Technical ReportPDF Available

Computational Diplomacy: Foreign Policy Communication in the Age of Algorithms and Automation

Authors:

Abstract

The structural shadow of uncertainty over diplomacy is stronger than ever. Some communicative rituals and practices of diplomacy are growing more obsolete, as modern political communication slides increasingly to short and sharp rhetoric, coupled with automation tools that bombard audiences at unprecedented levels. Diplomacy itself is hardly obsolete however, as the task of mediating and negotiating power relations is perhaps as important as it was during the Cold War. New power centers - in the form of technology companies and big data brokers - are changing the state-centric parameters of classical realism perhaps, but the inherent dynamics of power realignment still render diplomacy a crucial endeavor. To rise to the challenge however, modern diplomacy has to develop a strong computational capacity, able to adapt to the changing nature of digital communication and advances in automation.
Computational Diplomacy
H. Akın Ünver, Kadir Has University & EDAM
This paper was supported by
Cyber Governance and Digital Democracy 2017/3
November 2017
Cyber Governance and Digital Democracy 2017/3
1
Uncertainty is a foundational aspect of politics and diploma-
cy. Critical elections, armed conicts, ally/adversary behav-
ior, explicit/implicit threats are fundamentally uncertain, yet
core processes of statecraft, diplomacy and politics. That’s
why over centuries, diplomacy has grown into an art form
of managing high-risk uncertainties between nations and
institutions. Uncertainty isn’t trivial, or secondary, since it
has direct impact on policy and fortunes of nations through
costly miscalculations. Cognitive processes, misperception
and elite psychology have thus grown into central themes of
inquiry in international relations research through the Cold
War and continued to dene foreign policy research after
the fall of the Berlin Wall. War onset, results of diplomatic
negotiations or how people behave during emergencies or
crises, are all variables in a three-dimensional equation, de-
termining power relations at the global level.
Uncertainty is also a communicative process. How we under-
stand, contextualize and navigate through uncertainty de-
pends on verbal and nonverbal cues. That’s why historically,
articulation, tact and acumen grew into key qualities of good
emissaries and ambassadors, as they communicated power
relations between nations. This is also the main reason why
states always sought to narrow down their key communica-
tions into a small audience of highly qualied individuals,
specically trained in the art of uncertainty management. As
more people got involved in political processes, leaks be-
came more likely, but perhaps more importantly, sides lost
their common political language to navigate through periods
of uncertainty. As communication technology progressed,
diplomatic processes had to adapt in order to protect both
secrets and also common language. Take for example how
the invention of writing led to the development of seals of au-
thentication, printing press to mechanical cipher, telegram
to morse code and radio to modern encryption.
From this perspective, digital diplomacy sounds like an un-
natural progression in the long history of technology and
diplomatic communication. Politics online is anything but
hierarchical, unidirectional or secret. From Ministers of For-
eign Affairs to notetakers, all parties to a political or diplo-
matic process have equal access to social media platforms,
or can build a website. Quality of content determines fol-
lower count and diplomatic language is replaced by a new
type of tech language, where emojis, directness and snark
have a higher value than elaborate, long explanations. ‘It is
rather complicated’ is the motto of diplomacy, yet the craft
itself is moving into the domain of 140-word explanations
and emojis. In November 2016, Guardian reported that
Whatsapp was becoming the primary medium of commu-
nication among diplomatic circles, even during some of the
key voting and negotiating processes in the UN and the EU
headquarters.
2
The spread of Whatsapp among diplomatic
circles was alarming as the UK Foreign Ofce reported that
its diplomats were using the platform instead of the specially
designed encrypted messaging application. User-friendly
platforms always win over clunky, elaborate interfaces and
design is always often popular than security; especially for
Julian Borger Jennifer Rankin in Brussels and Kate Lyons, “The Rise and Rise of International Diplomacy by WhatsApp,” The Guardian, November 4, 2016, sec. Technology,
https://www.theguardian.com/technology/2016/nov/04/why-do-diplomats-use-this-alien-whatsapp-emoji-for-vladimir-putin.
Airaksinen, T. ‘Against all the odds: Machiavelli on Fortune in Politics’ in Donskis, L. (Ed.). (2011). Niccolò Machiavelli: history, power, and virtue (Vol. 226). Rodopi.
2
1
If one is to be able to handle, successfully, such drastic uncertainty which The (Machiavellian) Prince
meets in those political situations where conicts are a life and death issue and in which no rules apply,
one needs to be exible. One needs to be a fox and foxes ex the language.1
Computational Diplomacy
Foreign Policy Communication in the Age of Algorithms and Automation
H. Akın Ünver
Cyber Governance and Digital Democracy 2017/3
2
younger diplomats. Furthermore, as evidenced by a succes-
sion of high-prole leaks by Edward Snowden, Julian As-
sange, Chelsea Manning and many others, even the best-
kept state secrets can be leaked. In the past, such leaks
would involve intelligence agencies, or media companies.
Now all leaks are public and governed by the same atten-
tion economy metrics as advertisements. Furthermore, so-
cial media space is increasingly more vulnerable to ‘digital
spoilers’, such as trolls and bots, that amplify messages (in-
cluding leaks) by the millions.
States now have to craft more ingenuous policy positions
and elaborate national strategies as some of the deepest
secrets of their nation are distributed online, shared in the
millions, all while their governments are paralyzed without a
clear agenda on how to tackle such processes. Before diplo-
macy could adapt to ‘digital’, it is now faced with problems
from more advanced computational aspects of technology.
