FutureCoast: A Playful Way to Assess Public Perceptions for Better Climate Change Communication

ABSTRACT

We examine how the FutureCoast storytelling game creates an accessible, online space to explore the climate problem and its impacts, as well as to glean insights regarding player perceptions. Through FutureCoast, players imagine a climate-changed future by creating stories about an altered world. A total of 251 voicemail responses generated from game participants recruited through social media and other channels were coded and analyzed. Subject engagement with the storytelling game provided valuable data about climate change understanding, as well as rich, player-created narratives that document the complexity of public thinking about climate-changed futures. Commonly occurring themes include Adaptation, Challenge, Technology, Weather, Governance and Policy, and Food. FutureCoast participants perceived optimistic scenarios for technology, energy and mitigation, and pessimistic scenarios for weather, food, water and adaptation. From FutureCoast stories, we gain an understanding of public perceptions toward climate issues that can help communicators develop more informed and effective climate change communication strategies.

Key policy highlights

• Through playful approaches, such as FutureCoast, we can gain an understanding of public perceptions toward climate issues that can help communicators develop more informed and effective climate change communication strategies.

• Using novel approaches such as games to understand perceptions can elicit information from people who would otherwise not engage in surveys or other research methods.

• An innovation of the FutureCoast approach is its ability to produce rich, player-created narratives, which can be analyzed to uncover complex thinking about climate-changed futures. Responses may reveal where the public identifies and voices emerging issues earlier than experts.

• Identifying optimistic and pessimistic trends around climate issues gives communicators the opportunity to re-frame negative climate perceptions toward actions and solutions, thus empowering their audiences with information that can elicit climate action.

KEYWORDS:

• Climate change
• stories
• games
• engagement
• perception

Introduction

Though public awareness of climate change and concern about it has grown steadily in recent decades, public understanding of climate change and its impacts still differs from scientific knowledge in important ways (Weber & Stern, 2011). A recent survey documented this gap (Leiserowitz et al., 2023). Though 74% of Americans think that global warming is happening, 61% of Americans think of global warming as primarily caused by humans, and 66% are somewhat worried or very worried about it, less than half (48%) think that Americans are being harmed by it at present, and even fewer (41%) state that they have experienced its effects personally – rates much lower than current scientific assessments (USGCRP, 2023). Only 20% know that the broad majority (>90%) of scientists agree that human-caused global warming is now occurring. Recent research has identified such misperceptions as an obstacle to support for climate action (Bogert et al., 2023; Bohdanowicz, 2021; van der Linden, Leiserowitz, et al., 2015). In this context, it is crucial that educators and communicators connect with the public in ways that draw on current levels of awareness and concern to facilitate engagement and improve understanding of climate issues (Kumpu, 2022). New methods are necessary to create an encouraging space for climate change discourse (National Research Council, 2011; Turrin et al., 2020; van der Linden, Maibach, et al., 2015), especially methods that incorporate the cultural domains of values, place, power, and narrative (Munshi et al., 2020). Because climate change is infrequently discussed among friends and family (Leiserowitz et al., 2019; Maibach et al., 2016), and the majority of people are not engaged in formal, classroom education, there is a need for informal settings that promote climate literacy and normalize the reality of climate change (Pfirman et al., 2021; Spitzer, 2014).

Games and simulations have been pointed to as having potential for creating enjoyable engagement with climate change in both formal and informal educational settings (Eisenack & Reckien, 2013; Pfirman et al., 2021; Reckien & Eisenack, 2013; Wu & Lee, 2015). Games can lessen the intimidating nature of climate change by serving as a non-political and non-expert venue for discourse on the subject. Games about climate change may appeal to a broad audience and allow them to engage with climate issues extending beyond their own current experiences and expertise (Skains et al., 2022; van Beek et al., 2022). A study by Allen and Crowley (2017) shows that communication efforts that are relevant, participatory, and interconnected are more likely to promote climate engagement. Storytelling games in particular promote collaboration and the exchange of ideas, insights, and experiences among the contributors, influencing their beliefs (CRED & ecoAmerica, 2014; van Beek et al., 2022) and building their awareness of the importance of cooperating with others (Schenk and Susskind, 2014). By engaging in future-thinking and creating stories for a game, players must deliberate on climate change and potential risks, which requires recalling knowledge, comprehending it, and applying it (Anderson & Bloom, 2001). Other research looking at exercises with integrated futurity and behavior change have shown that merely bringing future risks into a person’s consciousness increases the likelihood that an individual will begin planning or make a change in lifestyle to prepare for or mitigate those risks (Spence et al., 2011). As Nisbet (2009) found, storytelling can be used to “bring diverse audiences together on common ground, shape personal behavior, or mobilize collective action” (p. 22). Considering the role of storylines, Young et al. (2020) stated that “[t]his alternative approach does not make predictions as such, but provides information and a space to consider future possibilities and potential actions associated with these.” A study of role-playing simulation games, in which players jointly build stories, showed that they can “provide effective science-policy engagement tools that allow players to engage in future thinking and corresponding meaning making” (van Beek et al., 2022)

Increasingly citizen and crowd science (Scheliga et al., 2018; Show, 2015) and participant (player, stakeholder) input to simulations are used in co-creation of approaches and solutions, most famously through Foldit (Eiben et al., 2012; Groh, 2012; McGonigal, 2011; Radchuk et al., 2017). However, to date, most research on climate games has focused on using them as vehicles to communicate information, raise awareness, motivate action, and role-play decision-making (Flood et al., 2018). Research is more limited on the use of games to elicit information from the players useful for developing climate communication, adaptation, or mitigation strategies (Arlati et al., 2021; van Beek et al., 2022; Wibeck & Neset, 2020). Here we analyze participant storytelling responses to the online FutureCoast prompt to envision “possible climate-changed futures” to gain insight into player perspectives (https://soundcloud.com/futurecoast, Supplemental Materials). FutureCoast is the first game of its kind to invite the public to contribute to such envisioning (FutureCoast, 2014). A 2020 review of “20 Incredible Environmental Games from the Last Decade” described the game in this way (King, 2020):

FutureCoast was a hauntingly-beautiful showcase for real people to share their ideas, concerns and predictions about climate change. It encouraged everyday people to face the reality (and alternate reality) of our world’s climate changed futures – this installation paved the way for the advent of “CliFi” – Climate Fiction.

Highly participatory (Radchuk et al., 2017), FutureCoast engages the public in deliberation on climate impacts while simultaneously providing information to us as researchers about the public’s understanding of and attitudes toward climate change, which is considered the “knowledge-attitude interface” (Sturgis & Allum, 2004). Traditional deficit model-based public understanding of science research assumes that a lack of information by the public is the driver for negative, skeptical, and/or biased risk perceptions (Bauer et al., 2006). Under the deficit model, polls and other quantitative methods of assessing public understanding that indicate a lack of public concern about climate change or skepticism about scientific consensus would imply a deficiency in their understanding of climate science. While this may be the case [for some], continuing to provide more decontextualized information may not improve public understanding of climate change and its risks (Bauer et al., 2006). Furthermore, as Allen and Crowley (2017) argue, public comprehension of climate science may not be the most relevant factor in inspiring the public to act on climate change and endorse meaningful public policy and programming to prepare for it. For example, Moerkerken et al. (2020) show that the likelihood of Dutch farmers to undertake measures to reduce their energy use and to mitigate their greenhouse gas emissions depends more on their understanding of, and deeper engagement with, agricultural systems than on their knowledge of climate change.

