Only the Date Is Settled

On 26 August 2026, the Independent Electoral and Boundaries Commission filed a gazette notice that did something almost nothing else about Kenya's 2027 general election can do: it became true by force of law rather than by observation. Gazette Notice No. 13771, issued under Article 88(4) of the Constitution, the Elections Act and the Elections (General) Regulations, 2012, sets Tuesday, 10 August 2027, as the date Kenyans next vote for president, Parliament and county government.¹ Everything else this paper examines — who sits on the commission that issued that notice, what machines will capture the vote, whose names are on the register those machines check against, where the constituency and ward lines are drawn, and who gets to audit any of it — is, as of this writing, somewhere between settled, actively contested, and still before a judge.

This paper is a companion to an earlier paper in this series, which argued that Kenya loses the most valuable thing a campaign season produces — a structured record of what doorstep conversations told candidates their constituents needed — the moment the campaign that gathered it dissolves. That paper looked at the constituency level, after a vote. This one looks at the national level, before one. The two do not share subject matter; where they do share a source — Kenya's constitutional count of constituencies and wards, and the Data Protection Act's 2022 record — I cite it again independently rather than leaning on the earlier paper's verification.

I want to be precise about what kind of claim this paper makes, because the subject invites a kind it is not making. "Institutional readiness" sounds like it should resolve to a verdict — ready, or not. This paper does not produce one. It produces something narrower: a dated inventory of what is verifiably in place, in dispute, or simply unknown, across the institutions whose job is to make the 2027 election count. Several entries are unresolved as this is written and will likely have resolved, one way or another, by the time anyone reads it. I have tried to say so plainly rather than guess the outcome.

1. Scope, method, and what would refute this

This paper rests on a different evidentiary base than its companion. The earlier paper argued from direct, repeated observation of campaign operations. This one has no fieldwork in it. It draws on three kinds of source, and I want to be exact about each.

Primary legal and constitutional text: the Constitution (Articles 88 and 89), the Elections Act, the Elections (General) Regulations 2012, and two decisions I read in full or in the closest available form — the Supreme Court's advisory opinion on IEBC's capacity to delimit boundaries without a quorate commission, and the High Court's September 2026 order on voter-registration technology access. Kenya Law's judgment database and the Kenya Gazette both refuse automated retrieval, a limitation this corpus has hit before; where I could not fetch a judgment directly, I used practitioner and press summaries that independently agree on the holding, and I say so at the point of citation.

IEBC's own public statements and gazette notices, as reported contemporaneously by one or usually several independent Kenyan outlets.

Reporting on a named government audit I could not fetch directly: the Auditor-General's special audit of IEBC's KIEMS-kit custody, known to me through two independent outlets that agree on the auditor, the period, and the general finding, but not on every subtotal — a discrepancy I flag rather than resolve by picking the more convenient number.

What this paper does not argue from: no interviews with IEBC, parties, or the ODPC; no independent technical audit of KIEMS or the registration system; no polling; no assessment of security-sector or media-environment readiness, both outside this paper's scope.

What would refute this paper's central claim — that, as of October 2026, Kenya's institutional readiness for 2027 consists of one settled fact and several unresolved disputes rather than a demonstrated state of preparedness — is given in full in §14. The sharpest version: if every open item below resolves well ahead of the election, in the direction that favours readiness, "mostly unresolved" becomes wrong and "resolved later than ideal, but resolved" should replace it. I would want that outcome, and would want it reported if it happens.

2. The one fixed point

Gazette Notice No. 13771 was published in Kenya Gazette Volume CXXVIII, No. 150, on 26 August 2026, following related notices (Nos. 13497–13503) six days earlier in the same volume.¹ It invokes Article 88(4) — the provision establishing IEBC's functions, including conducting and supervising elections — with the Elections Act and its 2012 regulations. From it, IEBC published a full statutory calendar: nominations running through May and June 2027, campaigning ending 7 August 2027, polling on 10 August 2027, and a dense sequence of intermediate deadlines running back through October 2026.²

This is, by design, the most legally determinate part of the process. A general election date gazetted under Article 88(4) does not move without a constitutional-level event — a declared emergency, a successful challenge to the gazettement, or a constitutional amendment. None has happened, and nothing found here suggests one is imminent. Everything in the rest of this paper is institutional capacity running against that fixed clock, not uncertainty about the clock.

3. Two and a half years with nobody home

The commission that issued the 2026 gazette notice did not exist, in the sense of having anyone legally authorised to sit on it, for more than two years immediately before that.

