Magazine | “It’s hard to see how you’re being tricked by algorithms”
Over the coming weeks, Folia will be dedicating a special print magazine to the theme of “democracy” “We are increasingly seeing only what the algorithm shows us,” says philosopher, UvA lecturer and former National Philosopher Daan Roovers. She investigated what this means for public debate.
For your thesis, you delved into the invisible processes behind our news. Why did you want to investigate that particular aspect of the public debate?
“When we talk about the public debate, we often talk about the input – the voices that participate – or the output, such as a specific talk show or newspaper article. But the most important step is actually the mediation between the two. We use the ugly word throughput to describe this. The classic example is that of a newsroom: at the NOS, RTL Nieuws or even at Folia, information is constantly filtered and selected throughout the day. For the public, this is a very unclear and opaque process: a kind of black box. But at the same time, this is where public debate is formed. It is, in fact, the engine room of debate.”
And that engine room has undergone considerable renovation recently.
“That’s right. In the past, we trusted that the engine room of the public debate was populated by professionals: journalists who adhere to certain codes, allowing us to entrust the process to them. But in today’s digitalised public sphere, the role of journalism has simply been greatly reduced. This means that the process of filtering and selecting is now largely carried out by algorithms from large tech companies. We are still dealing with that black box, but now often without any human involvement.”
What difference does it make whether that opaque process takes place in an editorial office or in an algorithm?
“Well, if you used to see a large building on Wibautstraat with the Volkskrant logo on it, for example, you at least knew that a few hundred people worked there who decided together what you would read. With this digital machinery, that is even more hidden from view. That is misleading. But more importantly: what happens in a newsroom is, in principle, traceable. You can reconstruct how the journalistic process unfolded: who made which decisions and on what basis. That doesn’t apply to the way algorithms select news.”
Even though people thought that the internet would actually lead to more transparency.
“When the internet emerged, that was indeed the idea. It could certainly have turned out that way, but the way the internet is structured now is actually very opaque. Google, for example, will never reveal how it arrives at its search results. Or even what exactly is being searched for. This has a lot to do with commercialisation. It has enabled such an enormous concentration of power, which means that these black boxes remain closed.”
What are the consequences of this?
“Above all, it means that you often have no idea why you do or do not get to see certain things. I find the story of cartoonist Jip van der Toorn very telling. She posts great critical cartoons on Instagram. But she noticed that when she posted a cartoon about Gaza or showed a Palestinian flag, her reach immediately plummeted. She then only got 20,000 views per post, instead of 150,000. She concluded that this could only be due to shadowbanning: the choice of a platform to either show or not show your content. As a solution, she now has a folder of bikini photos ready to go. When she posts them, she is immediately picked up by the algorithm again. But that is a very good example of a throughput process that you are not aware of yourself: you don’t know why you see something. And vice versa: as a creator, you don’t know why your message reaches many or few people.”
And such an algorithm probably weighs the content differently than a human editor would.
“Exactly. Those algorithms don’t necessarily use what I call ‘epistemic selection’ – selection based on knowledge or truth-finding. Not at all. For example, there is also something called rage bait: algorithms that specifically aim to make you angry, because anger and indignation generate a lot of interaction. That is commercially interesting for such a platform, but it is not necessarily intended to transfer knowledge to you.”
Do you still use social media?
“Less and less. I cancelled my Twitter account a few weeks ago. Now I only really use Bluesky, Mastodon and LinkedIn. And very occasionally I still use Instagram.”
And with all this background knowledge, how do you view your own timeline?
“Like everyone else, I suppose. It’s very difficult to look over your own shoulder all the time to see how you’re being deceived.”
Even after seven years of doctoral research?
“I’m afraid so.”
From 28 April, a new edition of Folia will once again be available in purple distribution racks. The free print magazine will be disposable on all campuses from that date onwards, featuring democracy as its central theme.
