The Hindsight Files 2020: Much More Than Politics

What do an oil company in Kenya, the Ministry of Health in Ghana and UNDP have in common? According to The Hindsight Files, they all hired Cambridge Analytica.

Key points

A proposal to Tullow Oil in The Hindsight Files outlines "academic research" surveys in the Turkana region of Kenya, but the results were intened to "inflence and modify" behaviour, misleading participants.

According to The Hindsight Files, The Ministry of Health in Ghana hired Cambridge Analytica to conduct a nationwide study on attitudes towards healthcare, the results of which Cambridge Analytica used in future campaigns for the ruling NDC party. This included encouraging opposition supporters not to vote.

In all the current cache of files we've so far found no mention of data protection, data security, or even an acknowledgement that the data collected was sensitive eg. health data.

News & Analysis

On New Year's Day, the Twitter account @HindsightFiles began publishing internal communications and documents from the now defunct SCL Group, dating from 2014-2018. They came from the hard drive of Brittany Kaiser, who held several senior positions at SCL Group including at one of its subsidiaries, Cambridge Analytica, and featured in the Netflix documentary "The Great Hack".

Privacy International first investigated Cambridge Analytica in 2017. We questioned the company's role in the Kenyan President's electoral campaign and the production of a series of online videos making incendiary claims about the opposition in a volatile political climate. Since then, thanks to the much-publicised Cambridge Analytica/Facebook scandal of 2018, we have learned much more about the ecosystem of companies where data exploitation is the business model, of which Cambridge Analytica was just one of many. The first Hindsight Files cover Kenya, Brazil, Malaysia, the USA (in files marked Iran and Bolton) and the second release covers Ghana. While protecting democracy is the reason Brittany Kaiser gives for releasing these documents, there is more to the files than politics and elections.

In order for companies like SCL Group to be able to carry off their bold claims of being able to change people's behaviour they need data, and lots of it, so we need to look at where they are getting it from. From information in The Hindsight Files, the different research projects SCL Group undertook meant the company held a lot of insights into people's behaviour and lots of data.

For example, a 92 page proposal to Tullow Oil, the largest oil and gas exploration and production company in Africa, is particularly revealing (see document 1 below). In 2012, Tullow Oil announced it had discovered oil in Turkana County in the north west of Kenya. Perhaps anticipating tension with Turkana's communities, at some point in 2015 SCL drafted a proposal for "research and consultancy" services in Kenya and also Ghana.

At first the proposal starts out describing PR services "to understand how best to engage positively with its host communities" through large-scale quantitative and qualitative research. But the goals of this research into the behaviours of local communities is hardly positive, as the goals of this extensive and detailed research is to better manipulate the people of Turkana. The proposal states,

"Additional deliverables will include comprehensive recommendations on the best cognitive and emotional strategies to influence and modify Turkana attitudes and behaviour. Also included will be conclusions on potential spoilers’ weaknesses and recommendations on how best to utilise these to the Tullow Oil’s advantage."

It is not clear if Tullow Oil went through with this proposal, but it will not come as a great surprise to most that an oil company would consider contracting this kind of company employing these kinds of tactics in a region described by SCL as, "the very least developed of Kenya’s counties and the Turkana themselves amongst its most idiosyncratic and fractious communities".

The proposed research violates even the most basic of research ethics guidelines, with SCL looking to actively mislead research participants as to the purpose of their questioning. They state that, “All research will be presented to respondents as an academic research study”, which is clearly misleading. The proposal does not mention any security arrangements for the data collected, privacy policy or any consideration that the information they are collecting is sensitive, let alone the psychological profiles to be compiled.  At the time, Kenya had no data protection law, meaning the company could collect any kind of information with no limitations or safeguards and do what they liked with it. For example, they could keep the data and feed it into future work on, say, elections.

From The Hindsight Files, this is what appears to have happened in Ghana. Several documents reference that SCL were commissioned by the Ministry of Health in 2014 to conduct a large study on health practices and requirements in Ghana. SCL conducted 30,000 household surveys, in 97 constituencies, across 10 regions. In a later report to the ruling National Democratic Congress (NDC) party, research based on 30,000 surveys etc, "allowed SCL to model the future vote distribution within each constituency in Ghana based on how respondents said that they would vote should there be an election tomorrow" (see document 2 below).

According to documents, SCL's work for the NDC included a negative campaign ultimately trying to get opposition supporters not to vote (see document 3 below). A further unaddressed letter on SCL Group letterhead boasts,

"We hold a large scale data set relating to various aspects of public health in Ghana, including people’s perceptions of healthcare facilities, health insurance, and factors influencing access to medicine and treatment. We also hold data on themes including which national and local issues are important to people, their perceptions of the state of the economy, popular media channels, and key influencers. In addition we have detailed demographic data".

The letter goes on, "SCL can offer expert data analytic services which could assist you [the unamed recipient] in several ways" (see document 4 below). It is worth remembering that the Ministry of Health in Ghana paid for the dataset that SCL is selling.

Another disappointing find in these files is that organisations like UNICEF and the United Nations Development Programme (UNDP) commissioned SCL to conduct extremely sensitive research from 2009-2012. In a section on "recent relevant projects", the Tullow Oil proposal states UNICEF commissioned SCL to produce a communication strategy on preventing child marriage and enhancing the protection of girls.  UNDP commissioned three studies- one on disarmament in South Sudan in 2011 and two studies on youth radicalisation in Somalia and Egypt. There are glowing references from government agencies in the UK and the USA relating to counter-terrorism projects.

In all the current cache of files we've so far found no mention of data protection, data security, or even an acknowledgement that the data is sensitive, as data relating to people's health cetainly is. It seems organisations have sleepwalked into giving SCL a huge amount of sensitive data without seemingly knowing what they will do with it.

Knowing what we know now about SCL Group and how they progressively ramped up their data collection methods, culminating in the 2018 Facebook/Cambridge Analytica scandal, organisations that have used their services should be coming clean about what happened to the data collected. While many believe that Cambridge Analytica and others are merely peddling snake oil disguised by clever marketing, this doesn't make the underlying intrusions into people's privacy or the potential to stoke tensions or even conflict in fragile regions any less real. More files are expected to be published, but already a solid picture is forming that nobody should be using companies like SCL or others to carry out such sensitive research, especially in places where no protection exists for the people being studied.