Let's Regulate Social Media

Let us guide you down the path of history, dear readers, and to the late ‘90s and early 2000s when the internet was just beginning its staggering rise.

We were there, and so we remember when there were considerable warnings and hand-wringing about the safety of children online. The concept of children being lured out into the hands of strangers through chatrooms was known, debated, discussed in newspapers, and so on. We recall being on MSN Messenger, and speaking with other teenagers on the other side of the world, though we had no way of knowing if they were who they said they were. Then social media exploded (or should we say spawned from the pits of hell) with the advent of the major platforms such as Facebook and Twitter, and everything immediately ran into the most obvious problem: the best and worst of humanity could now be amplified and projected internationally in a vast, global conversation.

The issue of user-generated content swiftly lead to Section 230 in the US, which at least allowed the nascent American social media sites to exist without fearing being sued for something said by their members. The idea of pre-teens being groomed on these sites was less of a thought back then, as the COPPA legislation effectively made online interaction with kids under 13 incredibly difficult… in theory. Children used the sites anyway and said they were 18.

Rest assured it has not escaped our notice that the ethos of the English-speaking networks were heavily steeped in American culture. They were subject to laws in different countries and did, in fact, follow them somewhat (see the German requirement to ban Nazi imagery, for example), but their being based in the United States did color almost everything they did. The free market, the pursuit of profit, and the absolutism of free speech, were the mantras of the day. The social juggernauts treat their users as a convenient resource, to be exploited as necessary, no matter their age. They care nothing for mental health, or social well-being, or whether their technology materially improves anyone’s life. (Contrast the Chinese walled-garden of social media, where the networks are subject to government control on various forms of content; they value collectivist social harmony above the ability to say whatever you like, and although such control can be criticized, we must admit that at least they are clear about their goals and the reasoning behind them. For now, please allow that much of what we say here is directed at the major US-based networks, as our experience of other networks is almost nil.)

In our opinion, the most troubling development in the last twenty years of social media is the development of algorithmic feeds. It may interest you to know, if you are not as old as us, that the feed of content that you encounter on logging into a social media site used to be sorted in order of most recent, and the content would have been solely from people you were connected to in the network. There were no “suggested posts”, or “you may be interested in”. The user viewed only what they had explicitly indicated they wanted. The launch of algorithmic feeds, in which both the posts shown and the order in which they appear are determined by a complex machine learning system geared towards generating the most engagement possible was bizarre, unsettling, and resisted by many, but ultimately foisted upon us. It is now the default on almost every social media site.

The point of such feeds is simple: a social media network lives and dies by how much time the users spend on it. More time = more opportunity to show advertising/collect data = more money. Machine learning systems are uniquely good at interacting with large user bases and “tuning” the feed as needed to make sure that engagement, whether good or bad, is maximized. This produced a very troubling effect in that inflammatory or controversial content became high-engagement by default. For example, a post claiming that abortions are the work of Satan could become a battleground between those who scoffed at such a notion, and those who wholeheartedly believed it. The merits of the claim are not taken into account by a dumb computer script; the algorithm pushes it to many feeds based on its high engagement signals, and never mind that the post is polarizing and all but guaranteed to make people viciously angry and argumentative with each other.

The algorithmic feeds also favored right-wing content. There is a story here, dear readers, in which we may supply some detail, for we were there back in the early 2000s and we witnessed right-wing behavior specifically in the creationist space. (A brief aside: creationists are those who believe that Earth is only 6,000 years old and the Bible is a literal document, such that Noah’s Ark was real, the Flood happened, the six days of creation happened, and all of modern science surrounding evolution and the geological record is wrong. They are, and always have been, the human equivalent of moldy bread.) We observed that the right-wing were relentless in their opinions and could not be convinced otherwise; as said by Jonathan Swift, “Reasoning will never make a Man correct an ill Opinion, which by Reasoning he never acquired.” Their faith was absolute and unshakeable, and lent them a vicious energy with which to attack their opponents.

The left-wing, by comparison, were argumentative even among each other, and, being mostly scientists, were well used to the push and pull of academic debate. The nature of science to hold theories tentatively, to test hypotheses, to listen to and consider alternative views, did not serve us well against those for whom the debate alone was the equivalent of a holy war upon their religion. It was frustrating, oh so frustrating, to read yet another topic containing the same debunked ideas from a commenter who had been told, repeatedly, that those ideas had been debunked, and yet they still pounded the digital table and insisted that we consider them again. But what we observed in that time still holds true today: reactionary politics held by virtue of irrational belief garners the most attention and even compounds upon itself, because reasonable discourse is boring in comparison.

And thus the feeds began to pull online sentiment to ugly, dark and despicable places. Hateful ideas born from irrational belief, such as white supremacy, surged and allowed networks of anger and hatred to form and become concentrated. The effect was not born from social media (we know well that it existed long beforehand) but social media was the ultimate expression of it due to the companies’ algorithms placing the highest value on the attention and engagement of users. Content that provoked an extreme emotional response was every networks’ stock in trade, and negative content in particular provoked the greatest response.

There are good and sensible scientists doing the work to analyze the effect that social media has on people. Of course, they found that heavy users have an increased level of stress, anxiety and depression. (Worth noting that they also found that social media allowed increased access to support networks; marginalized communities, such as trans people, often find connection and friendship online rather than in person.) Sensitive individuals being over-exposed to the algorithm constantly pumping inflammatory content in front of them is so well-known at this point that there is a word for it: doom-scrolling.