The previous chairperson's and commissioners' six-year, non-renewable terms lapsed in January 2023.³ From then until July 2025, IEBC had no chairperson and no commissioners — only a Secretariat, which cannot exercise the Commission's own powers. A selection panel recommended nominees in May 2025;⁴ Ruto's first gazettement attempt, on 10 June 2025, was itself declared unlawful for violating existing court orders, and had to be redone on 10 July 2025 before the nominees could lawfully take office.⁵ Erastus Edung Ethekon and six commissioners were sworn in on 17 July 2025, at the Supreme Court, before Chief Justice Martha Koome.⁶

Thirty months without anyone empowered to act was not merely an administrative inconvenience. The Banisa parliamentary seat fell vacant on an MP's death, and the Supreme Court upheld the Magarini MP's election nullification — both triggering by-elections the commission had no quorum to conduct.⁷ A petitioner, Adan Mohammed Abdullahi, brought Petition E148 of 2023 alleging the selection panel and Parliamentary Service Commission had deliberately slowed recruitment, with knock-on effects for both by-elections and boundary delimitation.⁸

The clearest statement of what the vacancy disabled came from the Supreme Court itself, almost two months after the vacancy had ended. IEBC's Secretariat — acting alone, since there were no commissioners to instruct it — had filed a reference on 4 July 2024 seeking an advisory opinion on the Commission's capacity to delimit boundaries without quorum.⁹ The Court's answer, delivered 5 September 2025, held that delimitation is reserved strictly to Commissioners under Article 88(4)(c) and cannot be delegated to the Secretariat; that IEBC was constitutionally incapacitated from delimitation and other mandated electoral processes without a quorate Commission; and that extending Article 89's delimitation timelines requires Parliament, not the Commission or courts acting alone.¹⁰ By then the new commissioners had held their seats for seven weeks — the ruling's value was retrospective, not operative — but it is the clearest confirmation that the vacancy was constitutionally disabling for exactly the functions this paper examines.

4. The kits: replacing, losing, and counting them

Kenya's biometric voter-identification and results-transmission system, KIEMS, is the piece of election technology most visible to a voter, and it is where the clearest concrete readiness signal sits.

IEBC owns roughly 59,000 KIEMS kits across two procurement cycles — about 45,352 bought for 2017 and roughly 14,000 for 2022.¹¹ For 2027, the plan is to replace the entire 2017-era fleet, now approaching a decade old, plus acquire roughly 5,000 contingency units, while retaining the 2022-era kits.¹¹ An international tender, IEBC/OIT/01/2026-2027, closed 1 September 2026 with a KES 30 million tender security.¹² Press cost estimates have not converged: KES 7 billion, 9.3 billion, and — in a mid-September retender with tightened specifications — 6.5 billion have each appeared in credible outlets.¹³ I treat these as an order-of-magnitude range, not a settled figure, and log the discrepancy in Appendix B rather than pick one.

The tender is unresolved in a specific, checkable way. The Public Procurement Administrative Review Board dismissed a challenge to three mandatory technical requirements on 3 September 2026.¹⁴ The challenger, Oilmax Ventures Ltd, then sought judicial review; Justice William Musyoka granted leave, suspended further procurement steps, and set a compliance mention for 7 October 2026 — this paper's publication date — with judgment due 16 October 2026.¹⁵ I do not know how that will resolve and have not guessed. It is the cleanest near-term falsification opportunity this paper names: either the ruling clears the single largest 2027 technology procurement with enough runway to manufacture, test and deploy before August 2027, or it does not.

Separate from buying new kits is the question of what happened to the old ones. Auditor-General Nancy Gathungu's special audit of IEBC's KIEMS custody, to 30 June 2024, found the commission could not reliably account for where a meaningful share of its fleet was, against a policy requiring every field-deployed kit's return to a central Nairobi warehouse. Two independent accounts agree on the finding but not the scale: one reports "over 200" kits immediately unaccounted for against an official IEBC figure of two, rising to "about 4,000" cumulatively since 2017;¹⁶ the other gives a breakdown — 159 by-election kits never returned, 30 destroyed in 2022 Wajir tallying-centre violence, 2 officially lost, 158 lost previously, 5 in 2023, 256 held unreturned in Nyeri, and ~3,400 with unupdated register status — that sums to neither headline total.¹⁷ I cannot reconcile these from secondary reporting, and the audit itself was not independently fetchable. Both accounts agree on the cause: internal controls could not be confirmed, and the authority to fix them was itself constrained by the absence of a commissioner to approve one — the §3 vacancy, as a concrete operational cost rather than an abstraction.¹⁶