From this perspective, MFAs have to adapt to three layers
of computational challenges. The rst of these is size (vol-
ume). Digital interconnectedness has increased the number
of actors and parties in key political processes, reducing the
level of inuence in diplomacy to the random online citizen.
Unprecedented volumes of information and opinion ows in
digital space, weakening states’ control over their nation’s
image, policy and branding. Government-led PR campaigns
tend to be more boring, less imaginative and ‘gray’, inevi-
tably losing against more creative and vibrant offerings of
non-state actors. Second, digital communication is instanta-
neous. Communicative processes of high-priority and high-
risk political issues rapidly proliferate online, far exceeding
states’ ability to respond through traditional modes and poli-
cy processes. States’ adaptation to digital platforms not only
requires an upgrade in tools, but also in the way of thinking
and responding to emergencies. Finally, not all information
that travels online is true. Often, in the heat of the moment,
misleading information or doctored images spread rapidly,
bringing in the necessity for states to verify and correct mis-
takes to prevent escalation. Recently however, states them-
selves have begun irting with these misleading content
types to gain the upper hand against their diplomatic rivals,
which is blurring the lines between states and non-state ac-
tors in legitimate digital political communication.
Digital diplomacy (or Internet, cyber, e-, Diplomacy)
emerged as a state reaction to its growing irrelevance in
digital space. process. States were forced to discover and
seize the potential of these capabilities, only after a succes-
sion of three high-prole international crises, that demon-
strated the extent to which they were left behind in the digital
communication revolution. The emergence of ‘digital diplo-
macy’ as a global concept coincides with the onset of the
2010 Arab Spring and Occupy movements that emerged
in 2011. When fuelled with enough grievances and social
organizational power, digital technologies allowed masses
to mobilize against, and threaten state power, sometimes
fundamentally changing power relations in their country,
such as in the cases of government toppling in Egypt and
Tunisia. These technologies were also challenging state nar-
ratives and framing of social and political events, bypassing
traditional modes of state propaganda, as well as existing
governmental controls on mass media outlets. As a result,
states began to strategize to establish digital representation
and communication practices, gradually evolving into digital
diplomacy practices that we know of today.
A second evolutionary trigger of digital diplomacy was the
rise of ‘citizen journalism’ phenomenon. The emergence
of a global caste of citizen reporters that established digi-
tal and real-time news dissemination networks online, was
perhaps the greatest challenge to states’ control over infor-
mation. Marginalized in the business model of state - mass
media relationship, the neglected practice of local report-
ing received an unprecedented boost with digital technol-
ogies, allowing any individual to be able to gather news at
the street level and share it with global audiences, bypass-
Cyber Governance and Digital Democracy 2017/3
3
ing the hegemonic state-media power. Then came another
trigger: the spread of extremist content online and digital
recruitment. ‘Digital radicalization’ became popularized with
the rise of the Islamic State in Iraq and Syria (ISIS) and the
term then almost exclusively referred to jihadi extremist mes-
saging online. Although equally large networks of radicaliza-
tion and extremist messaging can be observed in European
and American digital space, states’ digital focus is still very
much concentrated on online jihadi networks. Yet, growing
range of threats from other tints of global radicalization will
inevitably awaken states to the need to think about the phe-
nomenon as a trans-cultural topic. The need to counter all
ranges of radical messages and frames online will eventu-
ally add another layer of responsibility to digital diplomacy:
formulating and disseminating alternative religious, political
and social messages and preventing the spread of radical
content online.
The pace with which communication technologies evolve,
will render the concept of ‘digital diplomacy’ obsolete, be-
fore it even became mainstream. The evolutionary impulses
of political communication are being increasingly driven by
automation and computation, and seek to overload, over-
whelm and distract collective attention on a global scale.
Websites and social media actors are no longer necessarily
‘human’ and ‘bots’ - that are operated by a single program-
mer - can mass-produce digital content at a theoretically
innite scale. Such digital content can be factually false,
misleading or anachronistic and can easily ood social
media systems during emergencies and crises. They can
impair diplomatic communication and escalate inter-state
disagreements; worse, they can bypass governments and
inuence entire populations at key political junctures. By in-
creasing the likelihood of misperception, they also render
armed escalations more likely. To that end, automation sits
at the heart of modern diplomatic evolution and requires a
new framework and strategy that goes beyond ‘digital’.
Online Diplomatic Behavior:
Who Speaks on Behalf of the State?
Digital communication technologies have fundamentally
altered the nature of diplomacy by changing the very en-
vironment diplomats function. In October 2012, during the
U.S. Presidential election debate, Barack Obama and Mitt
Romney were at loggerheads with each other over the be-
havior of American diplomatic representatives in Libya and
Egypt. The debate was whether the members of US foreign
mission in Tripoli and Cairo were right in issuing online sup-
port and tweets for the anti-government protests, without a
clear government policy. Obama had to walk a ne line be-
tween his government’s necessity to control US diplomats’
online expression and not denouncing ongoing protests
against Muammar Ghadda and Hosni Mobarak.
3
The dig-
ital revolution and social media platforms reduced the costs
of political engagement for all parties and strengthened the
communicative agency of diplomats, but also created a
new domain of political authority; one that doesn’t neces-
sarily play along with ofine state interests. Should states
restrict their diplomats’ presence online and risk irrelevance,
or should they allow individual expressions of opinion and
dismiss them as ‘not representative of the state’?
Fast forward to August 2014, when western world capitals
were hotly debating Russian involvement in Ukraine (spe-
cically, whether Russian soldiers were directly engaged
in the clashes in Donbass region). Simultaneously, Russia
was mounting one of its best-funded and highest prole dip-
lomatic defenses against these ‘allegations’, insisting that
armed men in Donbass were pro-Russian locals. Around
the same time, a Russian signal corps sergeant - Alexander
Sotkin - began posting geotagged seles from within Ukrain-
ian territory, soon joined by other Russian artillery, logistics
and combat troops.
