iScience
Article
Conspiracy spillovers and geoengineering
Ramit Debnath,
David M. Reiner,
Benjamin K.
Sovacool, Finn
Mu
̈
ller-Hansen,
Tim Repke, R.
Michael Alvarez,
Shaun D.
Fitzgerald
rd545@cam.ac.uk
Highlights
Social media users react to
new climate technologies
like solar geoengineering
(SG)
Deep learning and natural
language processing
measured online
discourse’s toxicity
Conspiracy theories like
chemtrails affect online
public negativity related
to SG
SG conspiracies use UK,
US, India, and Sweden
geopolitics
Debnath et al., iScience
26
,
106166
March 17, 2023
ª
2023 The
Author(s).
https://doi.org/10.1016/
j.isci.2023.106166
ll
OPENACCESS
iScience
Article
Conspiracy spillovers and geoengineering
Ramit Debnath,
1
,
2
,
3
,
4
,
9
,
*
David M. Reiner,
4
Benjamin K. Sovacool,
5
,
6
,
7
Finn Mu
̈
ller-Hansen,
8
Tim Repke,
8
R. Michael Alvarez,
2
and Shaun D. Fitzgerald
3
SUMMARY
Geoengineering techniques such as solar radiation management (SRM) could
be part of a future technology portfolio to limit global temperature change. How-
ever, there is public opposition to research and deployment of SRM technologies.
We use 814,924 English-language tweets containing #geoengineering globally
over 13 years (2009–2021) to explore public emotions, perceptions, and attitudes
toward SRM using natural language processing, deep learning, and network anal-
ysis. We find that specific conspiracy theories influence public reactions toward
geoengineering, especially regarding ‘‘chemtrails’’ (whereby airplanes allegedly
spray poison or modify weather through contrails). Furthermore, conspiracies
tend to spillover, shaping regional debates in the UK, USA, India, and Sweden
and connecting with broader political considerations. We also find that positive
emotions rise on both the global and country scales following events related to
SRM governance, and negative and neutral emotions increase following SRM
projects and announcements of experiments. Finally, we also find that online
toxicity shapes the breadth of spillover effects, further influencing anti-SRM
views.
INTRODUCTION
Asthecallsforclimateactionintensify,
1
,
2
climate engineering technologies, in particular solar radiation
management (SRM), have received increasing attention, and public controversy has ensued. SRM includes
technologies such as space-based shields, stratospheric aerosols, cirrus cloud thinning, marine cloud
brightening, and increasing surface albedo.
3–5
While the broader conception of geoengineering may
also include greenhouse gas removal options (such as large-scale afforestation or direct air capture and
storage), most geoengineering debates focus on ‘‘solar geoengineering (SG)’’, often referred to as solar
radiation management or solar radiation modification (SRM).
There is opposition to both SG research and SG deployment. Opposition to SG research arises from not just
concerns about its potential methods (which are still hypothetical), and potential harms,
6
but the
uncertainties and risks associated with the path from research to problems after deployment, which are
politically expensive.
4
,
6
Outdoor SG experiments have already resu
lted in sustained opposition from civil
society and indigenous groups.
7
Concerns have reached a level where some have argued for a global treaty
prohibiting research, deployment, and use.
8
Calls to block SG research highlight the need to explore what
specific issues different groups are raising about SG and the mechanism(s) by which these concerns are
disseminated.
Concerns include the potential weaponization of weather control, attribution of weather-related disasters
to an SG program, deepening of conflict from perceived illegitimacy of its deployment across sensitive
geopolitical areas,
4
,
9–12
as well as temptations for SG to be used to tailor the climate for a single country’s
benefit,
10
,
13
including redistribution of health risks and income inequality.
14
,
15
Therefore, many experts call
for more research to better understand SG.
3
Others have urged social scientists to help inform the design
of future research.
16
For example, a 2021 US National Academies o
f Sciences, Engineer
ing, and Medicine
(NASEM) report
3
emphasized the need to better understand the dichotomy of uncertainties and risks asso-
ciated with SG research. The report concluded that SG research should include public programs for
accountability and called for more robust research governance, including scale limits on field experiments,
periodic program reassessment, and broader international cooperation.
