In the early hours of July 29, 10 days before Kenya’s presidential election, Chris Msando, a 44-year-old information technology manager at the Independent Electoral and Boundaries Commission, got behind the wheel of his Land Rover in Nairobi’s central business district.
Msando, a plump, bespectacled father of four, had only been in his role since May. His predecessor at the IEBC, the body overseeing Kenya’s elections, had been placed on leave after he allegedly refused to cooperate in an audit of its technical systems. In his three months on the job, Msando had gained a reputation for competence, and integrity, in a position of critical importance. Earlier that evening, he’d appeared on national television to explain the workings of the commission’s high-tech answer to the suspicions that had long dogged election results in the country.
Msando, though, would never get a chance to make sure the system functioned as intended. After leaving the broadcast studio, he spent the rest of the evening at a downtown club with friends, before getting in his car with three unknown individuals shortly before 2:00 a.m. Footage from city CCTV cameras shows the vehicle circling erratically through the Kenyan capital, until it disappears on a highway traveling north an hour later. The IT manager’s body, along with that of a 21-year-old female companion, was found later that morning in a wooded area roughly ten miles from the city center. An autopsy revealed Msando was tortured and then killed by strangulation.
Msando’s death came as the country braced for a political re-match between the incumbent president, Uhuru Kenyatta, son of Kenya’s first president, and Raila Odinga, a former prime minister and longtime opposition leader of the country. Odinga had run for president three times before, including 2007, when he lost to Mwai Kibaki under heavily disputed circumstances, touching off a horrific wave of violence that still clouds the country’s politics. Odinga, who heads the National Super Alliance, or NASA, ran again in 2013, and lost, narrowly, to Kenyatta. This year, in the vote on August 8 that followed Msando’s brutal killing, Odinga lost again to Kenyatta and his Jubilee Party. Three weeks later, the Supreme Court issued a stunning verdict overturning the result, only to have the country return Kenyatta to the presidency once more in a repeat vote held in October.
Nearly three weeks after the IEBC had declared Kenyatta’s August victory, Odinga paid a visit to the late Msando’s home. Msando’s six-year-old son surprised the aging politician with a question: Why had his father been murdered, and who did it? Odinga stared at the boy before responding. “We don’t know yet,” he said after the pause. “We are trying our best to find your father’s killers and I promise that we shall.”
Four months later, that promise remains unfulfilled. Msando’s killers have not been found, and the story of Kenya’s election failure has yet to be fully told. What we do know is that the technology Msando managed—systems the country hoped would help resolve entrenched anger over repeated disputes at the ballot box—failed in ways that offer lessons about what technology can and cannot do to protect the integrity of an election. Msando’s unsolved case remains a symbol of a fraught electoral process—one that was supposed to be legitimated through technology but may have ultimately been harmed by it.
An Election Disputed Before It Began
Although Kenya stands out as a relatively mature democracy in a region better known for despots, its elections have long been tainted by allegations of vote rigging. Given that context, the murder of Msando, keeper of the technology meant to ensure a clean vote, kicked off a raft of speculation. Almost immediately, Odinga’s NASA Coalition accused Kenyatta’s Jubilee Party of involvement. “There’s no doubt in my mind,” Odinga would tell reporters, “that this heinous crime was committed by people intending to interfere in the electoral process.” Others countered that Odinga could be behind the crime himself, as a way of discrediting an election he expected to lose.
Odinga, whose father was also a leading figure of Kenya’s post-independence politics, had reason to be suspicious of the election process. He has a relatively strong case that the election was stolen from him in 2007. In the aftermath of the crisis that followed that election, a South African judge, Johann Kriegler, released an independent report that detailed widespread vote manipulation. Kriegler’s findings included a smorgasbord of old-school rigging tactics: intimidation, bribery, ballot stuffing, and problems with the register that enabled as many as 1.2 million Kenyans to cast votes in the names of dead people. The conduct of the election, Kriegler noted, “was so materially defective,” that it was impossible to know who actually won. In the chaos that ensued, more than 1,000 people were killed and 500,000 displaced.
Although Kenya has avoided 2007-style violence this time around, dozens of people, mainly supporters of Odinga, have been killed by security forces since August. And questions about the performance of the voting system in 2017 linger. While parts of the IEBC’s systems, in Msando’s absence, worked as planned, others failed in ways that called into question the integrity of the result, which gave Kenyatta a win with 54 percent of the vote. After Odinga filed a legal challenge to the election outcome, the Supreme Court answered by overturning it. On September 1, in a startling announcement, Chief Justice David Maraga declared Kenyatta’s victory “invalid, null, and void,” and ordered a new vote within 60 days.