4
These goofy and clueless social media
posts on Russian media platforms VKontakte and Odnok,
eventually proliferated in Twitter, Facebook through mass
sharing, ultimately nullifying state-sponsored propaganda
efforts of Moscow.
Brian Fung, “Digital Diplomacy: Why It’s So Tough for Embassies to Get Social Media Right,” The Atlantic, October 17, 2012, https://www.theatlantic.com/international/
archive/2012/10/digital-diplomacy-why-its-so-tough-for-embassies-to-get-social-media-right/263744/.
3
Sean Gallagher, “The Sad, Strange Saga of Russia’s ‘Sergeant Sele,’” Ars Technica, August 14, 2014, https://arstechnica.com/information-technology/2014/08/
the-sad-strange-saga-of-russias-sergeant-sele/.
4
Cyber Governance and Digital Democracy 2017/3
4
Whether a nation, its diplomats, soldiers or ministers should
formally exist in digital space is a trickier question than the
mainstream debate suggests. MFAs, embassies and diplo-
mats are online actors willingly or not, and their ofine activ-
ities are getting inuenced increasingly by issues that have
a signicant digital component. Arguments in favor of digital
engagement include the ability to shift, mold and inuence
global public opinion with much lower costs compared to
traditional forms of public diplomacy. Politicians, diplomats,
ministries and international organizations already tap into
this inuencing capacity for positive (i.e. charm offensive,
cultural engagement, awareness-building), as well as nega-
tive (heated online arguments, defensive comments, propa-
ganda) reasons. For example following the coup attempt in
Turkey in July 2016, Turkish MFA has launched one of the
best-coordinated examples of digital diplomacy. A wide net-
work of Turkish embassies, as well as members of mission,
used their Twitter accounts to disseminate information, vide-
os and photos to build awareness in foreign capitals. In the
deep uncertainty of the immediate post-coup phase, MFA’s
sentiment-neutral messaging travelled farther and shared
more across foreign media outlets, compared to the govern-
ment-led combative and antagonistic messaging, that was
mostly shared within Turkey. The country has since then be-
came an instrumental part of global political engagement,
from the Jerusalem protests to Rohingya crisis. In a previous
piece, I discussed how Turkey got involved in the Jerusalem
crisis through a combination of ofcial and unofcial digital
campaigning.
5
Yet, mere presence of large numbers of government or insti-
tutional actors online doesn’t qualify as a successful social
media or nation-branding campaign. One common mistake
is to think of digital nation-branding as a one-way street, in
which ofcial government or institutional positions are simply
uttered online, without any direct engagement with interest-
ed online parties. Such campaigns are viewed as dull, unin-
teresting and unimaginative, effectively yielding questiona-
ble multiplier effects on states’ existing image. On the other
end of the spectrum however, are the ‘social media diplo-
mats/ministers’, that favor direct engagement with online ac-
tors, but share too much, effectively making statements that
do not converge with government policy, ending up gen-
erating controversy rather than engagement. The balance
isn’t always straightforward and tends to be heavily subjec-
tive. One of the good examples of this balance has been
Israeli MFA’s culture-oriented nation-branding campaign,
which steers clear of thorny political issues and follows di-
rect engagement with questions on Twtitter, Facebook and
Youtube on Israeli cuisine, artists and daily life.
6
How much should states commit to online representation,
just like other forms of diplomacy, requires a strategy. Al-
though there is increasingly more quality research on the
topic, we are still very much in the dark over what makes a
digital campaign successful and especially how do political
and non-political media campaigns are consumed and dis-
tributed differently. This is because what constitutes as ‘suc-
cess’ in digital space, is highly subjective and context-spe-
cic. This renders political social media campaigns hard to
control and even harder to measure in terms of their impact
and success. Often, well-led digital media efforts can suffo-
cate among other well-led brand or event campaigns, gen-
erating far less engagement, or yielding unintended diplo-
matic outcomes than desired. One example was the Russian
Airstrike Watch unit under the British Foreign Ofce, which
tweeted real-time information on Russian bombing opera-
tions in Syria.
7
Challenging Russia’s narrative of targeting,
frequency and strategy in Syria, this unit eventually generat-
ed a Russian backlash against the British Government itself.
While the campaign itself was apparently successful from
a digital media point of view, diplomatically it impaired Brit-
ain’s ability to pressure and bargain against Russia in Syria.
Furthermore, some of the best-led and best-distributed dig-
ital media campaigns failed to bring about the change they
Unver, Akin. ‘What Twitter can tell us about the Jerusalem protests’ The Washington Post. 28 August 2017. https://www.washingtonpost.com/news/monkey-cage/
wp/2017/08/26/what-twitter-can-tell-us-about-the-jerusalem-protests/?utm_term=.52f561258ead
Unver, Akin. ‘What Twitter can tell us about the Jerusalem protests’ The Washington Post. 28 August 2017. https://www.washingtonpost.com/news/monkey-cage/
wp/2017/08/26/what-twitter-can-tell-us-about-the-jerusalem-protests/?utm_term=.52f561258ead
6
5
Worley, Will, “Russians Stage ‘Retaliation Protest’ Outside British Embassy,” The Independent, November 4, 2016, http://www.independent.co.uk/news/world/europe/
russia-british-embassy-moscow-retaliation-protest-demonstration-aleppo-syria-a7398441.html.
