Social Media – Truly Viable For Crisis Response?
To mitigate the unpredictability of crises and the complexity of crisis response, affected individuals and first responders are using new technologies, particularly social media, to help themselves. Examples include:
- Concerned citizens used a wiki after Hurricane Katrina to organize, collaborate, and rapidly create the PeopleFinder and ShelterFinder systems (Murphy and Jennex, 2006).
- Citizens affected by the 2007 San Diego Wildfires used a wiki to pool knowledge on which homes burned and which survived when the local media failed to support their needs (Jennex, 2010).
- Mumbai citizens used twitter to report their status, let others know where to find friends, relatives, etc., and to solicit blood donations following the 2008 Mumbai terrorist attacks (Beaumont, 2008).
- Victims trapped by falling debris during the 2010 Haiti earthquake used texting and/or Facebook to alert their friends/family to their location and condition (Boodhoo, 2010).
These anecdotes provide evidence of the value of social media to individuals in responding to crisis. However, the question has to be asked, is social media reliable enough for individuals and organizations to include social media in their crisis response plans? This preface explores this question by looking at perceptions of respondents to a survey administered after the great Southwest Blackout centered on San Diego, California on September 8, 2011.
Plotnick and White (2010) describe social media as generally being attributed to the collaborative applications supported by Web 2.0 technologies. These include, but are not limited to, Twitter, Facebook, MySpace, wikis, and blogs. Blogs, wikis, and MySpace were the first applications becoming popular in the early 2000s, while Facebook and Twitter are more recent creations. While cloud computing infrastructure is making social media applications more resilient and reliable, the methods users utilize to interact with social media applications are not. Most users access social media applications using their laptops, home computers, or mobile/smart telephones. Cloud computing infrastructure can be supported by highly reliable server farms with self contained back up power supplies to ensure they remain operational should grid power be lost. Private users don’t always have this luxury. Home connections rely on grid power to run their computers. Should that power fail, batteries are usually available for laptops, and some users have uninterruptible power supplies (UPS) for their desktops. In both cases broadband users rely on their routers to connect to their Internet Service Provider (ISP) and these routers may or may not be connected to a UPS. Mobile connections rely on the cell phone infrastructure to connect to the Internet. The cell phone infrastructure relies on a series of cell towers to connect mobile phones to the telephony system. Back up batteries are included in cell phone towers should grid power be lost. Ultimately, social media application reliability is not just reliant on having a reliable platform on which to run. Reliability is also dependent upon having a reliable connection system (cell phone infrastructure, landline infrastructure, cable infrastructure) and interface system (mobile, laptop, desktop). The reliability issue then becomes that of being a complex system of inter-reliabilities between multiple systems owned and operated by different, and sometimes, competing, companies.
This author served as the embedded systems and contingency planning project manager for Southern California Edison in their Year 2000, Y2K, program. A major concern of the program was understanding the risk caused by the complexity of systems used to communicate and operate the power grid. The communication (telephone and cell phone) systems were found to be as described above and it took much cooperation between the various system providers to generate an understanding of the relationships and complexities in the communication system, Backup power systems, including battery backups, were reviewed as part of the overall contingency plan should Y2K issues cause failures in the power grid and communication and control systems. It was observed that battery testing and replacement intervals were often not met prior to Y2K resulting in increased awareness and commitment to perform critical battery testing and maintenance. One of the concerns of this preface and a motivation for this research is a concern that this is still a practice and a risk following a loss of power event that causes social media to be unsuitable for crisis response and management.