An Odinga victory was not meant to be. The court’s decision, which marked the first time a presidential vote had been annulled in Africa, was hailed as a milestone of judicial independence. But it would only leave the country more divided. Despite hailing the Supreme Court ruling, Odinga boycotted the repeat vote held on October 26, claiming the IEBC could not hold a credible poll; Kenyatta breezed through with 98 percent of the vote and was inaugurated in November for a second term. Odinga, who insists he won the poll in August, remains defiant, calling himself the “people’s president” and urging his followers to boycott government-affiliated businesses.
The Effort to Protect the Vote
The dimly lit wood-paneled offices of the IEBC, where I met with staff on a December afternoon, are hardly evocative of an institution transformed by tech. Yet Kenya’s election-management body, like its counterparts around the world, has embraced the digital revolution. According to David Carroll, director of democracy at the Carter Center, who’s monitored more than 25 years of elections for the Atlanta-based nonprofit, a majority of countries now incorporate at least some digital elements into their voting processes. The technology varies widely. Electronic voting machines, designed to eliminate spoilage associated with paper ballots, such as Florida’s infamous “hanging chads,” were first introduced in the 1980s. India, the world’s largest democracy, now uses EVMs, as they’re often abbreviated, in all of its polling stations, nearly a million. Other countries, including a growing number in Africa, have incorporated elements of the system used in Kenya, where voting still takes place on paper ballots, but registration and result transmission are digitized. In November, the self-declared state of Somaliland held the world’s first election where voters were identified through iris scans. Over the last decade, more than a dozen countries have conducted online voting pilots and one, Estonia, currently offers online voting to all its citizens.
Kenya’s embrace of voting tech is a consequence in part of the violence and chaos that started in 2007. To prevent cheating in the future, Kriegler, the South African judge, recommended introducing the two innovations that would become the heart of the IEBC’s toolkit: a biometric register, partly to help ensure all voters are living, and an electronic transmission system, intended to reduce anxiety-producing delays in the tabulation process and improve the transparency of figures from the polling station level on up.
In theory, Kenya, a leading African tech hub, was well-placed to absorb these innovations. In 2013, however, both systems encountered fatal problems. More than half of the biometric kits failed, some due to a lack of back-up batteries, forcing election officials to revert to the manual paper register. The transmission system widely malfunctioned as well and an alleged programming error vastly inflated the number of legitimately rejected votes. The IEBC ultimately abandoned the electronic tallying entirely, before certifying Kenyatta as that year’s winner—a process Odinga unsuccessfully challenged in court. Knowing that history, Kenyans, who began queuing to vote as early as 4:00 a.m. on August 8, 2017, had every reason to approach the systems in use this year with skepticism.
Did the promise of digitally enforced integrity make the election more transparent, or less? As Odinga supporters argue, did it facilitate vote rigging in cyberspace that simply superseded old-fashioned ballot-box stuffing? Or, as the government insists, did it raise voter expectations to unrealistic levels, and ultimately cast more doubt on Kenyatta’s victory than it should have? As governments around the world enhance the technology of their own voting processes, what can others learn from Kenya’s experience?
A good place to start, elections experts say, would be an honest assessment of what digital election systems can do well and what they cannot. John Githongo, an anti-corruption activist and chairman of the Nairobi-based African Center for Open Governance, argues that technology holds great promise in improving an election’s efficiency, particularly in a country like Kenya, where poor infrastructure prevents many polling stations from physically delivering results in a timely manner. The problem, he says, is when technology is introduced as a substitute for something it cannot change: a society’s fundamental level of trust.
“Technology can play a very positive role in elections,” Githongo tells me, “but it typically overpromises. It can improve the integrity of a process when there’s integrity to build on. But it cannot introduce integrity when there is none.”
What Happens When a “Foolproof” System Fails
2017 was supposed to be different. Even after Msando’s death, the IEBC swore by its technical systems. “Our technical team has worked on this technology for months. It is foolproof and can only suffer a hitch through human intervention,” the commission’s CEO told reporters just before the vote. But as the vote counting unfolded, it quickly became clear that something was amiss. An apparent server overload when results began to transmit caused delays in figures being posted, heightening suspicions. More worryingly, vote totals from nearly a quarter of all polling stations arrived at the national tallying center without required scans of official documents needed to verify the numbers. It later emerged that a significant number of forms from polling stations and the country’s 290 constituencies lacked security features like signatures, stamps, and official watermarks, meaning they could have easily been forgeries.