7
Cyber Governance and Digital Democracy 2017/3
5
desired. One of the best examples of a highly professional
digital media campaign was #bringbackourgirls - dedicat-
ed to 276 Nigerian students kidnapped by Boko Haram.
Although the campaign was widely shared and received
high-level of attention (including Presidential) it ended up
having virtually no real effect on the outcome of the hostage
situation. Some of the chibok girls were released, but long
after the campaign and with no visible response to the pop-
ularity of the digital effort.
8
These instances point to the necessity of a well-thought out
strategy that connects a nation’s ofine interests to its online
presence, in a way that reinforces both domains. Digital di-
plomacy doesn’t exist in a vacuum and is a highly interactive
process, where every tweet, like and share is a digital activi-
ty and are interpreted as state policy. A diplomat retweeting,
or liking a particular foreign policy article will be interpreted
by journalists as a formal government position and will be
produced into a news article in that light. Denials or clari-
cations will usually arrive much later than the incident enters
into news cycle and even then, clarications will not be as
popular as the made-up news. Especially in contested is-
sues, there will always be a challenge or a backlash against
MFA-led digital media campaigns. Not only rival embassies
or MFAs, but foreign government ofcials, celebrities and
random citizens alike will join the conversation. But these
essentially human-led accounts aren’t the main problem.
What happens when this challenge comes from thousands
of anonymous accounts, posting irrelevant or simply wrong
information by the thousands?
Computational Propaganda: Attention Economy
in International Crises
Marketing and advertising have predominantly relied on at-
tracting consumer attention as a form of currency. As ad-
vertising got digital, so did political advertising and prop-
aganda. Digital advertising is structured upon the premise
that our digital footprint can - and should - be monetized, at
increasingly lower marginal costs. The content that we share
online, along with people we follow, or personal information
we post across our social networks returns back to us in the
form of tailored advertisements. Political campaigns too, use
similar tools. Companies, politicians, ministries, celebrities
all compete for attention and produce content that has the
best likelihood of getting shared and surviving online. At-
tention has always been a scarce resource, although digital
media and communication increased its value substantially,
as the production costs of digital content became far cheap-
er than consumption costs. This was foreseen by Herbert
Simon, who has rst coined the term ‘attention economics’
9
in 1971 in reference to the overabundance of information,
with the likes of Thomas Davenport and John Beck
10
, Mi-
chael Goldhaber
11
and Georg Franck
12
expanding the term
as it relates to information overload in business and market-
ing. The term took on its more relevant form in social media
through Hubermann (et. al.) 2008 piece, which asserted that
the digital media has strengthened the agency of ordinary
people for whose attention multi-million dollar companies,
states, intelligence agencies and presidential candidates all
competed.
13
Maeve Shearlaw, “Did the #bringbackourgirls Campaign Make a Difference in Nigeria?,” The Guardian, April 14, 2015, sec. World news, https://www.theguardian.com/
world/2015/apr/14/nigeria-bringbackourgirls-campaign-one-year-on.
8
Thomas H. Davenport and John C. Beck, The Attention Economy: Understanding the New Currency of Business (Harvard Business Press, 2002).
Michael H. Goldhaber, “The Attention Economy and the Net,” First Monday 2, no. 4 (April 7, 1997), http://rstmonday.org/ojs/index.php/fm/article/view/519.
Georg Franck, “The Scientic Economy of Attention: A Novel Approach to the Collective Rationality of Science,” Scientometrics 55, no. 1 (September 1, 2002): 3–26, do-
i:10.1023/A:1016059402618.
Herbert A. Simon, “Human Nature in Politics: The Dialogue of Psychology with Political Science,” The American Political Science Review 79, no. 2 (1985): 293–304,
doi:10.2307/1956650.
10
11
12
9
Bernardo Huberman, Daniel M. Romero, and Fang Wu, “Crowdsourcing, Attention and Productivity,” Journal of Information Science 35 (October 17, 2008), doi:10.2139/
ssrn.1266996.
13
Cyber Governance and Digital Democracy 2017/3
6
This naturally changed how political messaging and prop-
aganda works. Competing over attention means producing
both higher quantities of and more striking (not necessarily
higher quality) digital content. Visuals that trigger people’s
extreme emotional response mechanisms of hate, outrage,
fear or lust eventually dominate social media metrics and
online behavior patterns, forcing the eld to evolve into more
darker corners of human nature. Automation sits at the heart
of this debate. Emotion eliciting content, produced at in-
creasingly larger quantities eventually became an arms race,
following an escalation pattern that brought about computer
programs called ‘bots’. These bots eventually took on tasks
of spamming, posting and resharing content in digital space
at a theoretically innite magnitude. Bots gradually evolved
from business and marketing related tasks to politics, her-
alding a new era of political communication: computational
propaganda.
According to Oxford’s ‘Computational Propaganda Project’,
the term refers to a vast network of automated agents, dis-
tributed across multiple media platforms in order to distract,
ood and mislead public opinion.