FutureCoast players can engage with the climate topic regardless of current experience or knowledge. Furthermore, through play, they use their sociological imagination (Cataldi & Iorio, 2022), which, as noted by Kumpu (2022) can “enhance understanding of public engagement and communication as systemic phenomena that reach far beyond individual attitudes, emotions, and behavior.” At the same time, responses provide an opportunity for communicators to observe and gauge the public’s understanding and perceptions about the issue. These stories from individuals can also identify preconceived ideas about the impacts of and solutions to climate change (Marx et al., 2007; Moezzi et al., 2017).

Materials and methods

Intervention: the FutureCoast game

The intervention, FutureCoast (2014) was created by game and experience designer Ken Eklund as part of the Polar Learning and Responding: PoLAR Climate Change Education Partnership (www.thepolarhub.org), which included academic researchers, game designers and other creative professionals, and members of non-governmental organizations and Indigenous Peoples. FutureCoast is an interactive online storytelling game that encourages users to explore a future reality by leaving voicemails, that per the gameplay, have leaked back from the future to listeners in present day (Supplemental Materials). By using voicemails as a form of expression, players no longer consider climate change and its impacts as abstract concepts occurring in the remote future, but rather they become visceral, first-person experiences for both players and listeners. Through this foresight narrative device, participants are encouraged to think concretely about the near-to-intermediate future (2020–2065) and what a climate-changed reality would be like. FutureCoast does not prescribe a certain degree of climate change or set of impacts occurring as a condition of the game; rather, a variety of futures are built from the players’ own existing knowledge and imagination.

FutureCoast crowdsourced voicemails during an active period from early January through the end of May 2014. This online experiment elicited open-ended, variable length, and unstructured narrative responses which could be coded along a number of continuous and categorical variables. The research objective was to explore the hypothesis that participant-generated content in interactive online storytelling games like FutureCoast can provide valuable insight into public understanding of climate change. Recognizing the potential for the game to elicit such a range of responses, the researchers prepared a coding scheme for analysis which covered a range of domains, including time horizon, the nature of climate impacts, the range of proposed responses, and the affect that was expressed.

Sample

Subjects were recruited through several channels: through Google searches; through links in websites; and through invitations distributed through social networks. Voicemails used in this study were collected from the FutureCoast Hotline from January to the end of May 2014. Participants called the Hotline and a prompt asked them to leave a voicemail from a possibly climate-changed future. Callers could then choose to first listen to previously recorded voicemails, or go directly on to leave their own. They were prompted to choose a year from 2020 to 2065, and to leave a voicemail of up to three minutes long, imagining that they are speaking during the year that they chose. This research was approved by Columbia University’s Institutional Review Board (protocol AAAR8243 titled “FutureCoast”).

Research protocol

A total of 265 voicemails were collected during the live period. Four were excluded from the dataset because they were duplicates or contained obscenities. Ten additional voicemails were excluded because they were in a foreign language or inaudible.

The final dataset consisted of 251 voicemails, with a URL to the audio file and a voice-to-text transcription for each. Voicemail transcriptions were manually reviewed, validated and re-transcribed if incorrect. Fourteen of the 251 usable voicemails had partial transcriptions due to one word or more being indecipherable; however, the content of the voicemails was considered intact and the voicemail was included in our analysis. The final data file included each voicemail duration and the player-selected year, as well as the ID number of the chronofact (an aspect of game play, described in greater detail below) when available.

Subject audio responses were stored, logged, and transcribed. We transcribed the audio into text, and then coded the responses. To analyze the composition of the voicemails, three types of codes were developed: content themes, sentiment groups, and climate change response categories (Supplemental Materials). Each code received a binary code indicating the presence (1) or absence (0) of that theme in each voicemail. Codes were evaluated for reliability by assessing the level of agreement among three independent coders.

Content themes were inductively established from the voicemails (Marceau et al., 2022). These codes characterize the climate impacts a player envisions. For example, a voicemail describing weather events such as extreme temperature or changes in precipitation would be coded for the Weather content theme. Each voicemail was initially coded by a single researcher, who detected themes in the transcribed voicemail. After the initial draft set of codes was established, the draft set was refined into a final set by the research group to reduce overlap while ensuring all major voicemail themes were represented. This process resulted in 14 unique content codes that are broad enough to encompass similar issues, but distinctive enough to minimize redundancy. Each voicemail was coded for a minimum of one content theme, but could be coded for an unlimited number of themes. Qualification for the code included direct mention of any theme names or parts of the theme description (Supplemental Materials).

Sentiment codes were evaluated to understand the emotional state of the fictional character speaking in the voicemail. Although the voicemails are fictional, the sentiment or affect a player chooses to act out can indicate their attitude and perception of a specific climate issue. For instance, a climate impact that arouses negative sentiments may indicate the player perceives that impact as a threat. By comparison, a demonstrated relaxed or happy disposition could be inferred as the opposite depending on the content of the voicemail.

Affect was determined by assessing the valence and arousal of the voicemail and speaker's voice. The valence indicates how attractive (positive), unattractive (negative), or neutral the context of the voicemail is to the player. The arousal is a measure of the voice itself and how activated, deactivated or neutral it is in the voicemail. A list of common sentiments was compiled (Supplemental Materials), and voicemails were coded for a minimum of one sentiment up to an unlimited number of sentiments. Similar sentiment codes were then clustered into five groups and used during subsequent rounds of validation coding.

While the content themes emerged from open coding of the voicemails, the climate change response (CCR) codes were predetermined by the research group based on existing recognized climate change dialogues within climate change communication literatures (Baden et al., 2022). They were included to assess the participants’ knowledge and awareness of climate change effects and potential response strategies. They included: (1) Adaptation, (2) Mitigation, (3) Challenges, (4) Opportunities, (5) use of Scientific Information, (6) Bottom-up Strategies, (7) Top-down Strategies, and (8) Physical impacts. Unlike the content themes and sentiment groups, for which all voicemails received at least one tag, CCR themes did not have a minimum requirement and were coded as present only if mentioned. Additionally, no CCR codes are mutually exclusive and it is possible for a single voicemail to be coded for two complementary codes such as Opportunity and Challenge. Voicemails that qualified for the CCR codes frequently demonstrated an understanding of climate-specific topics such as adaptation and mitigation. Some messages explicitly stated adaptation technologies or mitigation strategies, while others implied them. Challenges and Opportunities were often explicitly stated in voicemail messages or easily inferred based on the content of the voicemail. The use of Scientific Information was determined based on the presence of relevant climate science references and terminology and Physical Impacts referred to physical changes that occurred due to climate change. Bottom-up Strategies included voluntary individual or community responses, while Top-down Strategies included government or industry responses. Detailed descriptions of CCR themes can be found in the Supplemental Materials.

