Monday, 15 October 2018

Using Open Source Intelligence, Social Media Data To Forecast #NigeriaDecides2019


Introduction: This research is aimed at harvesting and collating social media data, open source intelligence or publicly available information on political trends, narratives, opinion polls and employing the dataset to forecast the likely winner in Nigeria’s forthcoming presidential election slated to take place on February 16, 2019. Granted there are about 24 candidates in the forthcoming presidential election but realpolitik, pragmatism implies the election would be a two-horse race between two northern Muslims of Fulani extraction - Atiku Abubakar (Peoples Democratic Party, PDP) and Muhammadu Buhari (All Progressives Congress, APC). Naysayers would argue that politics is local; that elections are not won on social media. Despite its flaws, open source intelligence, social media and big data analytics has proven to be veritable tools for inter-alia: predicting social unrest, economic crises, medical epidemic, natural disasters, terrorism and also gauging the ‘political pulse’.
It boils down to connecting the dots and using the known to figure out the unknown. This is doable as long as available data is valid, reliable, and there are no unforeseen circumstances that could precipitate a disruption. A 2017 report published on The Conversation titled, ‘’How artificial intelligence conquered democracy’’, depicts how machine-learning systems predicts which United States Congressional bills will pass by making algorithmic assessments of the text of the bill.

Caveat: Though an active citizen, good governance advocate, writer is not a card carrying member of any political party in Nigeria and no one commissioned him to embark on this task. My endeavour here as an analyst and an information junkie, is to glean data from overt, disparate sources and try to make an inference.

First, Some Clarity…

Before delving into the nitty-gritty of this thesis, let us reconcile the uninitiated with some terminologies they will come across in this essay. In the Intelligence Community (IC), the term "open" refers to overt, publicly available sources. It contrasts with close-source, covert or classified information. Hence, open-source intelligence (OSINT), a component of all-source intelligence is defined by the Central Intelligence Agency (CIA), as the ‘’insight gained from processing and analyzing public data sources such as broadcast TV and radio, social media, websites, specialized journals, conference proceedings, think tank studies, photos, and geospatial information, amongst others.’’ In 2014, a survey of more than 1,200 federal, state, and local law enforcement professionals in the United States established that approximately 80 percent used social media platforms as intelligence-gathering tools. Open source intelligence provided some insight to goings-on in Iran during the 2009 Green Revolution. Writing on the Observer, Micah Halpern, a political and foreign affairs commentator opines that, ‘’Open Source Intelligence should be the go-to tool we use for thoughtful analysis and surveys on every issue’’. In his September 25, 2018 essay on Foreign Policy Magazine titled, ‘’Google Maps Is a Better Spy Than James Bond’’, Nick Waters an ex-British Army officer and open source analyst says, ‘’Open-source intelligence is, in fact, potentially far more reliable and checkable by a democratic public than traditional closed sources’’. Nick argues that, ‘’One of the primary strengths of open-source investigation is that the sources and methodology are exactly that—open. Anyone with a computer and an internet connection can follow the reasoning of analysts, interrogate the same sources, and critique the techniques used’’. 

