Two election tallies will need one clear explanation

Picture a voter opening a results page late at night. Two national totals appear on the screen. One says verified; the other says unverified. They are close, but not identical. A party supporter has already circulated a screenshot of the larger number, cropped away the label and announced victory. The official explanation is several paragraphs long and arrives later. This is a hypothetical scene, but it captures the communication problem any proposal for parallel election tallies would have to solve.

Moses Kuria has proposed displaying an electronic presidential tally alongside one based on hard-copy forms verified at the national tallying centre, according to The Star’s report of 24 October. His stated purpose is comparison: people could examine whether the electronic figures and the subsequently verified documents agree. That is a proposal worth discussing on its merits. It is not, on the evidence reviewed here, an adopted IEBC system, and the distinction should remain clear throughout the debate.

The attractive part is the invitation to compare. Public access to the underlying record can make an institution easier to scrutinise. It can also reduce the mystique around a process that citizens are asked to trust at a moment of intense political pressure. But publishing more numbers does not automatically produce more understanding. The design has to explain what each number represents, what remains outstanding and how a correction changes the record. Otherwise transparency can become a larger supply of material for confusion.

I would begin the debate with the ordinary reader rather than the national scoreboard. What should a person be able to establish about a result from their own area? They should be able to find the relevant document, understand its status and see how it contributes to the displayed total. If something has changed, they should be able to identify the change and the reason. That is a more demanding test than whether a portal looks impressive during a demonstration.

A status label has to do real work. Unverified might mean that a document has arrived but has not yet been checked. It might mean that a number was entered manually, that a document is unreadable or that an inconsistency is unresolved. Those are different situations. A public explanation should distinguish them without forcing every voter to become a technical specialist. The smallest useful set of clear categories would be better than either one vague label or an incomprehensible catalogue of internal processes.

The same applies to completeness. A running total should say what it includes and what it excludes. A number covering only part of the country cannot be understood properly without that information. The page should explain its coverage in a way that remains visible on a phone and survives an ordinary screenshot. A percentage of records received may help, but even that needs a definition. Received, readable and checked should not be allowed to slide into one another merely because the display needs a short caption.

Corrections deserve particular attention. Suppose a transcription error is identified and fixed. The public should be able to see the previous entry, the corrected one and the explanation, with a timestamp. Quietly replacing a figure may leave an old screenshot circulating without any way for readers to reconcile it with the current page. A visible correction record would not prevent every false accusation. It would give journalists, observers and citizens something concrete with which to answer one.

That record must be designed before polling, rather than improvised during a dispute. A system can be technically capable of storing changes while offering no understandable public account of them. The people responsible should demonstrate how a correction appears to a reader, who authorises it and which document supports it. They should also explain how a user can report a possible error and how the institution will respond. An inbox that accepts complaints without returning an answer is not much of a public service.

There is a separate question about legal authority. A displayed total and the formal declaration of a result are not necessarily the same act. Any proposal must be examined against the applicable law and the duties of the relevant officials. This column does not claim to settle those questions. It argues that the public interface should explain them accurately once they are determined. Design should communicate the legal position, rather than create an impression of authority that the displayed information does not possess.

Political parties would have responsibilities too. A party that demands public access should be willing to preserve the qualifications attached to the information it republishes. Cropping out a warning or presenting a partial tally as a final result would undermine the very scrutiny the party says it supports. Newsrooms should adopt the same discipline. Speed is valuable, but speed without a clear account of status can leave audiences more confident and less informed at the same time.

There is no reason this explanation must be written in the language of a procurement specification. It could be tested in plain English, Kiswahili and other appropriate formats with people who have not helped build the system. Ask them what the page tells them, whether they think the result is final and how they would find a correction. If their answers repeatedly differ from the designers’ intentions, the problem is in the design. Blaming users for misunderstanding an election display is a poor substitute for improving it.

Accessibility belongs in that testing. The important information should remain usable with a modest phone, a slow connection or assistive technology. A visually elaborate national map may be useful to some readers while making the underlying records harder for others to reach. A simple searchable list and a clear route to documents may matter more. The test is whether people can inspect the information they need, not whether the launch presentation attracts compliments from people on fast connections.

An outage plan is equally important. If a public page becomes unavailable, the institution should have a previously announced way to communicate the interruption and restore access. An unexplained gap can become a political story before anyone knows its cause. That does not mean every technical failure is evidence of interference. It means that an election system should anticipate the public significance of ordinary failures and prepare a credible method for explaining them without speculation.

The proposal also needs a realistic account of resources and responsibility. Who maintains the display, checks the incoming data, answers questions and publishes corrections? How are those functions separated where necessary? What happens when several problems occur at once? A demonstration with a few sample records cannot answer every operational question. The public should be shown how the system performs under the conditions for which it is intended, including errors, incomplete records and heavy demand.

None of these questions is an argument against access. They are reasons to make access useful. Secrecy can force citizens to depend on competing partisan claims. Poorly explained disclosure can do the same thing by leaving those claims to interpret the official record. A well-designed process gives people both the evidence and enough explanation to assess it. The balance should favour openness while making uncertainty visible, rather than pretending that a live number is complete simply because it updates quickly.

Kuria’s idea should therefore be discussed through a published proposal that can be examined, criticised and tested. Its proponents should show the reader experience, the correction process and the relationship to the authoritative result. The electoral commission should explain its assessment in public if the proposal reaches it formally. A useful debate would focus on whether the design improves scrutiny and understanding, rather than whether supporting or rejecting it proves loyalty to a political camp.

The basic standard is easy to state even if it is difficult to implement. A voter should be able to tell what a figure means, how current it is and where it came from. A journalist should be able to explain why it changed. An observer should be able to trace it back to the relevant record. The institution should be able to defend the process without asking the public to take its competence on faith. Those are practical tests of a transparent system.

Two tallies may help people compare. They may also give partisans two numbers from which to choose. The difference will lie in the evidence, the explanation and the correction record around them. Before the debate becomes another contest over who wants a fair election, let the proposal answer the person looking at that screen late at night: what am I seeing, what is still missing, and what can I reliably conclude? That voter deserves an answer short enough to understand and solid enough to trust.