The million dollar question is how this all may be addressed, and (as Tech Enemy is based in Canada) we will begin with C-34, a bill that was recently introduced in the Canadian House of Commons, known as the Safe Social Media Bill. This is one such example of how a government may regulate the detrimental effects of social media.

The text of the Bill is dense, and so we shall refer to the summary for the sake of brevity. We have chosen and highlighted the relevant parts and included our comments after each section. Our primary concern is regarding the limitations imposed on the social media networks; the rest is more straightforwardly regarding the ability of the government to make regulations, and the ability of the people to make complaints, based on the bill.

It imposes the following on operators of regulated services (the networks):

• a duty to act responsibly[…] including by implementing measures that are adequate to mitigate the risk that users will be exposed to harmful content on the services, by providing tools to users to enable them to block other users and flag harmful content[…]
• a duty to act responsibly in respect of the regulated chatbot services that they operate, including by implementing measures that are adequate to mitigate the risk that the chatbot will communicate harmful content to a user[…]
• a duty to protect children[…] by integrating into those services design features that are set out in the regulations, by implementing minimum-age restrictions for accessing pornographic content on regulated services, and[…] by implementing minimum-age restrictions for being able to have an account[…]

For all that this bill is being described as one which primarily protects children online, its provisions are more generally broad and apply to harmful content across the entire sphere of social media. The interesting point about this is that the networks themselves have made some token effort at content moderation over the years, and these problems are well-known and at least somewhat understood. It is a somewhat standard response by a government to a technology that they grasp on a mere surface level.

That it is technically difficult, however, is not a reason to avoid implementing these measures. Machine learning can, in fact, assist in identifying harmful content at scale; BlueSky’s content moderation is known to make heavy use of such. The real issue here is that the networks simply do not want to implement such measures, even when it is technically possible to do so, because it would interfere with their pursuit of profit. Facebook is most well-known in this regard, as it is on record as refusing to implement safety measures to prevent young users from being groomed by sexual predators.

There is much hullabaloo about government overreach, “super-regulation”, and the creation of the Digital Safety Commission of Canada with major powers regarding how users must prove their age online, and to this we say… welcome to the party, pal. The government is wading into this area as a matter of necessity, and the area is so vast and difficult that a separate government body with its own powers and responsibility to handle it is entirely necessary. But characterizing this as an apocalypse scenario in which the government will determine how anyone can interact with the internet is farcical. The internet interprets censorship as damage and routes around it, as we are fond of saying, and it is a much bigger place than the major social media networks.

The bill’s other much-discussed provision, to add age-gating to social media networks, is nonsensical in our opinion. We already have the example of Australia to draw on in this regard; teenagers simply circumvent the age-gate or move to other, non-regulated networks. This should have been obvious from the start, as there is no way to tell who sits in front of a computer (or who holds a smartphone/tablet) without engaging in the most egregious kind of privacy invasion. In any case, let us posit how this may be achieved from a technical perspective.

Consider a site, BaseFook, that now requires age verification in order to allow access to its network. The bill has some information on what is expected for the operator of BaseFook in order to implement this feature, specifically what is considered “adequate”, and this boils down to the following:

• The implementation must actually work
• However they do it, they can’t use the age verification data for any other purpose
• They have to protect that data while they have it, and
• They must destroy it once the age verification is complete.

The bill also specifies that the implementation must comply with any other requirement in the regulations, which is a very government way of covering all their bases, but not especially relevant at this point.

First, the system could require a cross-reference against Canadian ID cards or passports, and scan the birth date. This could and indeed would be defeated almost immediately by some photo-editing unless the system enforces one account per ID, AND integrates with some kind of government database for validation.

Second, the system could take a photo of the user and estimate their age. This could also be defeated in a heartbeat simply by a teen putting someone else in front of the camera when the time comes to validate, or said teen using makeup to make themselves look much older. (We cannot believe that this has not occurred to anyone involved in this silliness.)

Third, the system could use an algorithm to scan the user’s behaviour and determine from machine learning whether the account is likely to belong to a teenager. Considering the sheer variability of human interests, this will produce an astronomical number of false positives, and would likely require a fallback to one of the other methods.

Fourth, and we all but guarantee this will happen, the parents of the teenager will create an account, run through the age verification, and then hand it over to their child with unfettered access. Despite the ubiquity of social media, there are many, many people who do not engage with it and would have no issues giving their teenager access so that they can talk to their friends. That it would also give them access to the more nefarious parts of humanity may not even cross their radar.

All in all, we are… concerned. Not so much about the intrusion of the government into our lives, but about whether their efforts will actually do anything meaningful to the status quo. In our opinion, playing whack-a-mole with “harmful content” is largely a waste of time when the real issue here is the core of the social media network algorithm itself, and the way in which it amplifies divisive content in pursuit of metrics and money. If we were to offer an opinion on how to solve this issue, we would suggest enforcing changes to that algorithm… and we are certain that every major social network would howl like a demented banshee and spend millions trying to stop such legislation from being passed. Anything that threatens their profits cannot be allowed, and, as we all know, their profits are the only thing they care about.

We live in strange times, dear readers. Be sure to pay attention in your own country, stay tech-savvy, and be skeptical if anyone talks about social media restrictions.