JP/Aman Rochman

Internet access in Papua has been limited for a “temporary” period these past weeks since student demonstrations erupted in August, followed by riots in which dozens were reportedly killed. Massive demonstrations in Jakarta and other cities also led to rumors that access to Twitter was being limited, which authorities denied. Greater distrust in the government is looming. 

Before that, we learned of the case of Baiq Nuril. Faced with persistent sexual harassment from her boss, the woman decided to record the lewd comments he made to her over the phone. When she shared this recording online with a colleague, her act of self-defense finally led to a sentence of six months in jail with a fine of Rp 500 million (US$35,755) for defamation under the Electronic Information and Transactions (ITE) Law.

This 2008 law is remarkable for its breadth, with its mandate to suppress defamation, hate speech, pornography and other rather vaguely defined types of expression, and fails to adequately distinguish between well-meaning whistleblowers like Nuril and potentially dangerous public figures like rock star turned politician Ahmad Dhani, who was sentenced earlier this year for his viral internet posts against “blasphemers”. Though in July Nuril was granted amnesty, the hundreds of other individuals who have been convicted under the cyber law have not been so fortunate. 

In practice, the law only holds individuals accountable. The platforms these people use to disseminate their speech are left out of the equation, even as those platforms profit from that content.

Consider the earlier case of Florence Sihombing, a student who was convicted under ITE for “defamatory” complaints offending residents of Yogyakarta, that she posted on Path in 2014. Florence’s comments, apparently triggered by a long gas station line, were no different from the day-to-day griping we all engage in. The only thing that distinguished them from a verbal complaint to a friend was the potential virality that electronic communication facilitates. If the government sees it fit to censor legitimate grievances, the least it could do is focus its attention on the actor who enables the ostensible damage — the platform. Yet, Path faced no consequences for the message it circulated.

The government approaches online speech issues with a sledgehammer when a scalpel is more appropriate. When it comes to the troublesome content ITE was designed to remedy, the authorities see just three options: punish individual users, ban the website or platform where the content was posted, or worse, limit the general public’s internet access.

Let’s take a page from Germany, which last year enacted the Network Enforcement Act. Under its terms, if a user issues a formal complaint about an internet post, the platform has 24 hours to determine whether that post violates German law and, if so, to remove it.

Failing to do so, the platform can be fined up to 50 million euro ($55.18 million). The premise of this law is that platforms share responsibility for the content they disseminate and profit from. Left with no choice but to adhere to the regulation, companies like Facebook recruited teams of moderators and developed algorithms that scan for harmful content.

Placing this burden back on the platform is particularly meaningful in a country like Indonesia, where democracy is still not fully entrenched and freedom of speech is not secure. Though not without its downsides, delegating responsibility to platforms has the benefit of allowing the government to step back from litigating every claim of defamation.

Since the likely worst-case scenario would be downgraded from prison time to the removal of a social media post, good faith actors like Nuril and Florence will not be disincentivized from speaking their truth online. Of course, platforms should alert the authorities to truly harmful content like child pornography or incitements to violence, but borderline cases of defamation could be better arbitrated beyond the chilling reach of state power. 

Indonesia currently ranks third worldwide in terms of the number of Facebook and Twitter users. Successful moderation will require machine learning. The major social media platforms have this technology, but unfortunately, these algorithms were “raised” in English. None has developed true facility in Indonesian. Unsurprisingly, these platforms cannot recognize dangerous, culturally specific terms such as halal darahnya (their blood is permitted under Islam to be spilled) or aseng asing (a derogatory term for the ethnic Chinese population). When the artificial intelligence processes these terms, it translates them literally, without the contextual knowledge to understand their true meaning. The challenge is to leverage the platforms to develop machine learning tools that understand the specifics of Indonesian society while keeping them independent of the state’s authority in ways that would threaten freedom of speech.

The best path forward would be the creation of one or more private and independent content moderation companies capable of developing a natural language processing (NLP) machine in Indonesian. These companies could then serve as a neutral arbiter for platforms, users and regulators. By creating a regulatory ecosystem spearheaded by a third party, Indonesia can take the first step in maintaining a digital discourse that is at once safe, free and sovereign.

Under this suggested regime, the state would mandate that social media companies must work with a content moderation company that uses an Indonesian NLP. Under the safe harbor clause, once the company allows for audit, a platform which relies on it can shed their liability. The government may still reserve the right to make specific takedown requests when it comes to emergency issues such as child pornography or incitement to violence. This balances individual rights, the platform’s needs for certainty when they make well-intended attempts to comply with the law, and social order. 

The robust detection algorithms that companies like Facebook use to find offensive content can and should be adapted to the Indonesian context. However, to succeed and earn popular legitimacy, these platforms will need to invest in Indonesia. It is not enough to have a sales and marketing office in Jakarta to push products. Companies must hire Indonesian linguists and engineers to work on the products themselves and to develop them in a way that makes sense for Indonesia’s particular social, legal and cultural circumstances.

__________________

Founder of Sintesa Consulting LLC, with a Ph.D from Columbia University Graduate School of Journalism

Popular

To my horror, it’s just a movie!

RI, Australia strengthen partnership on taxing crypto assets

Pentagon chief congratulates Prabowo on being elected President