The Count Needs Witnesses

Canada’s September 20, 2021 federal election offers a concrete way to separate those questions. This is a retrospective account, drawing on Elections Canada’s election archive and its March 2022 review of voting services in three First Nations communities. The later findings should not be mistaken for information available on election night.

The Missing Input

In Cat Lake, Poplar Hill and Pikangikum, three fly-in communities in the riding of Kenora, local election-day polling places had been planned and advertised on voter information cards. According to Elections Canada’s subsequent review, those polling places were cancelled ahead of election day.

The returning officer had learned of a conflict with hunting and cultural activities and replaced election-day polling with a single day of advance polling, based on his understanding of local needs. The review says the change was not communicated clearly to electors or community leadership. Headquarters learned about it on election day, too late to act.

This was not simply an inconvenience inferred by an outside critic. Elections Canada concluded that some electors could not vote because of communication breakdowns, identified gaps in electoral services and apologized.

The distinction matters. An attempt to accommodate community activities became a decision that some affected people did not clearly hear about. Local judgment and administrative flexibility can be useful. But an accommodation cannot be judged solely by the intention of the official arranging it. It must also be tested against the people expected to use it.

That is the power mechanism here: an administrative choice changed when participation was possible. Publishing an accurate total afterward could not restore the missed opportunity.

What the Record Can Show

Elections Canada’s archive also demonstrates the value of making a count inspectable. It links official voting results and a raw-data version for researchers. The agency explains that the official package adds context to the validated results, including updated registered-elector data, candidate demographics and poll-by-poll results, rather than changing the results themselves.

That detail gives readers more to examine than a winning tally. The archive also links an external audit report and describes its statutory purpose: checking whether election officers properly exercised specified powers and performed specified duties. That is a defined scrutiny function, not a universal certificate that every aspect of an election worked.

The archive’s publication of the Kenora review is important for the same reason. A public record should expose a failure, not require the institution to appear flawless before the record can be trusted.

But disclosure and repair remain different acts. The review establishes an acknowledged failure and a need to reconsider service delivery; it does not, by itself, establish that later changes solved the problem.

The coherenceist test is practical: can the people affected by a system make their experience count in its evaluation? Here, that means reading the account of missed participation alongside the voting totals, without treating either as a substitute for the other. Nothing in these sources establishes that the national result was invalid.

The lesson is narrower—and harder to evade. Keep the count inspectable. Keep the account of exclusion inspectable too. Otherwise, the next impeccably presented total may once again tell us less than we think.