Leading citizen journalism and sleuthing website - Bellingcat deployed painstaking OSINT (open source intelligence) techniques to establish what it claims is the real identity of the Russian military intelligence agents responsible for the nerve agent attack on former Russian spy Sergei Skripal and his daughter in the United Kingdom. Bellingcat reconciled Ruslan Boshirov as an alter ego of Anatoliy Vladimirovich Chepiga, a highly-decorated colonel of the Russian military intelligence agency, GRU and that the second suspect hitherto known as Alexander Petrov by UK authorities, is in fact the false name of Dr. Alexander Yevgenyevich Mishkin, a trained military doctor in the employ of the GRU.
Social media intelligence (SMI/SOCMINT) or social media forensics refers to the stack of tools that allows individuals or organizations to monitor social media channels, conversations, emerging trends. These nuggets of information are analyzed and synthesized into meaningful content and used to make strategic decisions across many disciplines. Just like everything in life, there is a drawback. The danger in social media intelligence (SOMINT) stems from the fact that some of the information are unverified, could have been outright disinformation or weaponized information disseminated by a bot.
Artificial intelligence (AI), is defined as the ''ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. Coined in 1959 by Arthur Samuel, the term ''Machine learning'' is a field of artificial intelligence (AI) that employs statistical techniques or algorithms to give computer systems the ability to "learn" (e.g., progressively improve performance on a specific task) from data and make predictions without being explicitly programmed. 
A similar concept, big-data analytics refers to the strategy of analyzing large volumes of data, or big data gathered from a wide variety of sources, including social networks, videos, digital images, amongst others, with a view to uncovering patterns and connections that might otherwise be invisible, and that might provide valuable insights.
Social Media, Data Analytics Disrupts Electioneering, Democracy
In our today’s data-driven world, social media has no doubt disrupted the way we communicate and is bracing to disrupt electioneering and by extension, democracy. During the 2016 U.S presidential election, data science firm Cambridge Analytica reportedly plunged an all-embracing advertising campaign to target persuadable voters based on their individual psychology. Similarly, UK Labour party’s successful digital media campaigns during the 2017 general election which saw them gain 21 seats from its Conservative opponents also buttresses the efficacy of social media to politics.

Cyberspace, Social Media As Election Battlegrounds
An average smartphone user clicks, taps and swipes his/her seductive gadget an amazing 2,617 times a day. The proliferation of smartphones and high rate of internet penetration entails many people are inadvertently addicted to their smartphones and social media. The Nigerian Communication Commission (NCC) says there are 103 million internet users in Nigeria as at July 2018. 
We also know that Nigeria boasts of the highest Facebook users in Africa; there were 26 million active Facebook users in Nigeria, as at May 2018. A plethora of Nigerians are also active on Twitter, Instagram, LinkedIn and other social media platforms. There are manifold implications: political campaigns, electioneering are no longer the exclusive preserve of traditional mass media/advertising platforms such as television, radio stations and newspapers. Now, cyberspace, social media is also political and campaign battleground. Another import of this is that analysis of social media, open source data, prevailing political narratives, trends, pulse of the country, could hint who Nigeria’s next president would be. By the way, recall that social media data, opinion polls favoured and foretold Buhari’s election victory in 2015. Cognizant of the huge role that social media played in Buhari’s victory in the 2015 presidential election, the Buhari administration through the Ministry of Information and Culture reportedly forked out N180 million to pay social media influencers in the 2017 federal budget. As a matter of fact, the Buhari Media Centre (BMC) was recently rebranded, rejigged and morphed into the ''Buhari New Media Centre'', BNMC with twitter handle: @BuhariCentre, WhatsApp forum, with membership cutting across the 36 states of the federation and the Federal Capital Territory (FCT).

Can Social Media Intelligence, Twitter Predict Election Outcomes?