Code validation

The initial draft of the codebook, with development of definitions for the content, sentiment, and climate response themes, was pre-tested by the three coders, noting themes which the coders found ambiguous or did not apply consistently. Discussions among the research team led to revisions of the codebook into a clearer, more explicit format, which was then applied.

In addition, the codes were validated through intercoder agreement analysis using the Reliability Calculation Program (ReCal). The agreement measure selected was Krippendorff’s Alpha due to its ability to handle nominal data, to compare more than two coders at once, and to account for chance agreement. The agreement threshold required for themes to be included in further analyses was .667, as suggested by Krippendorff (2004) for exploratory data.

Two of the three coders met the agreement threshold for sixteen of the original twenty-seven themes. The third coder was used as a tiebreaker for disagreements between the first two coders, which is a common method for resolving coder disagreements in content analysis (Edwards & Lampert, 1993; Lombard et al., 2002; McDonald et al., 2019; Reis et al., 2000). Nine themes were excluded due to low agreement coefficients. Two CCR themes with lower agreement (Adaptation and Challenge) were included in our analysis because they formed complementary pairs with other CCR themes that did meet the agreement threshold (Mitigation and Opportunity). Each voicemail was originally coded for at least one content theme and one sentiment group; however, some of these initial codes did not meet intercoder agreement. Thirty-four (34) voicemails did not receive a content code that met this threshold, and were therefore marked as “Other.” and were excluded from some of our thematic analyses (Supplementary Materials). Five (5) voicemails did not receive agreement for a sentiment code, and were marked as “Indeterminate.” These codes were only used if no other content theme or sentiment group was coded for.

Code communities

As part of obtaining useful information for communicators we performed a network analysis on all code types that reached intercoder agreement.

We used a fast greedy modularity maximization algorithm (Newman, 2006) to detect communities, or subgroups, within the network that were used in extracting additional information about public knowledge from the voicemails. The algorithm split the codes into two communities. We determined that the division is primarily driven by greater association with positive or negative sentiments. We selected names for the communities, optimists and pessimists, based on the sentiment codes included, coupled with presence of the Opportunity or Challenge CCR codes. The “pessimistic” or challenge-focused community included all the negative sentiment groups such as Scared/Panicked, Angry/Annoyed and Sad/Defeated. Themes that also fell into this community were Challenge, Conflict, Weather, Migration, and Adaptation which generally have negative connotations in the context of climate change. The “optimistic” community included the Happy/Excited and Neutral sentiments, and the content and CCR codes of Opportunity, Energy, Technology and Mitigation.

Complexity

Complexity is a widely used measure which characterizes the co-occurrence of multiple codes, a complement to frequency, a measure used to describe individual codes taken in isolation. The complexity of each code, a measure of how frequently it co-occurred with other codes, was determined by first assigning each voicemail a complexity score corresponding to the total number of content and climate themes it contained. Complexity scores were calculated by standardizing the respective totals and using the z-scores to represent the frequency of co-occurrence; each voicemail was compared to the average. Then the mean of the z-scores was calculated for each content and CCR code using all voicemails in which the code appeared. We considered what the complexity of certain codes, examined in conjunction with the code’s prevalence, could imply about players’ risk perceptions. In particular, we looked at the salience of some climate change issues and how well they are understood as a system in relation to other codes.

Results

The results demonstrate what callers found notable during this future-thinking exercise. They were analyzed to determine which climate change risks are more salient to the participating public, as shown by the frequency of mention and by the presence of emotional or affective responses.

Temporal setting

As part of the gameplay, players could choose when the voicemail they recorded took place. Their choice could be any year between 2020 and 2065, and they could then give further details on the time they described within the voicemail content, for example, by mentioning a season or holiday to contrast with expected weather events. Analysis of the years chosen revealed interesting characteristics in player choices. Peaks on the extreme ends of the period, years 2020 and 2065, are artificial in that they include all voicemails with earlier and later dates as seen in Figure 1. Peaks in years 2024 and 2059 are explained by the game prompt having mentioned them, which indicates the importance of what information is introduced to the public in framing exercises such as this. 2034 was 20 years after 2014 when the program ran. The round number years 2025, 2030, and 2050 experienced peaks in voicemails. These also happen to be common years scientists use to make projections, which could indicate the use or knowledge of such information by players. Aside from these peaks, the generally even distribution of voicemails indicates that players could engage broadly and envision across a time period that spans two generations.

Figure 1. Frequency of voicemail setting year chosen by player. Part of the player experience is choosing a year in which one’s voicemail comes from. This graph demonstrates the frequency of voicemails by year over the game-play period.

Codes and content

Content codes were the most robust category (Figure 2). Voicemail coding and validation resulted in ten content codes included in our analyses. The prevalence of each theme was calculated as a proportion of the number of voicemails the theme was mentioned in out of the total number of voicemails (Supplemental Materials). 13.5% of voicemails were coded as Other because the themes for which they were originally coded were excluded due to low intercoder agreement.

Figure 2. The prevalence of content and response code: complexity, expressed as average complexity z-score, showing the likelihood of each code co-occurring with other codes (top), and prevalence, expressed as the frequency of the code within all voice mails (bottom).

Of the nine (9) content codes that were included, Technology was the most prominent, occurring in 33% of voicemails, and Weather appeared in 31% of voicemails. The third most prevalent content code was Governance and Policy, which appeared in 22% of voicemails followed by the Food code with a 20% prevalence. The Energy code was present in 16% of voicemails, Health code in 14% and Water in 13%. Conflict and Migration were the least common of the content codes, and occurred in 7% of voicemails.

Four climate change response (CCR) themes were included in the analyses: Opportunity, Challenge, Adaptation, and Mitigation (Figure 2). These themes form two groups of complementary, but not mutually exclusive, CCR pairs: (1) Opportunities and Challenges, and (2) Adaptation and Mitigation. The Mitigation code appeared in approximately 15.5% of voicemails, while Adaptation had nearly triple that with references in 42% of voicemails. The Mitigation code included any technological, behavioral or political change to reduce greenhouse gas emissions. Common contexts in which this code fit included discussions of sustainable transportation, reducing energy demand and increasing efficiency, carbon accounting and replacing fossil fuels with renewable energies. Voicemails coded with Adaptation could include adaptation to physical phenomena such as building infrastructure, or changes in lifestyle in response to environmental pressures (e.g. risk-avoiding behavior).

The Opportunity code was prevalent in 21.5% of all voicemails and its counterpart, Challenge, appeared in almost three times as many voicemails (60%). Opportunities are described as positive outcomes or potential outcomes due to climate change, while challenges are unresolved difficulties. It should be noted that the high prevalence of the Challenge code may be because often an opportunity or solution could only be presented in a voicemail by stating a challenge, and thus would be coded for both codes, whereas a challenge could be presented without the explicit identification of an opportunity.

