In April 2018, while Kenya was still litigating a nullified presidential election, a venture founded by Kenyan students put a school's student government election on a smart contract. Nobody, including the people who wrote it, could see the count.

The Ruling That Framed the Project

In September 2017, Kenya's Supreme Court did something no court on the African continent had done before: it annulled a presidential election. The decision turned not on the ballots themselves but on what happened to the numbers between the polling station and the tallying centre. The court's problem was transmission. The count itself was never really the issue.

Seven months later, four students at the African Leadership Academy in Johannesburg built a voting system in which transmission could not be tampered with, and ran their school's election on it. Three of the four were Kenyan. They had been watching.

Four Founders in Search of a Use Case

The African Blockchain Initiative (ABI) had four founders, all students at the academy: Cyril Michino as chief executive and Amine Soufaih as chief operating officer, both in their second year, and Gibson Munene Njine as chief technology officer and Gerry Migwi as lead software developer, both in their first. Michino, Migwi and Njine are Kenyan. Soufaih is Moroccan.

They had started ABI the previous October as an education venture, and by early 2018 they faced a problem familiar to anyone trying to build something while still in school.

“People told us blockchain sounded promising, and then asked, reasonably, where exactly we thought they were going to use it,” Njine says. “You can explain the technology all day. Nobody is going to let you put it anywhere that matters.”

First, the Crypto-Simulator

Before the election came the Crypto-Simulator. In February 2018, bitcoin had peaked a few weeks earlier near $20,000 and was in the middle of giving most of it back, and across the continent a great many people were being sold a coin by somebody who had understood coins for about a fortnight.

ABI's answer was a simulated market where users traded with virtual money, so they could learn how the asset actually behaved, including what it feels like to be badly wrong, before putting rent into it. Around 400 people registered.

The team also set up six school chapters in universities and high schools, supplied them with weekly teaching material, and ran a course and a blog alongside.

“We were not against the technology, we were interested in it,” Njine says. “What we were against was the idea that the first thing you do with something you don't understand is put money into it. Let people lose fake money for a month. Then they can decide.”

How the Election Ran

The election system was written in Solidity. The academy already ran its student government vote electronically, so the team was replacing a mechanism rather than introducing one.

The smart contract took the eligibility conditions, opened voting at 7:00 a.m. and closed it at 6:00 p.m., and released the result automatically the moment it shut, simultaneously to the candidates, the electoral council, ABI and the entire student body. Once a vote was written it could not be altered. And at no point could anyone read a running total, the four developers included.

“Whenever someone votes, once that vote is in the system it can never be changed,” Michino said at the time. “No person in the middle can change it. Nobody has power over anything.”

Njine is careful about the scale of what that proves. “I wasn't trying to fix an election. I was twenty and it was a school,” he says. “But the question underneath it is real. Can you build a count that nobody has to be trusted to report honestly? That's the actual problem. Not the counting. The reporting.”

Power Removed by Architecture

The most notable consequence was institutional rather than technical. The academy's Electoral Student Council had always administered the vote. After the system ran, it could observe the process but could not affect it. A body that had held a power simply no longer held it, and nobody had to be persuaded to give it up: the architecture removed it.

Soufaih put the ambition plainly in an interview with the academy at the time: if it worked at school scale, the argument for using it at national scale stopped being hypothetical. “No country will really trust the system if it has not been used before,” he said.

The academy published the account under the headline “How African Blockchain Initiative Aims To Transfer Power To The People.” ABI was later rebranded Chaptr Global.

From Johannesburg to Microsoft

Njine left the academy in June 2019 for a fellowship at the Watson Institute in Boulder, Colorado, selected from nearly 400 applicants across more than 60 countries and named that year's first Sistla Tech Fellow. He went on to Babson College in Massachusetts on a full-tuition scholarship, graduated magna cum laude in 2023, and is now a software engineer at Microsoft in Redmond, Washington, on backend platform and cloud infrastructure.

Nothing in the intervening years has changed his view of what the 2018 project demonstrated, or of its limits.

“A school is not a country. The electorate is a few hundred people who all have email addresses and all have phones, and that is most of the hard problem solved before you start. Nobody should read what we did as proof that national elections can run this way,” he says. “What it does show is that the trust problem is a design problem. You can build a system where the question 'do you trust the people counting' doesn't arise, because nobody is counting. That part was never the hard part. The hard part is getting anyone to let you try it.”

Njine spends his working life around AI now, and says the pattern of 2018 has repeated almost exactly.

“Sometimes AI is the answer. Often it's just better automation, and knowing which is the whole job,” he says. “A rule that runs off your accounts. Two systems that finally talk to each other. Half the time that is the fix, and it's cheaper, and you can read exactly what it did. The thing being sold hardest is rarely the thing somebody needs. That was true of crypto and it's true of AI. Somebody has to be willing to say so, and it usually isn't the person selling it.”