In an interview with the International Center for Journalists in Nigeria (ICFJ) aftermath of the 2015 presidential election, Sunday Dare, a journalist and former aide to the national leader of the All Progress Congress (APC), Bola Tinubu, submitted that, ‘’The [2015] election was decided, dominated and directed by social media. The power of social media came out for this country. Social media played a central role as a watchdog in keeping the integrity of the process. Within minutes of votes being counted at a polling unit, the results were all over social media’’.
Nick Beauchamp, an assistant professor of political science at Northeastern University, said in an academic paper published in the American Journal of Political Science, in 2016, that his application of machine learning to more than 100 million tweets during the 2012 election cycle closely mirrors the results of state-level political polling. The academic paper asserts that machine learning and an analysis of Twitter trends outdo traditional polls on the national level. Similarly, a report released by Dublin City University in 2011 suggests that Twitter data could be used as an accurate indicator of election outcomes. An artificial Intelligence company - MogIA predicted the outcome of the last four presidential elections in the United States, including Trump’s win. MogIA never asked a single question of a single voter but utilized OSINT and artificial intelligence to analyze millions of datasets from public platforms such as: Google, Facebook, Twitter and YouTube in the U.S. and then created its predictions. Similarly, BrandsEye, a tool that looks at people's tweets, correctly predicted both the vote to leave the EU (Brexit referendum) and a Trump victory in the 2016 US election. Brandwatch’s analysis established that Trump had more Twitter mentions throughout most of the election cycle. At a time, Trump accumulated more than 4.9 million mentions while Clinton had just over 2.7 million.
According to a 2015 study titled 'How Africa Tweets', an annual study by Portland Communications which highlights Twitter trends, hashtag usage and online habits in Africa, Portland analyzed 1.6 billion geolocated tweets and the top 5,000 hashtags for 2015. The most tweets in Africa came from Egypt which accounts for about 500 million. Nigeria came in second with 360 million tweets followed by South Africa, Kenya, and Ghana. 
Given this result, it is safe to say that Nigeria has the second largest amount of Twitter users in Africa.
Similarly, Portland Communications July 2018 report which covered polls held in 10 African countries between June 2017 and May 2018, highlighted how Africans in the Diaspora influenced elections in Africa. The report says, "People continue to seek out the voices they trust with established journalists and news outlets consistently ranked in the top three influencers across all elections. With fake news and bots influencing conversations on social media, people continue to search for traditional sources of verified, accurate information’’.

Case Study: Nigeria’s 2015 Presidential Election Opinion Polls

Buhari led Jonathan whom he defeated during the 2015 presidential election in several opinion polls, including the poll administered by Jonathan’s Special Assistant on Social Media, Reno Omokri and also by online newspaper, Premium Times, JJ Omojuwa, the Africa Independent Television, AIT, World Stage Group.
Out of the total 9,206 respondents to the poll, 8,176 Nigerians said they would vote for Buhari. Buhari also bested Jonathan in another opinion poll by Daily Post newspaper. Out of a total of 9,204 votes in the poll, Buhari emerged winner by polling 6,129 votes (66.59 percent) and Jonathan, with 2,516 votes (27.34 percent). Buhari also emerged a winner in another poll conducted by political risk research and consulting firm, Eurasia Group. Buhari reportedly got 60 percent while Jonathan got 40 percent. The dissenting poll was the one conducted by Kevin, Charlyn and Kimberly Associates, a United Kingdom (UK)-based research and political risk consult.
2019 Election: How Buhari, Atiku Fair On Twitter, Online Polls
Northern Nigeria's only surviving newspaper - Daily Trust launched a twitter poll on October 12, 2018. The poll asked: ‘’who are you supporting in the 2019 presidential election? Retweet for @MBuhari, like for @atiku. Notwithstanding the unfair advantage of retweet button allocated to @MBuhari by Daily Trust, Atiku trounced Buhari in the poll. As at the time (1920 Hours, October 14, 2018) of writing this report, Atiku had amassed 14,500 ‘votes’ while Buhari lagged behind with an abysmal 4,699 ‘votes’. This poll seems to portray the pervasive disenchantment with the Buhari administration. I think this shows how #Atikulated many Nigerians are. At the peak of the #Change mantra and #SaiBaba frenzy in 2015, it is unthinkable that Atiku would trounce Buhari in a poll.
Daily Trust Newspaper Twitter Poll
Recall that the ruling APC has won every single opinion poll conducted by Daily Trust in the last 4 years (2014- 2018).
A Nairaland member's comment

Similarly, a recent online
poll titled Atiku vs. Buhari: Who can manage the economy better? Conducted by Nairametrics, Nigeria’s leading financial literacy and business intelligence website, more than half of the respondents to the poll (55%) say Atiku will be able to manage Nigeria’s fragile economy better than Buhari. Presidential candidate of the SDP, Donald Duke came second with 24% of the poll, while President Buhari scored the least (21%) among the trio.