The Great Southwest/San Diego Blackout
The Great San Diego/Southwest Blackout began at 3:38 PM on Thursday, September 8 2011 when a maintenance worker in Yuma, Arizona performing maintenance on a 500 kv transmission line caused that line to trip, stopping power flow to San Diego. The line was restored to operational status within several minutes but during that period of inoperability instabilities in the grid caused a power generator in Mexico to trip offline. The loss of the Mexican power generator caused further instabilities in the grid resulting in a low voltage situation that caused the San Onofre transmission substation and nuclear plant to trip offline. The San Onofre nuclear plant trip reduced available power while the trip of the transmission substation isolated the grid between San Diego and Los Angeles. The San Onofre transmission substation trip prevented the spread of the instabilities to Los Angeles and Orange County but also resulted in insufficient power for San Diego with the result that the San Diego/Southwest power grid collapsed and the region went into a blackout condition (CNN, 2011; Lee, 2011).
The blackout lasted approximately 15 hours affecting approximately 5 million people, 1.4 million of them in San Diego (Lee, 2011 and Page and Grieco, 2011). However, while the blackout was of relatively short duration, there were many significant problems. First, the blackout occurred at the start of evening rush hour and caused the failure of the traffic signaling system (virtually all traffic lights defaulted to blinking four way stops). This caused huge traffic congestion throughout San Diego with a dramatic increase in commute times (the author’s commute time went from 40 minutes to 2 hours, a fairly common example). Additionally, the trolley system failed with all trolleys stopped where they were at the time of the blackout as well as all train signaling systems causing non-electrical trains to have to reduce speed drastically. Finally, airports were forced to suspend operations due to reduced communication and control systems. Second, elevators, escalators, automatic doors, etc. in all buildings without backup power failed stranding a large number of people between floors or in stalled elevators. Third, water and water treatment equipment failed allowing some backflow and causing many residents to have to boil water before using it for humans. Fourth, refrigerators, freezers, stoves, televisions, and all home appliances without backup power failed. Among these failures were home computers and routers without UPS and laptops without charged batteries. Finally, and an additional motivation for this preface, the cell phone system experienced a great deal of degradation as observed by the author as he attempted to communicate with the local television stations to provide expert opinion on the causes and expected duration of the outage.
An early concern during the outage was that the blackout was the result of terrorist action. The concern was due to the failure of both main transmission lines into San Diego failed at nearly the same time. The author as well as the power company authorities was quick to ally this fear once it was recognized that the San Onofre substation failed on low voltage (this was recognized to be the correct action for the substation given the known failures in the rest of the grid system). However, getting this information to the affected population proved to be difficult as cell phone communications were failing for the author as he moved from cell tower to cell tower, and for the authorities as cell phone circuits were overloaded. These cell tower failures occurring within several minutes to an hour following blackout initiation is another motivator for this preface.
Recovery from the blackout began as soon as the cause of the San Onofre substation trip was understood and the restoration of the Southwest Powerlink (the original failed transmission line). Restoration began with restart of power generators and then the expanding recovery of distribution loops radiating out from the recovered power generators. The power grid was restored within 15 hours of blackout initiation.
This is an exploratory study that utilizes an exploratory survey to collect opinions and experiences related to the performance of social media during the Great San Diego/Southwest Blackout on September 8, 2011. The survey instrument was generated based on the research question of how reliable was social media during the Great San Diego/Southwest Blackout; and was distributed within three days of the event. Survey Monkey was used to administer the survey and San Diego State University’s Blackboard system was used to distribute a link to the survey to selected students at San Diego State University. Students were selected based on their having been students of the author. Responses were collected for two weeks with one follow up email sent. As this was an exploratory study with no theory or model testing attempted; survey results were analyzed using only descriptive statistics.
College of Business and Homeland Security students were selected as respondents as they are representative of the population expected to use social media, they were available, and they could be contacted rapidly. Rapid access was considered to be the primary requirement as it was important to collect survey responses while the event was still fresh in the memories of those solicited. Approximately 500 students were solicited. The exact number is not known due to the open invitation to pass the link on to acquaintances and family members as well as there being an issue with cross posting due to students taking multiple classes with the author. A total of 370 responses were received. All responses were used although some respondents did not answer all the items. To protect the identity of the respondents no personal identifiable information was collected. This eliminates potential follow up but was considered important for guaranteeing anonymity. The following reflects the demographics of the respondents.