14
The primary currency of
computational propaganda is a ‘bot’ (short for robot), which
is a computer program that performs pre-dened automat-
ed tasks. Bots can simply utter pre-designated set of words
on Twitter, like/dislike posts, or engage in more complicat-
ed assignments with the use of machine learning, such as
learning how to respond to social media posts. Bots prolif-
erate exogenous (and often unpopular) opinions, set dig-
ital agenda, generate engagement with online campaigns
and beef up follower counts in digital space. When they are
used in political processes, they signicantly amplify fringe
and radical positions, spam information providers (such as
journalists) and ood the discussion by hijacking dedicated
hashtags. States use bots to disrupt protests as well as each
others’ digital media campaigns and non-state actors too,
use bots to distract away from state-driven political agen-
das, or simply force participants into inaction, through mass
confusion. Samuel Woolley for example, distinguishes be-
tween three types of political bots: paid ‘follower bots’ that
increase politicians’ social media follower ranks, ‘roadblock
bots’ that hijack popular hashtags to confuse, mislead and
distract political opponents, and ‘propaganda bots’ that au-
tomatically attack online speech that is deemed ‘dangerous’
or unwanted - such as government criticism. How signicant
are bots in digital communication? A University of Southern
California study predicts 48 million Twitter accounts (15%
of all users as of 2015) are bots,
15
although measuring bot
activity is an inherently difcult task, due to their rapid pro-
liferation and disappearance during key moments. Regard-
less of how many bot accounts there are, what renders bots
troublesome is the fact that they can concentrate around
a single issue, dominate and overwhelm its online debate,
then disappear and reappear as necessary. As of 2016, the
world’s largest number of bot infections (one bot for every
1139 users) was in Turkey, making up 18.5% of all bots in
Europe.
16
This renders the country in a special position with
regard to computational propaganda, bot-driven political
engagement and automated propaganda efforts. Other usu-
al suspects are Russia, Italy and Germany, although lack
of reliable measurement prevents us from judging the true
place of China.
Bots and principles of automation that operate them are
getting more advanced every day. While it was easier to
recognize a bot from its monolithic and binary responses,
machine learning advances allow bots to continuously im-
prove and evolve their discursive options. Instead of dis-
seminating previously coded word combinations, bots are
now able to adapt to the language of a political movement,
Samuel C. Woolley and Philip N. Howard, “Automation, Algorithms, and Politics| Political Communication, Computational Propaganda, and Autonomous Agents — Introdu-
ction,” International Journal of Communication 10, no. 0 (October 12, 2016): 9.
14
Abel, Robert. ‘And the country with the most bot infections is.. Turkey’. SC Media. 5 October 2016. https://www.scmagazine.com/turkey-tops-the-list-in-the-number-of-
bot-infections/article/527590/
Onur Varol et al., “Online Human-Bot Interactions: Detection, Estimation, and Characterization,” arXiv:1703.03107 [Cs], March 8, 2017, http://arxiv.org/abs/1703.03107.
16
15
Cyber Governance and Digital Democracy 2017/3
7
reading increasingly more like a human account. It is not
hard to see journalists, diplomats or politicians getting em-
broiled into a quarrel with a well-programmed bot during cri-
sis situations. There are also some earlier examples of bots
copying communicative structures and lexicon of key po-
litical gures, signicantly adding to the high-level political
confusion during emergencies. Following recent scholarly
and policy emphasis on political bots on Twitter and Face-
book, most of these bots moved to Tinder – where people
would least expect them; in an online dating app. During the
British election campaign of July 2017, a group of political
programmers have developed a Tinder chatbot that would
disseminate targeted political messaging to 18-25 year olds
in battleground constituencies.
Factual truth in computational propaganda is often irrele-
vant to its success. In digital platforms, large numbers of
accounts uttering the same political position creates a band-
wagon effect (groupthink and availability cascade). The vol-
ume of digital consensus is usually more important than the
factual truth or the quality of the argument in getting those
messages shared and this is specically what bots are de-
signed to do. Although a number of fact-checking platforms
exist globally, the time required to disseminate false infor-
mation in large volumes is always shorter than verifying it
conclusively. Even when a false message is quickly refut-
ed, such messages still linger on due to conrmation bias
- namely, by people who believe that the message is true
because it ts their political preconceptions.
All of this incurs far greater weight over existing capacities
of diplomatic communication and public engagement chan-
nels. Balance of power in computational propaganda - like
cyber war - favors the offensive side as costs of defending
against such attacks require greater resources and better
coordination. Even when the defender is successful (i.e.
corrects disinformation quickly), psychological processes
of digital information consumption still linger on. This sig-
nicantly impairs individual embassies’ ability to formulate
responses and offer a counter-narratives in the heat of a cri-
sis. Even when these efforts are led by a central MFA with
enough resources to mount a real-time information veri-
cation crusade, content that will be posted online must go
through regular channels of institutional checks and bal-
ances (bureaucracy), creating signicant lags in ofcial re-
sponses. Conversely, a faster and less controlled information
campaign has the risk of sharing content that doesn’t reect
government policy or MFA view on issues. Such content can
unintentionally criticize government policy in another matter,
sapping the campaign efforts of the MFA. Russian MFA has
currently improved its use of digital communication technol-
ogies, specically Twitter, to reconstruct its image as a dull
and boring online actor. A recent example is an exchange
between CIA and Russian Ministry of Foreign Affairs. As CIA
tweeted a job call for Russian speaking new college gradu-
ates, Russian MFA responded with a tweet: ‘We are ready to
assist with experts & recommendations’.
A second issue relates to diplomatic use of automation it-
self. Should MFAs engage with bots in an endless battle for
narrative, or should diplomatic missions create their own
botnet and ‘ght’ with other bots? Engaging with bots is a
futile task, but given increasing difculties in recognizing
well-programmed bot accounts, most people cannot tell the
difference. An honest public diplomacy and digital engage-
ment attempt can descend into an unending exchange,
whereas refraining from engagement can impair digital di-
plomacy efforts if the account is actually human. One way
around this is to respond only to veried accounts, or at
least accounts that appear to have a real name. Another is
to work with data scientists to create inuence network maps
of political processes and understand which accounts are
the most inuential and top-drivers of political debate in any
given situation. Through engaging with central gures in an
engagement network, MFAs can adjust to the changing con-
tours of a political process and attain an efcient balance
between engagement and caution. A good example of such
inuence network analysis is the work of Efe Sevin, whose
work on MFA inuence metrics in digital diplomatic networks
has won the 2017 ISA ICOMM best paper award.
