Sun, Jan 05, 2020 - Page 6 News List

EDITORIAL: Spotting fake online content

Taiwan has laws prohibiting false news and the dissemination of opinion poll results shortly before elections, but to more effectively prevent unwanted influence, there should be a mechanism to flag posts from fake social media accounts.

A report published on the Liberty Times’ (the Taipei Times’ sister paper) Web site on Wednesday said that President Tsai Ing-wen’s (蔡英文) New Year’s Day speech garnered numerous comments that were critical of her, some offensive.

One commenter wrote: “Look at all these comments critical of the Democratic Progressive Party (DPP) on a ‘green’ news Web site. The writing is on the wall for the DPP.”

However, clicking on the user’s account revealed what appeared to be a fake account. The profile picture was a graphic and the account had few connections, most of which had Chinese names written in simplified characters, while there were political posts of a dubious nature on its timeline.

Most of the accounts that made the negative comments had similar characteristics. Unfortunately, some people engaged with the posts, seemingly unaware that they were most likely from fake accounts.

Facebook — which provides commenting functionality on the Web sites of the Taipei Times and the Liberty Times, as well as most other news Web sites in Taiwan — has made strides in eliminating false news reports. However, algorithms are imperfect and fact-checking is slow.

“Technology companies like to claim that algorithms are free of personal bias, yet they inevitably reflect the subjective decisions of those who designed them, and journalistic integrity is not a priority for engineers,” Nicky Woolf wrote in a Guardian opinion piece on Nov. 29, 2016.

A BBC report on Sept. 14 last year said that three years later, Facebook is still using an approach that relies on posts to be flagged either by an algorithm or a user, and then reviewed by a human.

Those who disseminate false reports are turning to images and videos to get their message across, and while Facebook has plans to stop those methods more effectively — including on Instagram, which it owns — the BBC report said: “Instagram hopes artificial intelligence will enable it to train algorithms to detect false content in the future, reducing the need for relying on fact-checking services and users.”

Unfortunately, there might be a long wait before such a system bears fruit, while some are skeptical that machines will ever make good decisions on their own.

A better approach might be to make algorithms act in a more subtle, but still proactive, manner. Posts by new accounts, those with few friends, or those with questionable timelines could be flagged and an identifier placed over the account’s username that warns people to use caution when engaging them. The identifiers could be removed if the account is verified, or after it has demonstrated its validity by engaging in more normal usage patterns.

The measures could also be applied to accounts that frequently post on Taiwanese pages, as well as Instagram and YouTube accounts created with foreign IP addresses.

Puma Shen (沈伯洋), an assistant professor at National Taipei University’s Graduate School of Criminology, on Dec. 22 said that China is increasingly targeting Taiwanese through YouTube videos uploaded through foreign IP addresses. An identifier next to such channels would alert Taiwanese to exercise caution.

Comments will be moderated. Keep comments relevant to the article. Remarks containing abusive and obscene language, personal attacks of any kind or promotion will be removed and the user banned. Final decision will be at the discretion of the Taipei Times.

TOP top