IEBC Wants Tamper-Proof Ballot Papers and 59,352 New Kits

The IEBC has begun procurement of the technology and secure printing that will run the 2027 General Election, advertising two international tenders.

The first covers the Integrated Elections Management System, its hardware and accessories — in practice the KIEMS kits used for biometric registration, voter identification and results transmission.

The second covers ballot papers, tactile ballot folders, the Register of Voters and the statutory election result forms.

The numbers

The commission has told MPs it requires 59,352 kits — 55,393 for polling stations and 3,959 for training. It plans total replacement of the 45,353 kits purchased in 2017 and reuse of the 14,000 purchased in 2022.

Reported tender values have been in the region of Sh7 billion, with an additional Sh9.3 billion for kits pushing election technology spending past Sh10 billion within an overall election budget of around Sh74 billion.

What "tamper-proof" ballot papers means

Secure printing with features that make forgery and substitution detectable: serialisation, security paper, watermarking or equivalent, and controlled print runs with reconciliation between what was printed, what was distributed and what was returned.

Tactile ballot folders are the accessibility component, allowing voters with visual impairment to mark a ballot independently.

The question procurement cannot answer

Who controls the system.

Kenyan media has reported that a bill on control of election technology and servers has been left unresolved as 2027 approaches. That is the question the 2017 presidential petition turned on — not whether votes were miscounted at the stream, but whether the commission could demonstrate the process between the stream and the declaration.

New hardware does not settle it. Custody, audit rights and what a court may inspect are legal questions, and they remain open.

What a campaign should conclude

The same thing as always, and it is entirely within your control.

The primary record of an election result is a signed form at a polling stream, and it exists before any transmission happens. An agent with a phone can photograph it, identified to the stream, before the paper moves.

A campaign with that has an independent tally it can defend regardless of how the transmission performs. A campaign without it is relying entirely on somebody else's system working and somebody else's paperwork being correct.

That is why we built the results feature the way we did — every figure tracing back to the agent and the form it came from, with the streams that have not reported listed rather than hidden. Not as a substitute for the official process, but as the independent record that makes it checkable.

Related: the results chain, form by form.

What goes wrong with election technology, historically

Not usually fraud. Usually failure.

Kenyan elections have seen KIEMS identification failures requiring fallback to manual register searches, transmission delays from areas with poor connectivity, kits arriving late or with degraded batteries, and result portals that lag or display inconsistently.

Each of those is an operational problem and each of them, in a contested election, becomes an integrity allegation — because a system that fails cannot demonstrate that it did not fail dishonestly.

That is the real argument for new hardware, and it is a better argument than any about tamper resistance: a system that works reliably removes the ambiguity that disputes grow in.

It also explains why a campaign's own record of the signed form at the stream matters most exactly when the technology has a bad night.

What a campaign should ask for

Two things, from the commission and from its own party.

The final polling station and stream list for your area, as soon as it is published. That number is your agent requirement and it is the single most important operational figure a campaign has.

Accreditation timelines for agents, because an agent without valid accreditation is not admitted, and discovering that at dawn is too late.