Respondents were nearly evenly split on gender with 184 (49.9%) being male and 185 (50.1%) being female (of those reporting gender; 3 respondents did not report gender). As expected, the majority of respondents were mid twenties (since they were predominately graduate level students) although given the great diversity of the San Diego State University student body and a request to pass the survey link on to family and friends, there was a spread of ages as reflected in table 1.
Table 1. Respondent age distribution
The respondents also reflect a diversity of social relationships in the San Diego area (the largest population area affected by the blackout) as reflected in Table 2.
Table 2. Family relationships in the blackout area
This data was collected as it was felt that the larger the local social network of the respondent the more likely the respondent was to utilize social media to contact/coordinate actions during the blackout. As can be seen in Table 2, over two thirds of the respondents had some family social network in San Diego.
Table 3 shows the numbers of respondents who use various social media tools. This Table shows that large numbers of the respondents routinely use cell phones, text messaging, Facebook, and the Internet. Combined with the family social networks of the respondents it was expected that these respondents would be representative of those affected persons in the San Diego area that would use social media to contact and coordinate during the blackout crisis.
Table 3. Social media routinely used
Ultimately it is concluded that the respondents are representative of those that should have knowledge of the performance and availability of social media during the blackout and are thus a representative sample of those possessing that knowledge. Additionally, the sample size of approximately 370 is sufficient to analyze and come to conclusions on the performance and availability of social media during the blackout crisis.
Table 4 shows that respondents primarily used their cell phones (over 90%), text messaging (approximately 90%), and Internet access via their cell phone (approximately 70%) during the blackout crisis. It is somewhat surprising that so few used Facebook (approximately 40%) or Twitter (approximately 11%) given the popularity of both. This could be a reflection of the age of the respondents (primarily in their 20s) and reflects that this age group utilizes other social media.
Table 4. Social media activities attempted during the blackout
Table 5 reflects the observed availability of the social media services used. It should be noted that over 70% of respondents for every social media tool reported loss or degradation of service. This is backed up by Table 6 that shows that approximately 60% of the respondents attempted to use social media at least hourly to every few minutes. This reflects the author’s observation of frequent messages of “all circuits are busy” to “service unavailable” messages received during the crisis and possibly reflects an expected saturation of capacity condition as the affected population attempted to contact and coordinate with their social networks. Table 7 is somewhat surprising in that it shows that over 95% of the respondents only attempted to communicate with 10 or fewer individuals. It is postulated that this smaller number of contacts reflects the focus on contacting/communicating with family and/or close personal friends rather than reaching out to all acquaintances. This is not an unexpected observation or result.
Table 5. Observed social media service availability
Table 6. How often did you attempt to use social media?
Table 7. How many people did you attempt to reach?
Table 8 reflects observed loss of service during the blackout. It is interesting to note that television virtually disappeared and that Internet connectivity and land line based phone service were severely impacted. It is promising that while cell phone service was degraded, over 75% respondents had at least some cell phone service. This reflects that cell phones may be safe to count on in the initial stages of a crisis. However, it is expected that had the blackout lasted more than a couple of days then virtually all local cell phone service would have been lost. Table 9 shows that over 80% of the respondents were less than satisfied with the cell phone and Internet coverage available during the blackout. This is disturbing in that it shows there is an expectation that there would be better service.
Table 8. Communication media lost during the blackout
Table 9. Satisfaction with cell/Internet coverage during the outage
The first issue to discuss is if San Diego is representative of modern cities and that if experience with social media performance and availability following a blackout is potentially applicable to other cities. San Diego is the 14th most wired city in the United States (Forbes, 2010) while the United States is ranked 7th (OECD) in wireless users and 15th in broadband users (IT-Hall). This doesn’t make San Diego a leading city for being wired, but it is in the top tier of wired cities. This has a couple of implications. The first is that for other wired cities the San Diego experience is applicable and perhaps a harbinger that reliance on social media for crisis response following a major disaster that severely disrupts power distribution will not be successful. The second is that for those cities much less wired than San Diego and thus more reliant on traditional media such as television and land lines the situation following a disaster that severely disrupts power distribution may be even worse given the very poor performance of these media/services in San Diego. However, it is a fair assumption to determine that San Diego is a representative city and that the Great San Diego/Southwest Blackout is representative of what may occur during blackouts in the developed world.