5. The money

In February 2025, IEBC requested KES 61.74 billion for the 2027 election across three fiscal years.¹⁸ By May 2026, thirteen months out, it had revised that upward to roughly KES 74.8 billion, while reporting the actual FY2026/27 Treasury allocation at only KES 41.3 billion — a reported shortfall of about KES 33.5 billion.¹⁹ I treat both figures as the commission's own stated position, not an independently audited reconciliation; what is verifiable is that MPs have pressed IEBC over the increase without the commission denying the shortfall.¹⁹ A funding gap this close to the election, on the line item already sitting in contested procurement, is a second concrete readiness signal alongside §4's kit-custody question, with its own test in §14.

6. The register: scale, and the fight over what's in it

Kenya ran a substantial registration push in the year before this paper was written. Routine Continuous Voter Registration added 267,249 voters between 29 September 2025 and April 2026; a time-limited Enhanced Continuous Voter Registration (ECVR) exercise, 30 March–28 April 2026, added a further 2,345,476 — a combined 2,612,725.²⁰ ECVR alone deployed 30,615 centres, 12,520 officials and 5,390 KIEMS kits across all 290 constituencies and 1,450 wards.²⁰ After ECVR, registration reverted to the ordinary constituency-office model, which IEBC itself has acknowledged presents access difficulties in marginalised and remote areas.²⁰ The KIEMS tender documentation separately projects a 28.5 million-voter 2027 electorate.¹²

I want to flag, not resolve, an inconsistency in the reported totals. Independent tracking aggregators put the national registered-voter count at approximately 22.1 million in June 2026 and 24.7 million in July — a jump suspiciously close to the entire CVR-plus-ECVR total IEBC reported concluded two months earlier.²¹ I cannot tell from secondary sources whether the July figure is genuinely new registration, delayed accounting of the already-completed exercise, or a reconciliation between two tracking methodologies, and would rather register the discrepancy than pick the cleaner-looking number.

Scale is not accuracy, and accuracy is currently in court. On 18 September 2026, Justice Gregory Mutai ordered IEBC to grant accredited parties and observers access to the source code and audit logs of the continuous voter-registration system, leaning on a KPMG audit finding 481,711 duplicate entries and 246,465 deceased-voter entries.²² I have not established whether this is the same KPMG audit from ahead of 2022 — which independently reported comparable findings at the time — cited again as unresolved, or a freshly commissioned audit reaching similar numbers; the distinction matters, and I flag rather than assume it in Appendix B. The court narrowed the petition to two questions — entitlement to technology-record access, and proper publication of the register — dismissing broader suitability challenges.²² Separately, in early October 2026, IEBC publicly rejected allegations that registration materials and KIEMS kits had moved across the border into Uganda, stating no kits were "misplaced or reported missing, intercepted or unlawfully moved."²³ I take no position on either matter; both are, as of this writing, open.

7. Boundaries: a cap nobody can unlock before 2027

Article 89(2) requires IEBC to review constituency and ward boundaries every eight to twelve years, completed at least twelve months before a general election — the same provision the Supreme Court's 2025 opinion addressed (§3).¹⁰ Two numeric caps sit underneath that cycle: Article 89(1) fixes constituencies at 290, and Section 26 of the County Governments Act, 2012, fixes wards at 1,450 — the figures already established, for a different purpose, in the companion paper.²⁴ IEBC's position, announced January 2026, is that resolving the conflict between public demand for new units and these caps requires Parliament; in the meantime it adopted a phased approach — policy work and data validation before 2027, no disruptive changes in the final twelve months, and the substantive review deferred until after 2027, contingent on validated census data.²⁵

That position is itself being challenged: a Bobasi-constituency boundary-review case has cleared a preliminary hurdle,²⁶ and a separate petition seeks to force a review before 2027 on pain of halting the election.²⁷ Neither had resolved at the time of writing. What matters most is not whether either succeeds, but whether success would land on a timeline incompatible with the twelve-months-before-election rule the Supreme Court identified as binding — why a forced pre-2027 boundary change leads §14's falsification conditions.

8. Data protection since 2022: busier, not broader

The companion paper documented the ODPC's one clear action on a concrete political-data complaint: the 2021 discovery that over 200 Kenyans were registered as party members without consent, met with a public statement and portal consent-mechanism changes, independently confirmed by the Carter Center and Privacy International.²⁸ This paper's question is narrower: has anything changed in that specific area since.