Affective response

Although risk can be defined objectively, members of the public take action based on their subjective risk perceptions, which often include an element of emotion. The ways in which individuals process risk relies heavily on information, and on feelings about that information; this phenomenon is called the affect heuristic (Leiserowitz & Smith, 2010; van der Linden et al., 2019). Climate change policy preferences of Americans are particularly influenced by certain emotions, such as worry and hope (Brosch, 2021). It is argued that affective reactions occur independently of and precede decision making, and thus influence risk perception (Leiserowitz, 2006). Although affect may distort risk perceptions, it is important to take into account the fact that emotion and reason interact, influence perceptions, and therefore influence behavior (Roeser, 2012; Skagerlund et al., 2020). We assessed participant climate risk perceptions using both the cognitive content and the affective content of the voicemails (Figure 3). The emotions associated with varying climate themes elucidate biases in the public consciousness and can inform climate communication strategy. Six sentiment groups were analyzed. Happy/Excited sentiments occurred in 30.7% of all voicemails and Neutral sentiments occurred in 25.5% of voicemails. Scared/Panicked, Sad/Defeated and Angry/Annoyed occurred respectively in 24%, 19% and 13.5% of voicemails for a combined total of 56.5% with negative sentiments. Five of the voicemails (2%) were coded as “Indeterminate” due to low intercoder agreement.

Figure 3. Bar graph showing the prevalence of sentiment codes where 1.00 equals code presence in 100% of voicemails.

Communities – optimists and pessimists

Based on the network analysis we conducted examining the relationship between the voicemail codes, codes fall into one of two communities, one that indicates optimism and the other pessimism about the future in the wake of climate change (Figure 4). The following pessimistic and challenge-focused voicemail was coded as Water, Energy, Governance/Policy, and Adaptation and Challenge, as well as sentiment groups Scared/Panicked and Sad/Defeated:

Hi, it's me. Listen, besides the scheduled brownout, the city just declared an emergency and turned off the water. To all of the city. If you've got water, bring home as much as you can. Get back to me if you can, I have no idea when it’s coming back on. Take care. Love you. Bye. ([Chronofact ID 168453-87112905] Soundcloud: https://soundcloud.com/futurecoast2/stay-home-if-you-have-water)

The “optimistic” community (solutions-oriented) includes the more positive sentiment groups, as well as Opportunity, Mitigation, Energy, and Technology. For example, the following voicemail demonstrated an optimistic sentiment (Happy/Excited) and was solution-oriented, containing the content themes of Food, Health, Energy and Technology, as well as both the Adaptation and Mitigation CCR themes:

Figure 4. Alluvial diagram showing the relationships between code types and the communities in which they occur most frequently. All codes considered in our analyses are found down the center of the diagram. The thickness of each band indicates relative code prevalence within its code category. The code category each code belongs to is indicated on the left of the diagram, and the community each code falls into is indicated to the right. (*Includes content codes that did not meet intercoder agreement. **Indicates climate change response (CCR) codes that did not have sufficient intercoder reliability to carry out full statistical analysis in other parts of the study but were included here as a complementary pair to other CCR codes.)

Hey Kelly, this is Sheila. Just wondering how you are. I have to say, I just had the best workout. In 10 min, I generated 15 kilowatts*. I wanna see you beat that in the feeds. Whoever wins gets to treat the other to some vat-grown beef on Friday. So, let's see if you beat me. Anyway, hope your day was better than mine. But I don't think that's possible! Bye. ([Chronofact ID 23585-77019284] SoundCloud: https://soundcloud.com/futurecoast/2020-the-best-workout. *note that humans cannot actually generate that much power)

Some of the more prevalent content themes reveal shared risk perceptions amongst players. More frequently occurring codes could indicate more salient climate topics for people, and when coupled with the expressed sentiment and CCR code, a clear picture is formed. For example, our results showed that Technology, which was the most frequently occurring content theme, appearing in over 33% of voicemails, was often associated with optimistic sentiments, as demonstrated in the following voicemail:

Hi, Bro. It's me. Welcome back to San Francisco. You won't recognize it though by the time you get here. We've hired this bunch of Dutch experts, you know, building dams and dikes along ocean beach and stuff like that. They've been doing a great job but on the positive, we're now breeding saltwater plants for hydroponic farms. And we're even raising salmon and other fish in rice ponds and catchment basins for salt water. But what's really exciting to me is that the most important is that our Pacific Reservoir. We've begun desalinization plants all along the coast, and we're supplying fresh water now not only to farms, but also to all the people around here. So, thankfully they have that breakthrough in science and technology to reduce the cost of desalinization. Chat with you when you land. Let me know. Bye-bye. ([Chronofact ID 65599-70920792] SoundCloud: https://soundcloud.com/futurecoast2/city-changes-san-francisco)

The topic of Weather, which was mentioned in over 30% of voicemails, was viewed much more pessimistically, and can demonstrate a depth of understanding of potential future risks:

Hey mom. Just wanted to let you know that Brody, Dad and I are heading back early. We thought maybe the heat would break, but they're saying it will stay like this at least until late next week. It wouldn't be so bad if we could still swim in the lake. The algae's back again. Oh, and the bugs up here have seriously taken over. I think next time we should bring mosquito nets. It was bad. I'm all bit up. You wouldn't believe what the cabin looked like when we got here. It was like a horror movie, I swear. Bugs everywhere. And no, I'm not exaggerating. Brody took pictures. Anyway, we should be back by tomorrow afternoon. We're gonna try to find a hotel with AC for tonight. Love you. ([Chronofact ID 94217-05124361] SoundCloud: https://soundcloud.com/futurecoast/2017-the-lake-like-a-horror)

Climate misconceptions

The complexity of climate change makes it difficult to know how to evaluate existing public knowledge, but stories provide an opportunity to see just how information is constructed in players’ minds and where there are disconnects between perception and fact. The playful narrative style of participation allowed us to obtain an unencumbered view of misconceptions held by players. For example, one prominent misconception was a scientifically unfounded relationship between climate change and earthquakes and other non-climate related risks (e.g. tectonic shifts, meteor showers, etc.):

Hi, Jack. This is Tony, we are scheduled to plan, to correct the issues with the tectonic plates of California. If not taken care of the massive Earthquakes will split California into 5 islands. One of which will drift up north killing thousands in Alaska. As you know the other four islands will become under chaos with dictators that will destroy the world starting with the US. We need the tectonic arrangement device. Please get in touch with me as soon as possible. Thanks, Jack, for joining the initiative. ([Chronofact ID 90750-88391460] SoundCloud: https://soundcloud.com/futurecoast2/tectonic-plate-initiative)

We also encountered exaggerations of near-term impacts of climate change. These could reflect serious misconceptions about climate change on the part of the caller, or playful humor. In the following voicemail which is supposed to occur in 2062, the player indicates that sea level rise (SLR) has inundated a large portion of the state of California, and that coastlines have reached Nevada. Such a dramatic impact is not projected by even the most aggressive sea level rise models within the next century:

