When computers count the votes in Ontario’s municipal elections, will they do so accurately? It’s possible that they do, but we do not have adequate assurances that this is the case. This is a serious problem that casts doubt on the democratic process.
Imagine for a moment that we handed the counting of all votes to one individual in each local election. Imagine further that no other election official ever saw a ballot that he counted, nor could they. He simply announced the result, and everyone was supposed to trust him. That approach would leave massive doubt as to the competence and integrity of the process.
A computer counting or registering the votes does not eliminate similar shortcomings. True, a computer is not swayed by political agendas or having a “horse in the race.” However, it is programmed by people who are. Questions of accuracy and integrity don’t disappear simply because technology is implemented. We should not blindly trust a hardware or software provider any more than we would a human being.
First, these machines do err sometimes. Examples were laid out in the preprint version of a paper published last month entitled, “It Doesn’t Take a Thief: Optical-Scan Voting Systems Fail Even Without Adversaries.” It was written by Philip B. Stark of University of California, Berkeley, and Aleksander Essex, Associate Professor of Software Engineering at Western University in London, Ontario.
As the paper explains, vote tabulators have had “many documented failures…ranging from unexpected ‘inputs’ such as fold-lines in ballots being incorrectly interpreted as voter marks… to procedural failures such as including test ballots in the election results, to ballot printers that are inadequate for the paper stock used, to … humid air at polling places causing paper jams.”
Unfortunately, some election officials can’t grasp the implications. “We have even encountered institutional settings in which the very possibility of tabulator error has not been fully internalized,” the authors explain.
Such problems are innocent in their motivation, but others include “adversarial threats to voting systems, including demonstrated attacks on memory cards.”
Other possibilities not mentioned in the paper are hypothetical but possible, that machines could be tilted wrongly in their programming or their results compromised by hacking. Given that power and political agendas are at stake in election results, should we think it strange if attempts to do so took place?
This is not to say that malfeasance is happening in Ontario. The real problem is that we cannot say with certainty that it is not happening, either. That’s because the kinds of checks to verify computer-tabulated votes are not being done. It takes a court order to trigger a complete verification of such a vote by an examination of paper ballots.
A more plausible option is a risk-limiting audit where a statistically-significant sample of the votes are verified to ensure they were counted correctly. Experts like Essex have recommended that risk-limiting audits be a routine part of computer-counted voting. So has Election Ontario’s Advisory Committee on Standards for Voting Technology. Yet, that doesn’t take place in Ontario’s provincial or municipal elections, in contrast to many jurisdictions in the United States and other countries.
Computerized voting where votes are registered online or at a machine without paper ballots present even greater challenges for accuracy verification. Unfortunately, a slim majority of Ontario municipalities used completely computerized voting with no paper ballots in 2022, and almost 70 percent will use this approach in 2026.
The way these systems are used in Ontario cannot instill confidence. Essex and two co-authors examined a different aspect in their paper “Verifiability Experiences in 2022 Online Elections.” They found that only nine per cent of the 222 Ontario municipalities that used online voting offered any mechanism by which a person could check that their vote was counted correctly.
Essex co-authored a paper in 2023 “Online Voting in Ontario Municipalities: A Standards-Based Review” that examined whether Ontario municipalities measured up to the Council of Europe’s standards for electronic voting. Only 14 per cent of the 49 standards and 93 implementation guidelines were fully met. Let’s hope they do better this year.
Some losing candidates from various municipalities have approached Essex over the years, asking if the computer voting is reliable. They could not reconcile their pre-election polling results with what happened on election day. Were proper verification systems in place, both candidates and voters would have more assurance. But we don’t.
Until Ontario is willing to implement robust and mandatory verification of computer voting processes, it should do what the province of Alberta did: ban computerized voting and vote tabulators from its municipal elections.