17
Sevin, Efe, ‘Traditional Meets Digital: Diplomatic Processes on Social Media’. Paper presented at the International Studies Association 2017 Annual Conference. https://
isaicomm.wordpress.com/icomm-awards/best-paper-award/#2017 Winner
17
Cyber Governance and Digital Democracy 2017/3
8
The last, and perhaps the greatest, challenge computational
propaganda poses to diplomats is the fact that their own
governments use them as well. Although current evidence
points mainly to authoritarian governments as the primary
users of bots as a form of political communication, democra-
cies too, rely on bot-driven political engagement during key
domestic events like elections or protests. In fact, Bradshaw
and Howard report that ‘the earliest reports of government in-
volvement in nudging public opinion involve democracies’.
18
This prevents any particular country to assume the moral
high ground in computational propaganda, often offsetting
image-building and PR work conducted by MFAs and diplo-
mats. Given the multitude of interests in a crisis (government
seeking to discredit a protest, intelligence services spread-
ing false information, journalists disseminating information
from the ground, diplomats trying to sustain nation’s image
and brand), bots can quickly create mass chaos, where mul-
tiple government agencies are working against each other.
The costs of uncertainty is greater for embassies in foreign
soil that may not have immediate access to their ministers
and either have to improvise to engage in real-time PR work
in their host country, or remain silent in order not to conduct
political work that may serve against home government. The
scale of computational propaganda is just too big for indi-
vidual embassies to make sense of and steer, signicantly
weakening their role during digital crises. The result revers-
es the diplomatic autonomy afforded by the digital media
revolution and reverts back to MFA-centric missions, effec-
tively weakening the effects of ‘digital diplomacy’.
Artificial Intelligence and Complex Tasks in
Foreign Affairs
Vyacheslav Polonksi wrote in his Independent column in Au-
gust: ‘There has never been a better time to be a politician.
But it’s an even better time to be a machine learning engineer
working for a politician’.
19
His view is supported by evidence
on how different A.I. tools help politicians run campaigns,
mine public opinion, predict voting patterns and engage
voters to build support on key issues. A.I. research and po-
litical science are currently overlapping across a wide range
of research agendas beyond elections and campaigning.
The notorious case of Cambridge Analytica - the data sci-
ence rm - which embarked on a micro-targeting campaign
by emotion mining through voter data based on fear-based
messaging,
20
is an important case. It is through this case
that we today know that A.I. can prole people through so-
cial media data, create tailored political advertisements that
cater to their emotional-psychological prole and success-
fully shift their political behavior as they interact with social
media platforms like Facebook, Twitter or Instagram.
Digital media isn’t the only place where A.I. is changing for-
eign policy. In June 2017, National Geospatial Intelligence
Agency - the US agency in charge of drone, satellite and
other aerial imagery network - decided to push for greater
Bradshaw, Samantha and Philip N. Howard ‘Troops, Trolls and Troublemakers: A Global Inventory of Organized Social Media Manipulation’. Oxford Project on Computational
Propaganda. Working Paper No. 2017.12. http://comprop.oii.ox.ac.uk/wp-content/uploads/sites/89/2017/07/Troops-Trolls-and-Troublemakers.pdf
18
Gillian Tett, “Trump, Cambridge Analytica and How Big Data Is Reshaping Politics,” Financial Times, September 29, 2017, https://www.ft.com/content/e66232e4-a30e-
11e7-9e4f-7f5e6a7c98a2.
Polonski, Vyacheslav, “How Articial Intelligence Conquered Democracy,” The Independent, August 9, 2017, http://www.independent.co.uk/news/long_reads/articial-intel-
ligence-democracy-elections-trump-brexit-clinton-a7883911.html.
20
19
Cyber Governance and Digital Democracy 2017/3
9
automation in visual data processing and collection.
21
Police
agencies in several developed countries are working on pre-
dictive models that collect and process real-time big data to
prevent crimes before happening, or stop popular mobiliza-
tion and protests before they grow in size and scope.
22
All
of this has implications on how threats are processed and
intelligence is collected on a global scale, in addition to how
states cooperate diplomatically to address the challenges
of these new technologies. The challenge of A.I. on human
conict was so great that more that in 2015 a number of
A.I. experts - including Stephen Hawking and Elon Musk -
signed an open letter calling for deeper research into the
nature of automation, along with its harmful effects.
23
One
of the best-known examples of concerns in the letter related
to ‘machine ethics’, exemplied by the question that ‘who
should a self-driving car choose to kill if accident is inevita-
ble?’ A similar debate brews in national security discussions
on autonomous weapons systems. In August 2017, Elon
Musk this time led a group of A.I. specialists with Alphabet’s
Mustafa Suleyman, in calling for the high contracting par-
ties to the UN Convention on Certain Conventional Weapons
(CCW) to ban on ‘killer robots’.
24
CCW had recently initiated
a formal session of discussions on the use of autonomous
weapons systems in conventional national militaries. Proce-
dural and budget-related problems had prevented a reso-
lution emerging from the meetings, pushing A.I. experts to
greater set of worries over delays. Regulating a new arms
race is a difcult task, just like regulating nuclear armament,
but one that needs to be addressed, given the enormous
destructive potential of such technologies. One main con-
sideration in the technology-diplomacy nexus is that if A.I.
progress outpaces international legislative capacity, large
number of countries will have to deal with the effects of tech-
nological processes that they haven’t approved or even un-
derstood.