Ah … hello sweets. Been awhile since we've talked. Yeah we just got done moving to Nevada it's so awesome we haven't been this close to the ocean since uh, since we lost our home in Santa Cruz. But it's great to be here, your mother loves Reno and I sure hope you decide to come and join us one of these days. Well, talk to you soon. Take care, and don't forget to eat your greens! Bye. ([Chronofact ID 85471-07987160, year 2062] SoundCloud: https://soundcloud.com/futurecoast2/all-moved-in-reno)

Discussion

This analysis shows that an exercise of future-thinking, a playful experience for many participants, generated abundant voicemail content regarding participant understanding of climate impacts and the future. FutureCoast’s crowdsourced climate futures provided us with information on which topics are of importance to the players, what opportunities and challenges they foresee, and how those potential futures make them feel. Through the game, players consider a wide variety of risks and solutions and personally engage with climate change. FutureCoast player stories, which emerged from engaging their sociological imaginations (Kumpu, 2022), produced a rich dataset from which we could observe their current state of knowledge, attitudes, priorities, and expectations. Consideration of cultural domains (Munshi et al., 2020) included narratives through story creation and frequent references to place to set the context; these domains also addressed their values and their concerns about power relationships. Players’ understanding encompassed a broad range of climate change topics and demonstrated a high level of thinking. As noted by Waddington and Fennewald (2018), bringing to the surface tacit and usually unexpressed knowledge is an important aspect of education, since it contributes to fuller understanding and more extensive use of that knowledge. The voicemail data also revealed misconceptions about climate change that otherwise could have gone unnoticed in a traditional survey or other analysis. This added qualitative and contextual information has been argued to be missing from traditional public understanding surveys, and eessential to fully assessing public knowledge (Bauer et al., 2006).

Player self-selection is an important component of the FutureCoast game design. A wholehearted participant is “self-motivated and self-directed, intensely interested and genuinely enthusiastic,” all of which are essential in future-thinking games (McGonigal, 2011, p. 138). In the case of FutureCoast, the premise that a player must opt-in contributes to the quality of content generated and genuineness of their expressed perceptions and attitudes.

We found that overall players had a broad and diverse understanding of expected climate change impacts and societal implications. Content spanned over 10 unique codes. For context, we compared the sectors listed in the table of contents from the 2014 National Climate Assessment (NCA) from the US Global Change Research Program (Climate Change Science Program (US), 2014) to the FutureCoast content themes (Figure 5). Note that we are using the 2014 NCA for comparison because it was contemporaneous with the voicemails analyzed. FutureCoast voicemail themes exhibited a greater emphasis on social issues, while the NCA table of contents had greater focus on scientific themes, natural systems, and specific communities. Voicemails discussed topics that covered resource availability, physical environmental impacts, economic and political ramifications, and the psychological effects of climate change. These findings show that the scope of climate change impacts envisioned by the players is on par with the priorities of experts, yet imply that social impacts are more salient for the public. While there are differences in the risks non-experts found when compared to experts, the players are able to comment on an equally wide range of themes. In addition, the majority of codes co-occurred with one or more other codes of the same code category. Drawing on the long tradition of research on mental associations within psychology (Andrews-Hanna et al., 2022), we suggest that this pattern reveals specific features of the players’ mental models (also known as representations, schemas, or cognitive frameworks). More specifically, it indicates that the players did not treat these codes as isolated elements, but rather that they link codes with other codes at a similar location in social and/or environmental systems (content with content, sentiment with sentiment, response with response); such groupings are known as categories within cognitive psychology (Markman & Rein, 2013), and have been connected with creative activity, such as construction of narratives within game-play, as in this study (Beaty & Kenett, 2023). This set of categories, or pattern of linkages, suggests that their mental models include the interconnectedness of these social and/or environmental systems when considering future climate impacts. Other recent work has also drawn on similar association tasks, asking individuals to provide verbal responses to prompts and examining the patterns of co-occurrence in responses to establish mental models. Lynam (2016) asks people to develop a narrative response to a hypothetical situation in which they explain adaptation to others; Dézma et al. (2024) prompted individuals to list a number of terms associated with a climate change prompt and analyzed their responses to show the relations between emotions and salient climate impacts. In a public health context, Emiliani et al. (2020) sought to examine the views of research subjects to life during the COVID pandemic; they used two association tasks: prompting subjects to compose a short narrative of their experiences and asking them to list associations to a novel prompt.

Figure 5. Comparative Analysis of the FutureCoast Codes and 2014 US National Climate Assessment Sectors (Climate Change Science Program (US), 2014). (*Includes voicemail content codes/themes that neither had sufficient intercoder reliability nor corresponded to a NCA sector (i.e. Economy, Infrastructure, Industry/Business, and Society). **Ecosystems/Landscapes did not have sufficient intercoder reliability to carry out full statistical analysis in other parts of the study, but was included here to demonstrate an overlap with the NCA sectors.)

To see if the voicemail themes that were not called out in the 2014 NCA table of contents emerged as important enough to highlight by the 2018 NCA (USGCRP, 2018), we first examined the table of contents. None of these voicemail themes (weather, technology, governance, policy, migration, or conflict) were represented in the new 2018 NCA content areas which were: Land Cover and Land-Use Change, Coastal Effects, Oceans and Marine Resources, Air Quality, Climate Effects on U.S. International Interests, and Sector Interactions, Multiple Stressors, and Complex Systems. However, analysis of the frequency of occurrence of the key term representing each theme, indicates two points (Table 1). First, there were frequent references to weather and policy in the 2014 assessment, so these were seen as important issues and it was a choice to not give them prominence in the table of contents. Second, the comparison of word frequency normalized by the number of pages, indicates that conflict and perhaps migration, emerged more prominently in the expert assessment in 2018 than in 2014 (Robinson, 2022).

Table 1. Frequency of occurrence based on word searches of the 2014 and 2018 National Climate Assessment reports.

	2014 NCA		2018 NCA		Ratio 2018/2014
	Freq	Freq/# report pages (829)	Freq	Freq/# report pages (1515)		Normalized by report page length
Conflict	24	3%	112	7%	4.7	2.7
Migration	83	10%	179	12%	2.2	1.2
Weather	390	47%	620	41%	1.6	0.9
Governance	50	6%	59	4%	1.2	0.7
Technology	114	14%	128	8%	1.1	0.6
Policy	282	34%	304	20%	1.1	0.6

Others have also used public engagement to identify emergent issues, which can then be translated into management decisions. However, such engagement activities often require more extensive interactions and involve smaller numbers of people than FutureCoast. As just one example, Weller et al., 2021, using a deliberative process called participatory technology assessment (pTA), noted that:

… in 2014 the National Aeronautics and Space Administration conducted a pTA forum about a proposed asteroid mission, where federal officials were surprised to find that the public independently desired a federal role in protecting Earth from asteroid collisions. These detailed citizen concerns fed into NASA’s decision to create an office for coordinating planetary defense.