Odinga’s team denounced the entire poll as a charade, claiming that unknown individuals used Msando’s login credentials to hack into the IEBC servers. There, NASA alleged, they introduced an algorithm that caused the server to effectively override all incoming figures and give a constant 11-point advantage to Kenyatta. NASA maintains that Odinga won the election outright, but it has yet to produce a smoking gun. Evidence it claimed showed manipulation of the IEBC servers was widely debunked. Multiple international monitors told me they have no concrete evidence to substantiate the claims of hacking.
Yet widespread tampering does appear to have taken place. According to the Supreme Court ruling, a significant number of the forms that were used to declare Kenyatta’s official victory could have been fakes. In an internal IEBC memo leaked to the press shortly after the court ruling, the body’s chairman, Wafula Chebukati, claimed someone had created an unauthorized account in his name on IEBC server and used it to conduct nearly 10,000 suspicious transactions. In an effort to gain a window into any behind-the-scenes manipulation, the court had ordered the IEBC to release the log-in trails of its servers, which would give a record of who accessed the systems and when. The body, however, provided pre-downloaded logs that it could not prove were from the servers in question. As the court’s full ruling noted, the IEBC had squandered a “golden opportunity” to debunk Odinga’s claims of hacking. This lack of transparency, combined with the paper form discrepancies, was enough for the court to conclude it could not verify the election’s results were accurate. Without ever declaring explicitly that fraud had occurred, the court ruled that the procedural problems with the election were severe enough to warrant a new vote.
Voting Systems That Helped While Hurting
In several ways, technology did improve the conduct of the August vote. Compared to the election of 2013, the biometric kits worked reasonably well in 2017 and appear to have minimized fraud associated with the register. Despite the irregularities the Supreme Court called out, scanned forms from a majority of polling stations did reach the national tallying center, and the IEBC posted data from individual polling stations on its website within a few days. Although these electronically transmitted figures were not official—and their credibility was undermined by the missing scans—the public presentation of this level of detail is far from standard practice. “In many countries, you don’t have access to polling station-level data for weeks, months—sometimes never—and without that kind of data it’s much harder assess things fully,” the Carter Center’s Carroll told me. “Had Kenya’s transmission system worked a little better, it would have been a real significant achievement.”
But the IEBC’s reliance on technology had unintended consequences. Both sides of the political divide, in fact, could reasonably argue that the existence of the electronic transmission system set the wrong expectations. To Odinga’s camp, the devices functioned as an extra layer of bureaucracy; by obscuring the vote tallying process, the technology could have given the government cover to manipulate the figures from behind the scenes. Kenyatta supporters counter that confusion regarding the operation of the technology opened up the tallying to excessive scrutiny, unfairly maligning a process that, though far from perfect, was more or less sound and upheld the will of Kenya’s voters. In our meeting at IEBC headquarters, the commission’s public affairs manager, Andrew Limo, accused Odinga’s NASA Coalition of “exploiting the mystic world of technology” in order to deceive the average voter into believing the whole process was a fraud. Neither side’s argument bodes well for the idea of technology as a cure-all for an ailing democracy. “This technology is meant to increase credibility, but in some ways it makes the process less transparent,” says Gabrielle Lynch, a professor of comparative politics at the University of Warwick who’s written widely about the Kenyan election. “It creates a kind of black box that you cannot see inside, which opens the door to suspicion.”
The technologies are also expensive. Kenya’s voting kits put taxpayers back an estimated $52 million for the August 8 election, and an additional $24 million when their vendor, the French firm OT-Morpho, reformatted the devices prior to the repeat poll in October. What’s more, using an outside supplier opens up the process to accusations of foreign interference. Following the August vote, NASA-affiliated members of parliament accused OT-Morpho of providing kickbacks to IEBC and Jubilee officials, while “willfully allowing” unauthorized access to its systems and therefore abetting rigging—charges the company denies. Others warn that technologies capturing voter data could be used to undermine citizens’ privacy. Marietje Schaake, a member of the European Parliament from the Netherlands, who led the EU’s election observation mission to Kenya, has noted with concern that the country does not have data protection laws. In theory, authorities have carte blanche to misuse data collected during the electoral process, including information from the biometric register. (In the months leading up to the election, Kenyatta’s team reportedly hired the UK-based Cambridge Analytica, which has garnered headlines because of its role in the election of U.S. President Donald Trump, to compile demographic data on Kenyan voters).