While cybersecurity and cyberwar received the lion’s share
of diplomatic attention in the last decade, the trend is rap-
idly evolving into more complex issues such as A.I. govern-
ance, norm-building in automation and regulating machine
learning in multinational political processes. The Economist
brought the issue into mainstream policy debate in May
2017, by arguing that ‘the world’s most valuable resource is
no longer oil, but data’.
25
The statement followed the current
nancial trends whereby big oil corporations are replaced
by big data rms in terms of capital, wealth and political in-
uence. Access to larger volumes of data is comparable to
access of hydrocarbons, as companies compete with each
other over the resource and state actors (at least try to)
compete over control over the companies. Diplomacy has to
bring in signicant levels of technology know-how into nego-
tiations as multilateral negotiations have to settle internation-
al norms on what Google can do with users’ search history,
Facebook with how they share personal information for sales
data and how Amazon, with advertising companies.
There are also ongoing experiments in bringing data clos-
er to the heart of diplomatic profession. Diplomatic reports
and activities are performed increasingly in digital domain,
Samuel Gibbs, “Elon Musk Leads 116 Experts Calling for Outright Ban of Killer Robots,” The Guardian, August 20, 2017, sec. Technology, https://www.theguardian.com/
technology/2017/aug/20/elon-musk-killer-robots-experts-outright-ban-lethal-autonomous-weapons-war.
Bernard Marr, “How Robots, IoT And Articial Intelligence Are Transforming The Police,” Forbes, September 2017, https://www.forbes.com/sites/bernardmarr/2017/09/19/
how-robots-iot-and-articial-intelligence-are-transforming-the-police/.
“The World’s Most Valuable Resource Is No Longer Oil, but Data,” The Economist, May 6, 2017, https://www.economist.com/news/leaders/21721656-data-economy-de-
mands-new-approach-antitrust-rules-worlds-most-valuable-resource.
Alex Hern, “Experts Including Elon Musk Call for Research to Avoid AI ‘Pitfalls,’” The Guardian, January 12, 2015, sec. Technology, https://www.theguardian.com/
technology/2015/jan/12/elon-musk-ai-articial-intelligence-pitfalls.
McLaughlin, Jenna, “Articial Intelligence Will Put Spies Out of Work, Too,” Foreign Policy, June 2017, https://foreignpolicy.com/2017/06/09/articial-intelligence-will-put-
spies-out-of-work-too/.
24
22
25
23
21
Cyber Governance and Digital Democracy 2017/3
10
rendering diplomacy a function of data - not the other way
around. Large volumes of country data (i.e. in the UN, or
EU) requires increasingly larger processing power, which in
turn, will inevitably lead to the development of A.I. platforms
in dealing with key country data. North Carolina government
ofce for example, is building chatbots that are continuously
improved to answer real-time constituency questions. Singa-
porean government too, is using Microsoft-based chatbots
systems to assist their citizens in key government services
such as registration, licensing and utility management. In
the foreseeable future, automated legal counsel, document
support, classifying and sorting diplomatic inquiries, trans-
lation and document drafting will become highly automated
and will directly concern the conduct of diplomacy. Which
company will handle that A.I. work and how to conceptualize
that company in an international legal framework, or what
the company will do with that processed data is another fu-
ture challenge - one that will certainly be a topic of intelli-
gence agencies. Similar questions go to automating con-
sular services, visa background checks, evaluations and
decisions on visa outcomes. Embassies will inevitably build
large databases for their diasporas and engage with them
in the most cost-efcient way, which will introduce multiple
elements of automation into the picture.
Further discussions need to be made on autonomous ne-
gotiation and the prospect of A.I.-based diplomacy in key
bargaining processes. There are negotiation support sys-
tems that are currently in development that are applied to
several legal processes, from job negotiation to bail terms.
An innite number of possibilities are calculated by A.I. plat-
forms that continuously process key variables. For example,
a hypothetical A.I. diplomat negotiating a trade agreement
would have real-time access to all available economic, so-
cial and political datasets, partnering with its rival A.I. diplo-
mat to come up with a set of mutually agreeable offers, at the
fraction of costs and time spent by real diplomats. Diplomat-
ic A.I. haggling would innitely save time in multilateral ne-
gotiations like the Paris climate change agreement and push
several deadlocked political processes into a set of solu-
tions. This doesn’t imply removing the human touch from di-
plomacy or politics, as personality and individual skill will still
matter. Rather, AI-based negotiation imagines a diplomat-
ic future where diplomatic bargaining is more streamlined,
with redundant or time consuming tasks are outsourced to
bots, whereas more important, ‘high politics’ processes are
still managed by human diplomats. Proponents of AI diplo-
macy argue that human error and personal egos - primary
causes of escalation - will be removed from the majority of
negotiating topics, eventually leading to greater cooperation
between nations. Critiques on the other hand question how
objective AI diplomacy can be, since algorithms reect bi-
ases and personality traits of another caste: programmers.
Whether AI can truly be freed from human error and ego is a
hotly contested topic within computer science itself and its
answers aren’t necessarily convincing to the critiques of AI
diplomacy.