Turning from analysis of themes to analysis of sentiments, it is important for researchers and communicators to distinguish between issues that evoke negative emotions from those that evoke positive emotions (Brosch, 2021). As Sweeny et al. (2006) stated, “optimism fosters a positive mindset to undertake challenges with confidence.” Our data show that visions of the future fall into two communities: challenge-focused with negative sentiments (pessimistic), or solutions-focused with positive or neutral sentiments (optimistic). Not surprisingly, these findings indicate that optimists are solutions-oriented. Typically, voicemails that fall within the optimistic community include technological improvements that mitigate negative effects of climate change, the adoption of positive behavioral changes, or actions to reduce greenhouse gas emissions and energy consumption. The pessimistic community tends to see more challenges than opportunities with regard to climate change. Voicemails in this community include scenarios in which society did not act to reduce the impacts of climate change or prepare for it adequately, leading to more stress and suffering for individuals.

Examination of player-created voicemails uncovered patterns in the perception of specific climate risks among participants. Communicators can use these classifications and findings of optimistic and pessimistic perceptions to improve framing of climate change issues and response options. The following are examples we found, along with recommendations for more effective, solutions-oriented communication. The Adaptation CCR code was used in over 40% of voicemails and found to be part of the pessimistic community of codes. Mitigation occurred in about 15% and was in the optimistic community. Mitigation was found to be slightly more complex (co-occurring with more codes) than Adaptation, although when compared to all codes both had some of the highest complexity scores (Figure 1). Optimistic views of mitigation sometimes indicated that players envision a future where the production of greenhouse gasses was reduced, a measure reducing the severity of negative climate impacts. The premise of adapting to climate change leans pessimistic because it implies the inevitability of negative impacts and a need to adapt to them. This corresponds to our finding that the Adaptation code is more closely associated with the pessimistic community of voicemails. If these analyses of Adaptation and Mitigation are indicative of public perceptions more generally, communicators can respond by presenting solutions to pessimistic perceptions, or in the case of adaptation, frame it as a solution itself.

Of interest to communicators is the appearance of some content codes such as Food, Water, and Health with predominantly negative sentiments, which may indicate a lack of perceived solutions for these issues. Pessimism about food and water resources in the context of climate change may reflect a general public expectation of resource scarcity. On the other hand, Energy and Technology were more commonly associated with positive sentiments and optimism, much like the so-called technological optimist, described by Costanza (2000) as one with a worldview “in which technological progress is assumed to be able to solve all current and future social problems.” We also found that misconceptions persist, such as crediting climate change for unrelated geophysical threats and an accelerated expected timeline for some impacts.

The more frequently occurring codes such as Technology and Weather are less complex generally, meaning they did not overlap with other codes as frequently as some of the other codes did. These findings could be indicative of either a shallow understanding of how dominant themes relate to human systems or of the more straightforward nature of these issues, thus making them easily perceived by the public. The four most complex themes, Water, Energy, Conflict and Food all have a lower prevalence, occurring in less than 20% of voicemails individually. These topics may be discussed less often because the risk is more complex and therefore harder to understand and articulate.

In addition to Technology being generally a less complex code, it was found to be more closely associated with the optimistic community of voicemails that generally perceived less climate change risk or more solutions. These findings could indicate that participants believe technology will be the silver bullet, mitigating or offsetting many of the negative impacts. The coupling with a lower complexity score may indicate an unclear understanding of exactly how technology will play a part in solutions-making. While it is true that technological innovation is a fundamental part of addressing climate change, there are challenges in developing and implementing these technologies. With these perspectives in mind, communicators could add value to the public’s understanding of technology and climate change by specifying exactly what activities could benefit from them, what time frames for technological development are in comparison with the time frame of the risk, and identifying the political and economic barriers to implementation of technologies.

Weather was another salient topic for FutureCoast participants in a climate-changed future, but less complex than most other content codes (except for Technology and Health). Weather was found to be tied to the pessimistic community of voicemails and frequently appeared in content related to catastrophic, unpredictable extreme weather events. Some voicemails indicated an understanding that a risk of climate change is shifting weather patterns and/or the intensification of some weather events, a perception that has been documented in public surveys previously (Hamilton & Lemcke-Stampone, 2014). In general, voicemails that demonstrated an understanding of how changes in weather will disrupt human systems included a single extreme event. For example, flash flooding was associated with infrastructure and property damages, and drought with food and water availability. These associations indicate that communicators have the opportunity to educate the public on how more subtle or prolonged changes in weather patterns, such as droughts, require responsible management (Bales et al., 2015). This type of public education can empower citizens to engage in local planning initiatives, as well as make solutions to anticipated weather changes clearer and less difficult to comprehend.

Food and Water were both frequently occurring codes that also co-occurred with each other (in 15 voicemails) and are associated with the pessimistic community of voicemails. Often these topics occurred in the context of resource scarcity. In some voicemails food or water availability challenges were presented with solutions. For example, one voicemail described substituting traditional farm-grown meat with lab-grown meat, envisioned as a less carbon and water-intensive alternative. Another voicemail described the opportunity for their community to build a desalination plant that produced enough water to supply the community and create revenue from the sale of excess water. Yet solutions-oriented content was not the dominant theme for the Food and Water codes. The majority of voicemails only presented risks and expressed negative sentiments. There is an opportunity for communicators and educators to relay positive information around resource use that could impact behavior in the present. For example, eating a plant-based diet tends to use less energy and water, and has the positive side effect of being healthier for the individual.

In sum, we find that the public has a complex, structured understanding of climate change, with certain emotions associated with specific content themes. Though we recognize that they represent only a sample of the general population, we suggest that the overall length and complexity of the voicemails – characteristics shared among the participants – show engagement and concern about the climate issue. We suggest as well that the association of certain climate change response codes and content codes with either optimism or pessimism could offer usable insights to communicators.

Limitations

We recognize that our research design, sampling strategy, and data analysis place limits on the scope of our discussion and interpretation. Specifically, we address the effects of gameplay on the content players produced, the representativeness of our participants when compared to random samples, and methods by which we analyzed codes and developed communities.

This experiment was not a randomized control trial; we analyzed input into a serious game from a self-selected population. The game-based format in which this content was gathered could influence the responses in ways that would make them differ from non-game research formats, such as writing an essay, or speaking with an interviewer. The content of the voicemails may not be representative of what participants truly think could happen in the future; the effects of acting out voicemail in a game setting could cause an exaggeration of narratives. For example, perhaps more negative voicemails with extreme climate change effects are more enjoyable or easier to make than positive voicemails about solutions. Likewise, misconceptions identified via voicemail content may have been exaggerated or generated as part of the playfulness of the game. The voicemail content could have also been affected by other aspects of the activity. Participants who listened to other voicemails before leaving their own could have been influenced by them and framed their voicemail similarly or used similar themes. That said, we believe we can learn from the topics the players chose and how they represented them.

Unlike traditional public understanding of science surveys, players were not a random sample of the public, but were instead self-selected. This is an inherent component of the game and could introduce a sample bias. We do consider self-selection, or voluntary participation as essential to the game format as it encourages wholehearted (i.e. voluntary and committed) participation. However, it does mean that the participants chose to play FutureCoast knowing that it was about climate change and they were willing to engage on this topic. Furthermore, people participated because they had something that they wanted to say: this could be one reason why their voicemails were such rich narratives demonstrating complex thinking.