ODPC enforcement generally has intensified — 51 complaints determined in 2024, 96 in 2025, a shift practitioners describe as "foundational compliance" giving way to "structured accountability," including, for the first time, explicit rejection of implied consent and recommendations to prosecute directors for obstruction.²⁹ April 2026 draft guidance covered the transport sector, cross-border transfers, Data Protection Officers, and institutional policy.³⁰ None of it is political-party- or campaign-specific. I checked the ODPC's own 2026 determinations directly: every entry involves an insurance broker, credit company, or similar private firm — none a political or election-related complaint.³¹ I searched separately for any enforcement action against a political actor over campaign data or contact-list provenance since 2022, and found none. The Registrar of Political Parties is developing its own data-protection policy with the ODPC, still in draft.³²

The honest conclusion is continuity, not change: the ODPC has become busier and more consequential in general, and has not, as far as either paper could establish, extended that activity into the political-data domain the companion paper flagged in 2022 — the same negative finding, carried forward rather than re-derived.

9. By-elections as a live signal

Kenya held several by-elections in 2026, each a smaller rehearsal of the 2027 machinery. On 26 February, UDA swept the Isiolo South parliamentary seat (Tubi Mohammed Tubi, 7,352 votes) and three county-assembly seats.³³ On 14 May, by-elections covered Emurua Dikirr constituency and wards in Samburu and Elgeyo-Marakwet.³⁴ On 16 July, Sammy Douglas Waweru Kamau of the Democratic Change Party won Ol Kalou with 41,266 votes against a registered electorate of 73,480.³⁵ I report the winner's tally and registered-voter count as given; I do not report turnout, since available sources give only the winner's count rather than total valid votes cast, and inferring turnout from one candidate's tally would overstate it.

I found no systematic, independently compiled account of KIEMS performance, transmission reliability, or dispute rates across these contests — a gap worth naming rather than filling with an inference. What the by-elections do confirm is that the reconstituted commission has been able to organise multiple contests since regaining quorum in July 2025: a modest, checkable fact after the thirty-month gap in which it could not.

10. A deliberate non-link: the PBO Act litigation

This corpus's existing research flags a precedent — Otieno & 2 others v Attorney General & another; Katiba Institute & 9 others, Petition E519 of 2024, [2025] KEHC 8557 (KLR) — as worth checking here, since the surname also appears attached to 2026 commentary on boundary-review delay. I checked, and the connection does not hold. The Otieno PBO case concerns the Public Benefit Organizations Act, 2013 — a different statute, regulator and petitioner — striking down mandatory NGO re-registration and compulsory donor-data disclosure.³⁶ The 2026 boundary-review commentary is by a different person who happens to share the surname, is an opinion piece rather than a filing, and raises no PBO Act issue.²⁷ Beyond the coincidental name, I found no substantive link. I record the check rather than silently drop it, because verifying applies as much to ruling a connection out as to establishing one.

11. Objections I take seriously

"This is a list of open cases and unresolved numbers; it doesn't answer whether Kenya is ready." Correct, by design: this paper deliberately avoids a composite verdict, because the evidence does not support one. A dated, checkable inventory is more honest, this far out, than a confident summary judgment.

"A busy docket of election litigation is normal institutional friction, not dysfunction — you're treating checks and balances as a red flag." A serious objection, and not wrong on its own terms. The companion paper made a similar point about the ODPC: the 2021 case showed the machinery working because there was something concrete to grip. The same could apply to the Oilmax challenge, the Njoroge case, and the boundary petitions — each a mechanism catching exactly this kind of dispute as intended. This paper's actual claim is narrower: several of these disputes run close enough to the election that resolution timing, not only outcome, now matters. §14 treats timing, not litigation itself, as the risk.

"Citing a thirty-month-old vacancy as a current problem is stale — the commission has functioned for over a year." Fair; I do not claim commissioners are currently absent. I do claim some consequences — the Supreme Court's confirmation of incapacity during the vacancy, the audit finding kit-tracking broke down partly for lack of a commissioner to approve a fix — persist into the current picture. A reader who thinks §4's problem has since been fully remediated should treat that as an open question this paper did not find a public answer to.

"Flux has a commercial interest in portraying Kenyan civic institutions as under-built." This applies more to the companion paper, which names BungeConnect, and to an earlier paper built around it, than to this one. Nothing in this paper's subject — national election technology, the voter register, boundary delimitation — is something BungeConnect sells into or benefits from either way. The narrower, honest version: this paper sits in a series whose broader thesis does serve Flux's interest in aggregate, and a reader entitled to discount the corpus's framing on that basis may discount this paper's too, even though its specific findings touch no Flux product.

