Do not let algorithms decide the elections

Local elections are at the end of this month, but I have already cast my vote on my phone.

My feed on the social media app Threads has been filled with posts tearing a certain mayor’s entire governing record to shreds. In my mother’s Line group chat, that same mayor is hailed as a flawless hero. My mother and I live in the same place, yet we inhabit parallel worlds. This is not a political divide, but an algorithmic one. The battlegrounds in this year’s elections are not campaign rallies; they are the smartphones in our hands.

As local elections involve candidates that many people might not be familiar with, voters rely heavily on the Internet to stay informed. They turn to Line groups and Facebook, precisely the places where the echo chamber effect is most pronounced.

If you read or search for posts about air pollution, these sites feed you more news about air pollution. For urban development, it might push posts about land justice. Over time, a person’s social media feed could become focused on the issues and opinions the user trained it to find.

Social media have made it easier than ever to target specific groups. The now defunct Cambridge Analytica made headlines for accessing the Facebook data of 87 million users, which it allegedly harnessed to create microtargeted advertising during the 2016 US presidential election.

In Taiwan, a city council candidate could spend a few thousand New Taiwan dollars to target a specific group of voters, potentially with misleading information about a rival. This is not a campaign tactic; it is psychological warfare. Any person could be the victim of paid propaganda, which they then repeat online.

Then there is “deepfake” technology. People used to worry that deepfakes would target the president or a celebrity, but now, even village wardens and city councilors are potential targets.

Imagine a photograph of a candidate accepting a bribe or an audio recording of them saying they would sell off public land suddenly circulating in group chats. Before anyone can even verify its authenticity, voters would already be headed to the polls. Local fact-checking resources are weaker than those at the national level, yet the damage from misinformation is far more direct.

Then nine-in-one elections affect everything from the roads outside our homes to the schools our children attend. We cannot allow algorithms to dictate their outcome.

The Central Election Commission and the Ministry of Digital Affairs must establish lines of defense.

First, require major international platforms to establish a transparency database that discloses advertisers of political content and ad-targeting parameters.

Second, establish an emergency takedown and labeling mechanism for fabricated or manipulated content, requiring platforms to label reported “deepfake” videos and suspend their recommendation within 24 hours in the week before the election.

Third, city and county governments should send text messages to voters debunking disinformation.

Social media can encourage people to take an interest in local governance, but it can also turn them against each another. When we cast our ballots later this month, we should vote for those who serve their communities; we should not cast our ballots out of fear or prejudice fed to us on social media. Do not let algorithms decide who the next mayor is be before we do.