The second issue to discuss is what does this all mean? It is clear that access to Internet sites was severely hampered. Home users virtually had no Internet so any Internet based communication via social media on the web would have failed. Additionally, any Internet based crisis response systems that can be accessed by home users would have been unavailable to those reliant on home based connections. Additionally, over 65% of respondents access the Internet via their cell phones and almost 90% use text messaging. This implies that a large number of respondents would have been able to access the Internet and use social media as well as use texting social media. Given that almost 70% of respondents attempted to contact someone within a few minutes to an hour of the blackout starting it is fairly safe to say there were many attempts to use social media to update status (about 35% tried Facebook, 11% Twitter, and 90% text messaging). The observation is that users expected to use their social media and mobile during a crisis. It is interesting that users did not expect a blackout to affect their cell phones or social media. Given that cell tower infrastructure uses battery backups it is interesting that there was so much cell phone degradation (77% reported less than usual signal strength, 83% reported lost coverage or degraded service). This wouldn’t be expected, although circuits being busy would be expected. It is suspected that batteries failed to perform as expected (although there has been no published accounts verifying this). This is troubling from a crisis response viewpoint as it would be expected that cell phone service should be fully available (as long as the cell phone tower infrastructure is physically intact) for 8 hours (per FCC order and as confirmed by FCC order 07-177 based on a review of communication failures following Hurricane Katrina) (Note that this rule also requires that phone switches and routers have 24 to 48 hour backup power supplies with additional fuel for generators on site with the equipment) (compliance with the FCC orders was to be within 12 months of the date of the orders (approximately October, 2008)). The implication is that perhaps the cell phone system is much more complex than expected and that battery backups are not all that are needed to ensure system operability following loss of grid power. Another possibility is that cell phone tower back up battery maintenance and testing is not sufficient, much like what was found during Y2K approximately 12 years prior. That backup battery maintenance and testing is not sufficient is a very troubling but somewhat expected possibility as these programs tend to suffer during tough economic times as experienced world wide the last few years.
In summary, users expected to be able to use their cell phones for Internet and social media access during the Great San Diego/Southwest blackout. However, most of them were unsuccessful or had limited success. That there was an expectation of available service is a reasonable expectation. Federal Communication Commission rules had specified, based on an analysis of Hurricane Katrina, that all cell phone towers be equipped with an 8 hour battery backup and that other phone system equipment be equipped with 24-48 hour backup power systems with available fuel supply. Crisis response planners also expect cell phones and perhaps social media to be available following a loss of power grid event. However, this did not happen during the Great San Diego/Southwest Blackout.