External election observers also struggle with digitized elections. Those missions have traditionally monitored elections’ wider political contexts and physical voting processes, but they admit they are less equipped to assess the performance of electronic devices or what’s happening inside servers. Following the Supreme Court decision in September, both Kenyan and international media were widely critical of foreign observers, including the co-head of the Carter Center mission, former U.S. Secretary of State John Kerry, whom they accused of giving the vote a premature clean bill of health. While Kerry and others’ statements were nuanced and stressed the need to avoid drawing immediate conclusions, the saga underscores how hard it can be to know what’s going on behind the scenes, particularly in a high-tech election like Kenya’s. According to Carroll, this is unlikely to change: Wading deep into an election’s digital underbelly, for example, by certifying computer code or software, would be too intrusive a step for missions mandated to function as independent observers. Nonetheless, in a typical election “observers want to know that there was some kind of independent certification of the digital and electronic systems,” he says. “What’s key is knowing that the systems in place allow for various windows to verify electoral data, and that they were subject to an independent technical review which the major parties were satisfied with.”
How Much Can Technology Fix?
One of the many ironies of Kenya’s 2017 election is that it was arguably the country’s most legitimate of the past decade. Kenyatta’s victories in 2013 and 2017 came in large part from his savvy exploitation of Kenya’s ethnic voting blocs. By making a former opponent, William Ruto, his running mate, Kenyatta captured the vote of many Kalenjin, one of the country’s largest communities, who’d largely voted for Odinga in 2007. In the run-up to the 2017 vote, Kenyatta and Ruto made overtures to swing and opposition areas, convincing local politicians to “defect” from the opposition. Their tactics were far from transparent: Many Kenyans told me they suspect the defectors were bribed, and a report this month by Privacy International revealed a Texas-based company that once also worked for Donald Trump was contracted to manipulate search and social media results in favor of Kenyatta. Still, given that some pre-election polls showed Kenyatta with as much as a four-point lead, irregularities that occurred during the voting process may just have padded his margin of victory.
Now, with trust in Kenya’s voting systems compromised, signs of an impending technological backlash have emerged. In October, in a reaction to the Supreme Court’s annulment of Kenyatta’s victory, his allies in parliament passed amendments to the country’s election law. The amended clauses, which make it more difficult for the judiciary to annul a vote based on discrepancies between electronic and paper results or problems with the electronic transmission system, are currently being challenged in court. If allowed to stand, however, critics say they’d severely undermine the electronic systems’ ability to function as a watchdog, effectively returning the country to a manual tallying system like the one that permitted massive fraud in 2007. “In some ways, this would be like going back to square one,” says Petronella Mukaindo, an attorney at the Kenya National Commission on Human Rights, who monitored the legislation drafting process closely.
Some argue that Kenya should double down on voting technology rather than retreating from it. Bitange Ndemo, a University of Nairobi professor and one of the country’s most prominent tech evangelists, pointed to M-Pesa, a homegrown mobile platform used by almost two-thirds of Kenya’s adult population to transfer money and make payments on their phones. A system that let Kenyans vote on their mobile phones, he argues, could be secured through the use of blockchains, tamper-proof ledgers of transactions behind crypto currencies like bitcoin, which were used in an elections context for the first time last year in Colombia. “We have so much faith in sending money through M-Pesa,” Ndemo tells me. “I don’t see why we cannot work toward having faith in voting from a mobile platform.”
Ndemo’s techno-optimism, however, points to the underlying challenge: Voting innovations, in Kenya and elsewhere, ultimately need to be endorsed by governments. These include incumbent regimes bent on rigging, which may embrace technologies that give elections the appearance of transparency, yet be wary of those that promise to eliminate fraud once and for all. In Kenya, where the last three presidential contests have been subject to dispute, faith in the credibility of elections—no matter what technology is used—may now be beyond repair. Githongo, the anti-corruption activist, already anticipates problems in 2022. “There is no election that can happen that will be clean,” Githongo tells me. “The IEBC is not trusted—they cannot pull off a process that will be accepted by a majority of the Kenyan people.”
For all the questions about Kenya’s now concluded election season, not least regarding the mystery of Msando’s death, one verdict is clear: Trust, perhaps the most fundamental ingredient to a successful electoral process, and by extension, successful democratic governance, cannot be manufactured. Kenya had an opportunity to function as a global proving ground for technological systems that could in theory help creaking democracies elsewhere. Instead, those systems have been tainted by a failure that is largely political. The election was, in a sense, disrupted. But democratic disruption tends to lead less often to creation than to chaos.