PDP’s candidate, Atiku trounced Buhari in an earlier twitter poll conducted by pro-Buhari groups - Mark Essien via @markessien and @YNaija.  Atiku also floored Buhari in a 6-day Twitter poll conducted by Japheth Omojuwa, a prominent Buhari proponent. In the first poll, Atiku garnered 35% of the votes compared to Buhari’s 32%. In the first online poll conducted by Mark Essien, Atiku polled 43% as against Buhari’s 35%. In the second poll, Atiku secured 39% of the votes while Buhari came second with 34%. In both polls, respondents were asked: “Who would you vote for President if elections were held today?” Of the total combined votes of 81,065, Atiku secured 29,672 representing a cumulative percentage of 36.6% compared to Buhari’s 26,590 representing 32.8%. If both polls - Mark Essien and Omojuwa’s are combined, Atiku will enjoy a 9% lead over Buhari.

Polls by NOIPolls Limited in October 2015 showed that Buhari had 80% approval rating. Between June 2015 and April 2016, President Buhari’s approval rating sank to 42 per cent. A survey conducted by the renowned NOI Polls Limited asserts that President Muhammadu Buhari's job approval rating dropped to its worst in 18 months, at 38% as at February 2018. This is the president's second lowest score since he was sworn-in over two years ago. In another poll conducted sometime in May 2018 ‘amongst 4,000 Nigerians’ across Nigeria’s 36 states and the Federal Capital Territory by Buharimeter, a brainchild of the Centre for Democracy and Development (CDD) which was launched to track the delivery of President Buhari’s campaign promises, Nigerians rated President Buhari below average even in his three cardinal campaign promises: Corruption, Security and Economy. According to this poll, President Buhari’s approval rating stood at 40, a decline of 17 per cent from 57 per cent rating recorded in the 2017 Buharimeter National Survey. Another poll conducted on www.politicsngr.com which lasted between May 21 - June 3, 2018, respondents were asked, ‘Will you vote for President Buhari in 2019’?  5,227 votes were recorded. Of the figure, 3,687 votes representing 70 per cent voted against Buhari re-election bid while 1,538 votes representing 29.4 per cent supported his ambition.

However, President Buhari defeated Atiku was in a Twitter poll organized by @SaharaReporters on November 26, 2017. 49 percent (a total of 21,704 Twitter users participated in the poll) of respondents say they would vote for Buhari (APC) while 26 percent opted for Atiku Abubakar (PDP). 25 percent said they would vote for neither candidate.
Atiku trending worldwide on Twitter after he emerged PDP candidate. Atiku @ number 1 while Buhari @ 12
Conclusion:

Politicians and their foot soldiers are seemingly the most optimistic people in the world. I am reminded that a week, let alone four months to a presidential election is a long time in politics. The gospel truth is that Buhari has lost the mojo, massive goodwill he enjoyed in 2015. I think insecurity and more importantly the economy would tilt the balance in the 2019 presidential election. People, Nigerians are suffering! Allan J. Lichtman, distinguished professor of history at American University who correctly predicted 30 years of United States Presidential elections including that of Trump which many opinion polls got wrong, says, ‘’presidential elections don’t work the way we think they do,” “They’re not decided by the turns of the campaigns, the speeches, the debates, the fundraising. Rather, presidential elections are fundamentally referenda on the performance of the party holding the White House. If that performance is good enough, they get four more years. If it’s not, they’re turned out and the challenging party wins.” I think Prof Lichtman’s hypothesis pertinent to Nigeria. If the current mood of the nation, open source, social media intelligence is anything to go by, the odds are against President Buhari. Barring any unforeseen circumstances, hate or love him, Atiku is in all probability, the next president of Nigeria.