Because no demographic information was collected and there was no way to contact the callers, it is not possible to glean anything regarding what they learned from the experience of creating and leaving the voicemails. That said, FutureCoast was a 2014 Climate CoLab Judges’ Choice Winner in the category “Shifting Behavior for a Changing Climate,” indicating that the judges considered the game to have value for behavior change. Also, van Beek et al. (2022) found that the storytelling part of a future-oriented role-playing game, where participants imagined living in a future year (2118), was effective in making tipping points more real.

There are other ways these data could be analyzed. We grouped codes that did not meet the intercoder agreement threshold into a category of “other.” These could have been handled differently. There is the potential that relationships between codes that were included and those that were excluded were removed from our analysis. Additional research on climate games could include methods such as A/B testing, and follow-up surveys of participants.

We recognize as well that our data from the previous decade does not perfectly represent the perceptions and concerns of the American population at present. However, we note that American beliefs and attitudes toward climate change, though reflecting growing awareness, shift slowly. The largest public opinion survey on this topic, conducted regularly since 2008, shows that increases and declines of most key variables, such as those who think that global warming is happening, have changed by less than 10% since the research was conducted (Leiserowitz et al., 2023). In addition, the date of the sample does not detract from the finding that playful approaches can elicit information on public thinking and associations that are useful for developing communication strategies.

Conclusions

A recent review of climate change communication advises practitioners, “engage, don’t preach” (Creutzig & Kapmeier, 2020), and many have noted the power of storytelling to engage. As the University of Bristol’s The Uncertainty Handbook: A Practical Guide for Climate Change Communicators states, “Tell a human story, not a scientific one” and “Communicate through images and stories” (Corner et al., 2015, pp. 12, 16). In a similar vein, the Frameworks Institute’s How to Talk About Climate Change and Oceans notes “the greatest communication challenge for climate scientists and other translators is not a war between cognition and emotion, but rather the lack of consistent and complete storytelling” (Bales et al., 2015, p. 2). Storytelling is key for compelling climate communication. In recent years, researchers and climate policy organizations have adopted storylines as a means of representing complex, uncertain projections. For example, the IPCC increasingly uses storylines to provide overall structure to their reports (Abram et al., 2019; Sillmann et al., 2021). We argue that stories from all sides of climate change discussions (both expert and non-expert) should be heard to increase public engagement and awareness as well as to improve understanding of public perspectives (Radchuk et al., 2017). The FutureCoast storyspace created a place for players to engage personally with climate change and make sense of a wide variety of interconnected impacts within a community of other non-experts (Jack et al., 2020).

Analysis of FutureCoast story content supports the claims of researchers who state that games allow players to connect to personal goals, affiliate with connect to a community, and create a meaningful story, all of which heightens engagement (Groh, 2012; Katsaliaki & Mustafee, 2015). The playful nature of FutureCoast gives players a space to enter an alternate reality of their creation based on their own impressions and start a peer-to-peer discourse about climate change issues concerning them. The availability of other crowdsourced narratives serves to pique players’ interests in a range of climate change issues, and encourage them to brainstorm solutions. The visceral stories they create and listen to are memorable and encourage prolonged interest in, and possibly action on, climate issues.

Communicators possessing a contextual understanding of public perceptions of climate change, such as attitudes and the perceived workings of systems and impacts, have the potential to help develop more informed and effective climate change communication strategies. Using novel approaches such as games to understand perceptions can elicit information from people who would otherwise not engage. Part of the innovation of FutureCoast is the ability to analyze rich, player-created narratives that uncover complex thinking about a climate-changed future. Activities like FutureCoast, and the subsequent analysis of the data they produce, reveal that the public recognize a wide variety of climate change impacts and responses; participants show engagement with climate change through the complexity of their responses and their emotional arousal – indicating opportunities for communicators. Responses also reveal where there are deficiencies in public understanding beyond what can be obtained through the traditional question-answer format of most polls and surveys, such as often not connecting risk with action. At the same time, responses reveal where the public identify and voice emerging issues earlier than experts. For example, conflict and migration themes emerged only recently, between 2014 and 2018, in the National Climate Assessments, despite extensive discussion of the topic in the media, in part because expert assessments undergo extensive reviews which can limit or delay the coverage of sensitive new topics (Piguet, 2022).

Identifying optimistic and pessimistic trends around climate issues gives communicators the opportunity to re-frame negative climate perceptions toward solutions, thus empowering their audiences with information that can elicit climate action. This research points toward the potential for using game responses to examine participant perceptions of other future threats, whether in the climate realm or in other areas, such as emerging diseases, international relations, and financial disruptions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

We acknowledge support from the National Science Foundation [award number 1043271]: CCEP-I Polar Learning and Responding: POLAR Climate Partnership [award number 1239783]: CCEP-II Polar Learning and Responding: POLAR Climate Change Education Partnership.

Appendices

Appendix 1. FutureCoast description

About FutureCoast. Archive: https://web.archive.org/web/20141006135550/http://futurecoast.org/about/.

See also: Wayback Machine (archive.org) captured the site in 2016: https://web.archive.org/web/20161111190732/http://futurecoast.org/.

FutureCoast (2014) was created by game and experience designer Ken Eklund as part of the Polar Learning and Responding: PoLAR Climate Change Education Partnership (www.thepolarhub.org), FutureCoast is an interactive online storytelling game that encourages users to explore a future reality by leaving voicemails, that, according to the gameplay, have leaked back from the future to listeners in the present day. Online media components included two FutureCoast websites, one in active gameplay with voicemails, and a second informational site that gave the context of the FutureCoast game, as well as resources on climate change, player-created blogs, and social media feeds. Several characters were created by the game designers to heighten player experience. Overall, approximately 20,000 people, primarily in the United States, had direct interaction with the game during the spring of 2014 via the game websites, voicemails, social media, “chronofalls,” and facilitated experiences.

The voicemails analyzed in this study were gathered as follows. An open call was made via social media and press coverage for anyone to call into the toll-free FutureCoast hotline and leave a voicemail from the future. In the fictional frame of the game, software of the future sprang a “space-time leak” in its voicemail system and audio artifacts manifested as physical objects called “chronofacts” (as pictured in Figure A1) during events called “chronofalls.” In real-life, players engage through social media to find geocached chronofacts staged in major cities in the U.S. and Europe that could lead to new voicemails. This approach was designed to draw players into the story and out into the real world to learn about local places at risk of change. Online, the growing library of player-created voicemails is supplemented by blogs and in-character videos.

All voicemails and supplemental media were available for use online during its active period; select audio is now archived at SoundCloud (https://soundcloud.com/futurecoast).

Figure A1. Image of a chronofact, a real-life object that enhanced authenticity of gameplay.