Future Trajectories in Computational Diplomacy
Contextualizing the rapid shift in communication technol-
ogies is a deeply confusing endeavor for diplomacy. The
sheer size of data produced and transmitted globally every-
day, prevents a proper framing of how MFAs can adapt to
these rapid changes, as well as disagreements over what
they exactly need to adapt to. Diplomacy has certainly more
Cyber Governance and Digital Democracy 2017/3
11
tools available at its disposal compared to a decade ago,
but also a larger, more diverse audience, less time and
greater uncertainty over communication. The elaborate sets
of communicative processes that emphasized trust and
common language developed over the course of centu-
ries are becoming increasingly obsolete, unnecessary and
slow. States too, have new sets of interests online and in
digital platforms that are built on automation. They all want
to build online inuence, get acknowledged and steer key
political processes that go on every minute, at different lay-
ers of global interconnectedness. The information related
to these processes no longer come solely from embassies,
intelligence agencies, or conventional media corporations,
but in platforms where such information is disseminated
equally to everyone. While these information intermediaries
are certainly not obsolete, they are also not as important as
they used to be and will have to devise new capabilities and
competencies in order not to end up irrelevant.
In August 2017, Facebook’s A.I. negotiator project went
sour as the chatbot started developing its own language
that cannot be understood by outsiders and learned how
to lie. This issue was a hotly debated topic at the Oxford
Internet Institute when I was a visiting fellow there. A retired
member of the British diplomatic service who was attending
a meeting, upon hearing this news, calmly told me: ‘Good.
Machines have learned how to conduct diplomacy’.
In coming years, MFAs will have to adapt to computational
diplomacy through a multitude of approaches:
Verication and Counter-Messaging: Depending on
nancial and human resources, MFAs can employ a
number of automation measures themselves. The pri-
mary use of automation will have to challenge and ver-
ify information owing at enormous volumes during cri-
ses and emergencies. Such verication tools can also
automatically get translated into multiple languages for
the use of a country’s embassies in foreign countries.
Furthermore, MFAs can use automation to detect tran-
sient inuence networks and misinformation sources
in real-time, helping substantially with attribution and
fact-checking.
Data/Sentiment Mining: Real-time scanning of social
media platforms through pre-determined set of word
combinations may allow MFAs to mine opinion and
sentiment, in issues related to foreign policy and po-
litical engagement. Whether a particular issue is de-
bated more within a certain age group, nationality or
geographic area is an important policy variable for dip-
lomatic missions. These engagement proles usually
vary between different policy agendas, so automated
data mining is usually a better way of proling engage-
ment compared to snapshots.
Digital Content Creation: Currently, several sports
media companies are experimenting with automated
content writing, that are growing more sophisticated in
a way that will soon eclipse human-based reporting.
Automated content curation is key for journalism and
can also be adopted by MFAs in similar fashion for na-
tion-branding, agenda-setting and awareness-building
campaigns. When combined with data/sentiment min-
ing, A.I.-based diplomacy efforts can also incorporate
automated policy writing, which can communicate a
pre-set policy preference across multiple language
platforms.
Diaspora/Business Engagement: Address, person-
al information, nancial assets and investment data
that belong to Diaspora groups abroad and business
commitments all benet from automation. Communica-
tion with both groups can be signicantly streamlined
through the use of chatbots that provide information on
elections, registration, taxes or trade agreement terms
to better mobilize and inform them with regard to both
home country and host country requirements. Chat-
bots are also potentially lifesaving in natural disasters
or other emergency situations to connect a large Dias-
pora group with necessary professional help.
Micro-negotiations: Micro-negotiators - bots that run
multiple rounds of negotiations based on vast sets of
data - can be vital to multilateral negotiations and save
signicant time to reach agreements. When the issue
that is negotiated is data-heavy (trade, infrastructure,
nancial and other numerical policy issues), micro-ne-
gotiators can do a far better and faster job than human
negotiators. These micro-negotiators can either reach
an agreement on their own, or assist senior negotiators
in determining political aspects of a settlement. In mul-
tilateral and multinational summits, micro-negotiators
can be even more valuable as sides can focus on more
human-centric aspects of negotiations. For example,
in discussing foreign aid sum and disaster relief, mi-
cro-negotiators can rapidly designate key aid areas
through geospatial imagery analysis and real-time so-
cial media posts from the region, signicantly easing
and streamlining aid negotiations.
Cyber Governance and Digital Democracy 2017/3
12
Automation doesn’t change the fact that diplomats and em-
bassies still matter. Foreign policy, like all politics, is a factor
of human condition, including sense, gut feeling and cultural
cues, along with its imperfections. However, there is a clear
trajectory whereby states that can best adapt to automation
- in war, foreign policy and economy - will develop more
efcient ways of dealing with the challenges of an intercon-
nected, data-centric world. Diplomacy too, can retain its
relevance and inuence over politics between nations, so
long as it can properly designate areas where automation
can help and where it can’t. Although all states will come up
with their own answers to these questions, based on their
own individual interests and needs, the common direction
in which automation and foreign policy is headed is more or
less similar for all countries. In the future, diplomacy has to
build data processing and management capabilities, with
dedicated departments and scientists supporting diplomats
and negotiators on the ground. The structure of this new
framework will also heavily depend on regime type, scope
of foreign interests and alliance behavior.
The structural shadow of uncertainty over diplomacy is
stronger than ever. Some communicative rituals and prac-
tices of diplomacy are growing more obsolete, as modern
political communication slides increasingly to short and
sharp rhetoric, coupled with automation tools that bom-
bard audiences at unprecedented levels. Diplomacy itself
is hardly obsolete however, as the task of mediating and
negotiating power relations is perhaps as important as it was
during the Cold War. New power centers - in the form of
technology companies and big data brokers - are changing
the state-centric parameters of classical realism perhaps,
but the inherent dynamics of power realignment still render
diplomacy a crucial endeavor. To rise to the challenge how-
ever, modern diplomacy has to develop a strong compu-
tational capacity, able to adapt to the changing nature of
digital communication and advances in automation.
Cyber Governance and Digital Democracy 2017/3
November 2017
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.