3
,
6
,
17
1
Cambridge Zero, University
of Cambridge, Cambridge
CB3 0HE, UK
2
Division of Humanities and
Social Science, California
Institute of Technology,
Pasadena, CA 91125, USA
3
Centre for Climate Repair,
Department of Engineering,
University of Cambridge,
Cambridge CB2 1PZ, UK
4
Energy Policy Research
Group, Judge Business
School, University of
Cambridge, Cambridge CB2
1AG, UK
5
University of Sussex Business
School, Brighton BN1 9SN,
UK
6
Institute for Global
Sustainability, Boston
University, Boston, MA 02215,
USA
7
Department of Business
Development and
Technology, Aarhus
University, 7400 Herning,
Denmark
8
Mercator Research Institute
on Global Commons and
Climate Change, 10829
Berlin, Germany
9
Lead contact
*Correspondence:
rd545@cam.ac.uk
https://doi.org/10.1016/j.isci.
2023.106166
iScience
26
, 106166, March 17, 2023
ª
2023 The Author(s).
This is an open access article under the CC BY license (
http://creativecommons.org/licenses/by/4.0/
).
1
ll
OPENACCESS
Other concerns focus on the politics and ethics of SG. Efforts to integrate Global South actors into SRM
initiatives remain limited.
18
,
19
Also, the development of climate engi
neering options often fails to tap
into truly inclusive or responsible forms of innovation or technology design.
20–22
The concerns and criticisms of SG discussed above largely reflect expert opinion rather than public percep-
tions and attitudes toward SG. This paper seeks to help
address this knowledge gap by analyzing historical
and social media interactions concerning various global SG events and investigating potential correlations
with other controversial topics, such as conspiracy theories.
Recent studies have shown that conspiracy theories, propelled by fake news and polarization (i.e. disinfor-
mation campaigns), can act as a contagion of inten
se negative emotions in digital social networks.
23–25
Disinformation campaigns can have a pernicious and lo
ng-lasting negative impact on beliefs about climate
change and on public acceptance of more sustainable lifestyles. For example, misinformation campaigns
on social media in Brazil condoned and then perpetuated renewed deforestation of the Amazon rainfor-
est.
26
In Europe, false conspiracy theories were spread (and believed), connecting plans to build a new
gas pipeline with the mafia’s involvement
and the spread of infectious diseases.
27
One patently false con-
spiracy theory dating back to the mid-1990s is ‘‘chemtrail
s’’ (explored in greater depth later), which alleges
condensation trails (contrails) from aircraft are intentionally seeded with chemical or biological compounds
for various nefarious purposes such as population control, military testing or, most relevant to our case,
weather and climate modification including SG. Up to 40% of Americans believe this theory to be ‘‘some-
what true’’, which has influenced social attitudes about climate policy, and geoengineering.
25
On climate change, there is growing evidence of a ‘‘spillover’’ effect that leverages local conflict/contro-
versies to cascade controversies in order to shift a policy agenda deliberately
28–30
, similar to conventional
agenda-setting. In this paper, we expand on the conceptual application of the ‘‘spillover effect’’ to evaluate
conspiracy theories across geopolitical boundaries
and their agenda-setting impact on public emotions
and online toxicity perceptions of SG research.
This paper is distinct from previous studies critiquing SG governance challenges and their associated con-
troversies related to climate action, on which there is already a rich literature,
3
,
4
,
6
,
7
,
16
,
25
,
31–33
including social
media mining-based SG conspiracy analysis.
25
One apparent gap in Tingley and Wagner’s
25
study is that
the authors had a narrow focus on the ‘‘chemtrails’’ conspiracy theory in their searches for data collection.
Therefore, they could not measure any spillover effect from other conspiracy theories in geoengineering
debates.
This study uses digital data from social media to capture cross-sectional variation in public emotions
following major SG announcements o
n a global and country scale (UK, USA, Sweden, and India) using geo-
specific hashtags. At the same time, we investigat
e whether social media is used to propagate and
influence conspiracy theories on SG, thereby deepening our understanding of potential spillovers of
controversial research topics into the public domain. Themes and ideas condensed in hashtags spread be-
tween user groups and regions in ways that are influenced by other themes and beliefs attached to the
same tweets. Hence, as a result of ‘‘conspiracy spillovers’’, we find that the online communication of SG
associated with conspiracy theories spreads by tak
ing advantage of the geopolitical controversies.
Our scope is limited to using social media interaction
data as a proxy for cross-sectional public opinion on
geoengineering. It may be non-representative of the broader public opinion. However, recent studies have
shown that it is legitimate to say that social media is a breeding ground for conspiracy theories and hence
deserving of attention.
34
Specifically, Twitter-led climate communication is increasingly seen to engage its
user toward climate action
35–37
using tools like hashtags.