Ultimately the Great Southwest Blackout can be considered a massive, unplanned, backup battery test. Did cell phones and social media pass the test? This is somewhat debatable. On one hand 83.4% of respondents lost or had their cell phone service degraded, 93.4% had their Internet service lost or degraded, and 60.5% were not satisfied with their level of service or coverage following the blackout. On the other hand 78.5% of respondents had some degree of cell phone service, 35.5% had some degree of Internet service, and 32.5% were satisfied with level of service or coverage. While many may consider it surprising so many respondents had some service during a blackout; if we judge these results from a crisis response lens the blackout test is failed. The duration of the blackout was less than 24 hours. It is proposed that for a service to be considered for crisis response support it should be reliable and rugged enough to survive for a sufficient period of time to allow for utility/maintenance first responders to either preserve the system or implement backup systems. Less than 24 hours is not sufficient to ensure all cell towers, Internet access points, routers, and switches are in service. Backup power supplies should provide the time for this to occur and FCC rules require it. However, it does not appear that backup power supplies worked as anticipated resulting in the widespread system/service outages. This author, as previously mentioned, served as the embedded systems and contingency planning project manager for Southern California Edison in their Year 2000, Y2K, program. Backup power systems, including battery backups were reviewed as part of the overall contingency plan should Y2K issues cause failures in the power and communication and control systems. It was observed that battery testing and replacement intervals were often not met prior to Y2K resulting in increased awareness and commitment to perform critical battery testing and maintenance. A similar review was conducted by the FCC following Hurricane Katrina. It is concluded that much like the observed behavior of many organizations who vow to be prepared for the next crisis just after the current crisis ends but then have their enthusiasm and commitment wane as time passes and the “next” crisis does not occur, the same has occurred with backup battery testing and maintenance. The recommendation is that government and industry experts evaluate cell phone battery backup maintenance and test procedures for improvement. Additionally, once cell phone systems can be reasonably considered reliable, social media source sites should be evaluated for inclusion in crisis response management planning and if found to be useful, propose and publish design guidelines that will make social media systems reliable and rugged enough to qualify as crisis response systems (i.e. propose similar rules for battery/power supply backup, etc. as is currently in force for the cell phone system).
Additionally, social media providers generally did not expect that their services would be used for crisis response. Facebook, Twitter, wikis, et cetera were designed to be used to create communities, communicate with friends, and collaborate. The companies that created them did not design them to be reliable in crisis situations. This creates a problem for crisis response managers. Users will use the systems they are familiar with and use every day during a crisis (as evidenced by this study). It is only natural that users will use their social media first. The United States Emergency Broadcasting System does not include social media nor does most government managed crisis response plans, however, given the widespread unavailability of television and to some degree radio following the loss of the power grid, they should assess the viability of including social media as part of emergency broadcasting. Jennex (2010) noted that individuals have led the way in applying social media to crisis response and proposed that organizations use a knowledge management strategy approach to incorporate social media into their crisis response planning. Howe, Jennex, Bressler, and Frost (2011) discuss how self organizing groups are using social media to plan and prepare for large scale crises. The issue is users using innovative systems to do innovative things that the system designers never intended. The conclusion is that a discussion needs to occur between the crisis response community and the social media companies as to what is being done with social media in the field and what changes/enhancements . etc. are necessary to make social media reliable. Included in this discussion should be infrastructure concerns including how to harden and improve the reliability of social media in a crisis situation.
The title of this preface is “social media: truly viable for crisis response?” The conclusion of previous papers is that functionally the answer is yes, social media provide communication functionality that users want in crisis response (Plotnick and White, 2010; Jennex, 2010; and Howe, Jennex, Bressler, and Frost, 2011). However, the conclusion of this preface is that while the functionality of social media is useful, the maturity of social media from a reliability point of view is not sufficient to warrant including social media as operational crisis response systems at this time. Social media are fine for crisis response planning, but not for operational crisis response and it will take some thoughtful redesign of social media infrastructure before they are acceptable for operational crisis response.
Howe, A. W., Jennex, M. E., Bressler, G. H., & Frost, E. G. (2011). Exercise 24: Using social media for crisis response. International Journal of Information Systems for Crisis Response and Management, 3(4), 36-54.
Jennex, M. E. (2010). Implementing social media in crisis response using knowledge management. International Journal of Information Systems for Crisis Response and Management, 2(4), 20-32.
Murphy, T., & Jennex, M. E. (2006). Knowledge management, emergency response, and Hurricane Katrina. International Journal of Intelligent Control Systems, 11(4), 199-208.
Plotnick, C., & White, L. (2010). A social media tsunami: The approaching wave. International Journal of Information Systems for Crisis Response and Management, 2(1), i-iv.
Raman, M., Ryan, T., Jennex, M. E., & Olfman, L. (2010). Wiki technology and emergency response: An action research study. International Journal of Information Systems for Crisis Response and Management, 2(1), 49-69.