alerts are marked with an icon of an orange bell and extend beyond the Twitter channel.
You as a Twitter user sign up for alerts through individual organizations’ Twitter pages or special alerts pages, like this one for Ushahidi. In addition to seeing alerts in your Twitter timeline, you can also receive them as push and/or SMS notifications on your phone. Whether you have or haven’t signed up for an organization’s alerts, the visibility of a Twitter Alert presently varies depending on how the Tweet is accessed.
Below is a screenshot of search results for #Iraq and Kirkuk, including a recent alert Tweet from @FCOtravel, one of the UK’s Foreign and Commonwealth Office accounts, which describes its mission in its Twitter bio as “Consular assistance for British nationals overseas from the Foreign & Commonwealth Office.”
No orange icon is visible. Nor is it visible when you expand the Tweet within these search results. If you click through to the Tweet details, however, it is clear.
Both of these examples come from accessing Twitter via the web. The next two examples come from accessing Twitter via the iOS app. From this point of access, neither search results nor Tweet details display the orange icon.
Device place matters: users see different things not only in conjunction with different location settings, but also when accessing services through different devices.
Following and DIY Citizenship Mashups
A number of emergency services send push or SMS notifications similar to Twitter Alerts. Earthquake alerts in Japan and AMBER alerts in the US use push notifications; your local school’s weather alerts probably have an SMS option; etc. What’s different, however, is that here Twitter provides the infrastructure to span devices and carriers. Twitter scaffolds contact between these places.
Historically, the categories of both who helps in natural disasters and who is helped have largely been organized around place, in this case mapped according to its political and geographic dimensions, by government agencies and relief organizations with parallel structures. Recently, social media has broadened the category of who helps — and in doing so, new technological places have joined political and geographic ones. Twitter and other platforms have been used to extend the reach of donation initiatives and to support techniques like crisis mapping and digital volunteering.
Both digital response techniques and donation patterns have attracted considerable attention from the news media and academia. But the ongoing transformation of who has been helped hasn’t garnered the same notice. These changes are more subtle, less eye-catching, less amenable to being woven into shiny progress narratives. At the same time, they suggest a significant shift in how individuals and institutions understand places, and consequently in the organization of power.
Twitter Alerts is grounded in the Twitter following relationship, and the following relationship is defined through the technological places of Twitter. A US resident can sign up for alerts from tenki.jp地震情報 in Japan, the NSW Rural Fire Service in Australia, and Rio’s Centro de Operações in Brazil. Individuals can choose civic memberships from a global pool. This makes sense: Labor, migration, and tourism patterns don’t fit neatly into a single set of national borders. Neither do many affiliations, communities, or social interactions. At its most extreme, this can yield a kind of DIY citizenship mashup, with individuals creating portfolios of government service relationships that are personally meaningful to them.
This is significantly different from the standard ‘the internet allows us to be global citizens’ line. In drawing on reputation and traditional institutions, the opt-in relationships enabled by Twitter Alerts don’t create a new network as much as reconfigure older, territory-linked services. Governmental and relief groups become overlapping, supranational organizations that serve not only resident citizens but also an informal, nonresident community.
Just as postal systems remade geographic places into zones determined by politics and history, social media technologies are remaking them today. But this is a decidedly different moment of history and politics, and the power balance among people, corporations, and states has shifted. The resultant new zones — and new configurations of zones — will further change this balance.