This example shows the complexity of information that can be represented in one voicemail:

Look, I just bought one of your hover homes and I'm mad as hell. You told me that I would not have to go up land anymore and search for higher ground anymore, when it flooded but the engine on this damn thing is on the fritz and I can't fix it myself. There's supposed to be a ten-year warranty, and I've got two feet, two feet of water inside my hover home. So you need to get your butts out here right now, find out what's wrong, and reimburse me for water damage. This is ludicrous, this is ludicrous. I don't understand where this is covered in your hover home contract. I didn't sign anything that said there's gonna be water damage. You gotta take care of this right now. I'm done. I'm done. ([Chronofact ID 61376-59860032] SoundCloud: https://soundcloud.com/futurecoast2/an-unhappy-customer-hover-home)

Included in this voicemail is a suite of assumptions: (1) the homeowner has repeatedly had to leave their home due to flooding, (2) it is worth it to homeowners to invest in buying hover homes, and (3) hover homes are available for purchase indicating that businesses have responded to customer needs for technological solutions and that they are confident enough in their solution that they provide a 10-year warranty.

This voicemail is also interesting because it was created in response to an earlier one (a sales pitch/robocall for the Hover Home company) made by another creator. It is an example of how listening to voicemails can inspire voicemails which often either (a) continue the same future, as in this example, or (b) give voice to an alternate or contrasting future.

Appendix 2: Voicemail coding

Coding categories, code descriptions and prevalence of each voicemail code. Prevalence is a proportion of the frequency each code was found within its code category (*Includes content or sentiment codes that were originally coded for but did not reach the threshold of intercoder agreement. **Indicates climate change response (CCR) codes that did not have sufficient intercoder reliability to carry out full statistical analysis in other parts of the study but were included here to demonstrate as a complementary pair to other CCR codes.)

Table

Code Category	Code	Description of Code	Prevalence Score
Content	Technology	Technological innovations or mainstream use of current technology; these can include products, adaptive solutions, and geoengineering.	0.33
Content	Weather	Extreme heat or cold, extreme weather events (e.g. flooding, drought, cyclones, etc.), weather pattern shifts, seasonal change; does not include acid rain.	0.31
Content	Governance/Policy	Municipal/state/national government action, policy (e.g. carbon accounting policies); does not include informal community groups/actions.	0.22
Content	Food	Agriculture, fishing, and general mentions of shortage/rationing.	0.20
Content	Energy	Renewable energy or fossil fuel infrastructure, policy, technology, cost, availability, etc.	0.16
Content	Health	Biotechnology, genetic modification, adverse health impacts (e.g. respiratory distress, heat stroke, etc.), healthcare costs.	0.14
Content	Other*	Economic issues, infrastructure, industry & business, social issues, ecosystems & landscapes.	0.14
Content	Water	Water shortages and rationing, water disputes, saltwater intrusion.	0.13
Content	Conflict	Rebellions & protests, war.	0.07
Content	Migration	Human migration, both large scale or individual.	0.07
Climate Change Response	Challenge**	Negative effects of climate change (e.g. resource scarcity, energy & transportation interruptions, conflict, disasters).	0.60
Climate Change Response	Adaptation**	Adaptive responses to climate change – any change in behavior/lifestyle due to environmental factors (e.g. adaptive technology, adaptive policy, behavior change, migration, resilient infrastructure).	0.42
Climate Change Response	Opportunity	Positive outcomes due to climate change (e.g. agriculture and industry expansion, increase in jobs).	0.22
Climate Change Response	Mitigation	Technological or behavioral change that reduces emissions (e.g. sustainable transportation, carbon counting, lowering energy consumption, renewable energy).	0.16
Sentiment	Happy/ Excited	Happy, Pleased, Content, Excited, Enthused, Hopeful	0.31
Sentiment	Neutral/ Robotic	Neutral, Emotionless, Robotic, Calm, Relaxed, Peaceful	0.26
Sentiment	Scared/ Panicked	Scared, Concerned, Worried, Distressed, Panicked, Shocked	0.24
Sentiment	Sad/ Defeated	Defeated, Pessimistic, Resigned, Sad, Disappointed, Regretful	0.19
Sentiment	Angry/ Annoyed	Angry, Irritated, Annoyed, Frustrated	0.14
Sentiment	Indeterminate*	Sentiment could not be determined.	0.02

Coding notes for sentiments

We characterize the affect of the caller to map positive and negative feelings toward each climate change issue encountered in the voicemail content. In order to obtain a more complete understanding of how players perceived climate risks, we ascribed sentiments to each voicemail by assessing the valence (positivity, negativity or neutrality) toward the content and the arousal (activation, deactivation, neutrality) of the players voice. These sentiments (i.e. happy, sad, angry, etc.) were then clustered with like sentiments into groups. Sentiments such as happy, pleased, content, excited and hopeful were all characterized with a positive valence and activated arousal. Anger, frustration, irritation and annoyance all exhibit a negative valence and activated arousal. Sentiments with a negative valence but deactivated arousal included feelings of defeat, sadness, disappointment and regret. Feelings of fear, concern, distress, panic and shock were also characterized by a negative valence and slightly activated arousal. Voicemails missing a pronounced sentiment fell into a group characterized by a neutral valence and arousal; these could be described as robotic, emotionless, or calm sentiments. There was also a small proportion of voicemails that had no or indecipherable dialogue and thus were coded as “Indeterminate” sentiment.

Coding notes for communities

Using a modularity maximization algorithm we detected communities, or subgroups, within the network of voicemail codes (content codes, CCR codes and sentiment groupings). A principal components analysis was conducted to determine the optimal number of groups to use in the analysis. This algorithm splits the codes into two communities. We then determined that the division is primarily driven by greater association with positive or negative sentiments, and therefore referred to them as the optimistic and pessimistic communities.

Appendix 3. Voicemails cited

“Stay Home If You Have Water” – 2020 – Chronofact ID 168453-87112905 Posted Feb. 2, 2014 by anonymous https://soundcloud.com/futurecoast2/stay-home-if-you-have-water.

“The Best Workout” – 2020 – Chronofact ID 23585-77019284 Posted Feb. 2, 2014 by anonymous https://soundcloud.com/futurecoast/2020-the-best-workout.

“City Changes (San Francisco)” – 2020 – Chronofact ID 65599-70920792 Posted Apr. 22, 2014 by anonymous https://soundcloud.com/futurecoast2/city-changes-san-francisco.

“Like A Horror Movie” – 2020 – Chronofact ID 94217-05124361 Posted Feb. 2, 2014 by anonymous https://soundcloud.com/futurecoast/2017-the-lake-like-a-horror.

“Tectonic Plate Initiative” – 2058 – Chronofact ID 9075088391460 Posted Apr. 9, 2014 by anonymous https://soundcloud.com/futurecoast2/tectonic-plate-initiative.

“All Moved in (Reno)” – 2062 – Chronofact ID 85471-07987160 Posted Apr. 19, 2014 by anonymous https://soundcloud.com/futurecoast2/all-moved-in-reno.

“An Unhappy Customer (Hover Home)” – 2065 – Chronofact ID 61376-59860032 Posted Feb. 20, 2014 by anonymous https://soundcloud.com/futurecoast2/an-unhappy-customer-hover-home.