Written By:
© Don Okereke, a security analyst/consultant/open source intelligence junkie/writer/active citizen

October 14, 2018

Follow me on Twitter: @DonOkereke

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

***This essay is an adaptation of my presentation: ‘’Social Media and Data-Driven Targeting in 2019 General Election’’ presented During The 2018 Cyber Secure Nigeria Conference Organized By The Cyber Security Experts Association of Nigeria (CSEAN), At Four Points By Sheraton Hotel, Victoria Island, Lagos, Nigeria on Wednesday, May 9-10, 2018.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

References:
1.                  ‘’How the internet flips elections and alters our thoughts’’
2.                 ‘’The Data That Turned the World Upside Down’’
3.                 ‘’How our likes helped trump win’’
5.                 'Information warfare': How Russians interfered in 2016 election
6.                 ‘’Facebook Failed To Protect 30 Million Users From Having Their Data Harvested By Trump Campaign Affiliate'
7.                  Research in India suggests Google search results (Search Engine Manipulation Effect, SEME) can influence an election.
8.                  “Big data revolutionized the way American politicians win elections. In the process, it broke American politics.”
9.                  The great British Brexit robbery: how our democracy was hijacked
10.             Cambridge Analytica and the Secret Agenda of a Facebook Quiz
11.            Cambridge Analytica boasts of dirty tricks to swing elections
12.            Murder, Fake News, And Hacking Concerns Cloud Disputed Kenyan Election
13.            ‘’Practical ways to restructure Nigeria before 2019 election’’
14.            85% of Nigerians use smartphones for product research
15.            How NC is meeting the threat to election security http://www.charlotteobserver.com/opinion/op-ed/article202454724.html
16.            Strengthening the Integrity and Transparency of Elections in the Age of Social Media https://en.unesco.org/news/strengthening-integrity-and-transparency-elections-age-social-media
17.            Improving the communications and information ecosystem to protect the integrity of elections https://en.unesco.org/sites/default/files/633_18_gni_intergrity_of_elections_final_report_web.pdf
18.            Online Political Microtargeting: Promises and Threats for Democracy, Frederik J. Zuiderveen Borgesius, Judith Möller, Sanne Kruikemeier, Ronan Ó Fathaigh, Kristina Irion, Tom Dobber, Balazs Bodo, Claes de Vreese*
19.            Troops, Trolls and Troublemakers: A Global Inventory of Organized Social Media Manipulation, http://comprop.oii.ox.ac.uk/wp-content/uploads/sites/89/2017/07/Troops-Trolls-and-Troublemakers.pdf
20.            Measuring the reach of "fake news" and online disinformation in Europe, http://reutersinstitute.politics.ox.ac.uk/our-research/measuring-reach-fake-news-and-online-disinformation-europe
21.            The “election management” company uses big data and psychometric profiling in operations designed to suppress voter segments., https://globalsecurityreview.com/cambridge-analytica-darker-side-big-data/
22.            Is political micro-targeting hijacking European democracy? https://policyreview.info/articles/news/political-micro-targeting-hijacking-european-democracy/753
23.            'It might work too well': the dark art of political advertising online, https://www.theguardian.com/technology/2018/mar/19/facebook-political-ads-social-media-history-online-democracy
24.            Why Facebook Showed You That Ad for the Candidate You Hate, https://www.nytimes.com/2016/11/08/us/politics/facebook-ads-campaign.html
25.            Russian Hackers Hijack Satellite To Steal Data from Thousands of Hacked Computers, https://thehackernews.com/2015/09/hacking-satellite.html
26.            How Disinformation And Distortions on social media affected elections worldwide, https://www.npr.org/sections/alltechconsidered/2017/11/16/564542100/how-disinformation-and-distortions-on-social-media-affected-elections-worldwide
27.            The Power of Social Media in the Nigerian Election https://www.icfj.org/blogs/power-social-media-nigerian-election
28.            EXCLUSIVE — Research: Google Search Manipulation Can Swing Nearly 80 Percent of Undecided Voters, www.breitbart.com/tech/2018/04/24/exclusive-research-google-search-manipulation-can-swing-nearly-80-percent-undecided-voters/
30.            Can Social Media Predict Election Results? Evidence From New Zealand
Michael P. Cameron, Patrick Barrett &Bob Stewardson

31.            Analysis of social media did a better job at predicting Trump’s win than the polls