12. What follows

No single item in §§2–10 is decisive on its own. A reconstituted commission running by-elections, a tender that may or may not clear court with enough runway, an unreconciled kit-custody record, a register whose accuracy is itself litigated, a contested boundary freeze, and a regulator busier everywhere except the one place a prior paper found a gap — together these describe an institutional picture still being assembled in public, with several pieces due to resolve within months of this paper's publication. What a paper like this can usefully do, ten months out, is not predict the resolution but state precisely what to check and when — which §14 attempts.

13. The claim I would defend

Narrower than a readiness verdict. As of October 2026, exactly one fact about Kenya's institutional preparedness for the 2027 election is settled beyond dispute — the gazetted date of 10 August 2027 — and every other dimension examined here (commission capacity, technology procurement, register accuracy, boundary delimitation, political-data enforcement) is currently either contested in court, administratively unresolved, or not independently verifiable from the public record. That is a claim about the current state of the evidence, not that any of these will stay unresolved — several are scheduled to resolve within weeks of this paper's publication date.

14. What would falsify this

In decreasing order of damage to the argument.

The gazetted date is moved, rescinded, or successfully challenged before polling. The most significant thing this paper could be wrong about.

The Oilmax judicial review (judgment due 16 October 2026) delays KIEMS procurement past the point new kits can plausibly be deployed before August 2027. The clearest near-term test, likely answered before most readers see this paper.

A court forces a boundary change before 2027 on a timeline conflicting with the twelve-months-before-election rule the Supreme Court identified as binding.

The access the Njoroge ruling ordered reveals active manipulation of the register, not the legacy duplicate and deceased-voter entries the KPMG findings describe. That would upgrade §6's "unresolved accuracy question" into something materially more serious.

An independent body reconciles the §4 KIEMS custody numbers, and the reconciled figure is substantially worse than any press figure cited here.

The ODPC takes enforcement action against a political actor for campaign- or voter-data handling before 2027. This would falsify §8's "no action found" claim — and, unlike every other item here, it is one I would be glad to see, since it would close a gap this paper and its companion both treat as a real weakness.

15. Limitations, and what this paper does not claim

This paper does not predict who wins in 2027, the margin, or whether the result will be disputed. It makes no claim about election-related violence risk or the security environment. It does not measure public trust; no polling appears here, and that absence is logged in Appendix B.

It does not claim IEBC, the ODPC, or the judiciary are failing institutions. A reconstituted commission running functioning by-elections, a procurement dispute resolving through ordinary judicial review, a court ordering transparency over a disputed system — these are at least as consistent with institutions functioning as designed as with institutions in crisis, a point §11 engages directly.

It does not resolve the numeric discrepancies in §4 or §6; both are stated as open. And it does not claim any finding here reflects on BungeConnect or any Flux product: nothing in this paper's subject matter is something Flux builds into or has evidence about beyond what is cited.

Appendix A — Evidence table

Sourced rows are cited in the footnotes. Empty rows are measurements this argument would benefit from and does not have.

P = press reporting, mostly corroborated across two or more outlets. J = judgment/advisory opinion, via the fullest available account where Kenya Law's database could not be fetched directly. Rows 13, 15, 16 and 17 would most change this paper's framing if filled; row 15 is §14's sharpest falsification test.

Appendix B — Citation register

CITATION-NEEDED — reconciled KIEMS custody figures (row 7). Two press accounts agree on the finding, not the totals; the audit itself was not independently fetchable.

CITATION-NEEDED — provenance of the Njoroge-case KPMG figures (rows 12–13). Unclear whether these are a fresh 2026 audit or the pre-2022 audit, carried forward.

CITATION-NEEDED — a single confirmed 2027 KIEMS cost (row 6). KES 6.5bn, 7bn and 9.3bn have each appeared in credible reporting; none is identified as final and board-approved.

CITATION-NEEDED — the June–July 2026 registered-voter discrepancy (row 11). A ~2.6 million one-month jump, matching an already-reported total from two months earlier, is unexplained in either source.

CITATION-NEEDED — ODPC enforcement specific to political data, 2022–2026 (row 15). No action found against a political actor for campaign- or contact-list handling since the 2021 case in the companion paper — the same open question, carried forward rather than re-derived.

CITATION-NEEDED — systematic KIEMS performance data across 2026 by-elections (row 16). No independently compiled account located.

CITATION-NEEDED — causal weight of the vacancy on 2027-specific preparedness (row 17). The audit links it to one tracking failure (§4); no broader measurement was found.