Users on Twitter employ hashtags to indicate the general theme or topic they wish to associate with their
tweets. This allows one to investigate how information propagates throughout the platform. As hashtags
show an attention-seeking behavior, it makes them a viable candidate
38
for studying the public reaction to
complex governance issues of SG and its interface with conspiracy theories. We use natural language
processing, deep learning, and network analysis (see
STAR Methods
) to explore users’ reactions
in 814,924 English-language tweets containing #geo
engineering across 13 years (January 2009–November
2021).
ll
OPENACCESS
2
iScience
26
, 106166, March 17, 2023
iScienc
e
Article
RESULTS
Public emotion on social media shifts following SG announcements
This section presents three longitudinal results based on the natural language processing (NLP) and
lexicon-based emotional analysis of the #geoengineering Twitter dataset. Here, we establish the dynamic
relationship between the user reactions to major solar geoengineering (SG) projects and governance
events in the public domain.
The first finding is a strong relationship between Twitter activity concerning major SG projects and climate
action or negotiation events (see
Figure 1
A). For example, the number of #geoengineering tweets
increased around the launch of the EXPECT project in Norway and when the IPCC actively considered inva-
sive measures to halt climate change in May 2014. A significant rise in Twitter activity was also observed
following reporting about the SCoPEx (Stratospheric Controlled Perturbation Experiment) project, led
by David Keith and colleagues at Harvard University and supported by a consortium of funders that also
included Bill Gates, a long-standing favorite targ
et of conspiracy theorists. The project was announced
in December 2015, and SG protests were held at COP21 in Paris (
Figure 1
A). Twitter volume on #geoengin-
eering peaked when the SCoPeX proje
ct was officially launched in Apri
l 2017. This was the most significant
surge in Twitter activity on #geoengineering, with close to 8,000 interactions per day (an
300% increase
relative to February 2017 levels) in the peak period (June 2017,
Figure 1
A). Twitter activity returned to pre-
SCoPEx levels by August 2017, followed by new but smal
ler peaks in the chemtrails conspiracy-related on-
line interactions. We also see from
Figure 1
that ‘‘opchemtrails’’ topics disappeared toward the end of
December 2017.
The trendline in
Figure 1
A can be divided into two phases of SG project development. Pre-2016,
there were more governance-related actions involving various multilateral organizations. However,
post-2016, significant announcements were made regarding practical implementation pilot projects
and experiments.
Secondly, the moving-average emotion analysis in
Figure 1
B shows that positive emotions (i.e. opti-
mism, joy, and trust) are more prominent, recurrent, and comprise a larger share of the #geoengineer-
ing social media conversation. The share of negative (i.e., disgust, sadness, fear, and anger) and
neutral (i.e., anticipation and surprise) emotions increase over time, with distinct peaks between
June 2017 and April 2018 (
Figure 1
B). We also see that emotions are more volatile in the 2009–
2012 period, with short-lived spikes in the share of negative (rising from 15% to 30%) and neutral
(24.5%–40%) emotions following governance-related events such as the publication of the Royal Soci-
ety report and the Oxford Principles both in 2009 (
Figure 1
B). A similar increase in emotion volatility
was observed in late 2010 (positive emotions increased from 40% to 55%), which coincided with the
launch of the SPICE project and calls for a moratorium on SG technologies at the UN Convention
on Biodiversity citing gaps in scientific knowledge and risk management.
39
The share of negative
and neutral emotions also spiked during the cancellation of the field-trial aspects of the SPICE project
in January 2012 (see
Figure 1
B).
Moving averages in
Figure 1
B (2018–2021) show public emotions possibly modulated by conspiracy the-
ories like HAARP (an ionospheric research program largely funded by the US military) and chemtrails. Later
in
Figures 2
and
3
A, we show the broadening of conspiracy theories to include the Covid-related ‘‘antivaxx’’
movement (late 2021) through the semantic structure of t
he online conversation. In line with findings in the
literature,
25
,
40–42
this shows the possibility that there are spillover effects between different conspiracy
theories. We also found increases in shares of neg
ative (from 20.4% to 28.6%) and neutral (from 24.3% to
47.7%) emotions following SG-related field experiments (
Figure 1
B). The share of neutral emotions, in
particular, grew substantially following attempted S
G field experiments. In contrast, negative emotions
rose following announcements of these projects.
The 2021 NASEM consensus report was critical for poli
cy debates, emphasizing that SG research should be
catalogedinapublicregistryandresearchonsocialdimensionsofSGshouldbeprioritized.
3
When this
report was released, which outlined the type of research that should be conducted, how it should be
governed,andwhoshouldfundit,themostspecificemotionchangewastheriseinneutralandpositive
emotions (
Figure 1
B).
ll
OPENACCESS
iScience
26
, 106166, March 17, 2023
3
iScienc
e
Article