What kind of a person does Netflix favourites think I am?

April 16, 2018; Publisher: ; Format:

I look at my algorithm-generated ‘Recommendations for Lizzie’, and I don’t like that person – or the control involved in the process

Each time I scroll down to Netflix’s “Recommendations for Lizzie” feed, my heart sinks in shame. Like a house of black mirrors from a Freudian funfair, it looks like a listicle of romcoms featuring a hapless white lady, with a kind heart and an unfortunate failing, struggling to get it together with the square-jawed-but-equally-flawed male lead. Honestly, it could be a cut-and-paste job from the Wikipedia filmography of Jennifer Lawrence, Jennifer Aniston, Jennifer Lopez or various actors named Ryan.

Netflix, in all its machine-learned wisdom, appears to know me better than myself. More than 80% of shows you watch on Netflix are discovered through its recommendations. Maybe I let my mouse hover over The Notebook for too long, and I am certainly paying the price for clicking on The Christmas Prince (which was, to be fair, a hilarious hot mess both in terms of continuity and constitutionality. As such, I consider my interest to have been at least in part professional and have no regrets). But the personal insight offered by Netflix’s algorithm leaves me cold.

You like romcoms? Good for you, no judgment. But I find this feed of saccharine narratives insidious, especially when these films codify gender roles and relationship goals in an infantilising manner. Sure, occasionally I watch them (to be honest, not actually that often, but whatever), however it is definitely not all I want to see. Watching only my Netflix recommendations would be like using the internet only to look at cat pictures: reasonable on one level, but you would undeniably also miss out on some interesting stuff.

Netflix’s biggest competitor, according to its CEO, is not other streaming services, but sleep. This observation is both sinister and unintentionally revealing about the quality of its content. But I also get it: by the time I am scrolling around looking for something to watch, I am usually seeking what gambling researchers call “the zone” – a state of wellbeing characterised by a sense of relief and an absence of stress.

I just want to tune out, and Netflix has a financial interest in perpetuating that state, uninterrupted, for as long as possible. Escapism is an important component of emotional wellbeing, and thankfully a Netflix binge is less harmful than a gambling addiction. But I would not want Netflix’s bottom line to be a determinant for cultural production and consumption.

Consuming culture should be about delight and surprise; it should be comforting and unsettling; it should be thought-provoking and create space for self-reflection. It need not be all of these things at once, but a balanced diet helps us understand each other across social divides and make sense of the human experience. Just as we should resist outsourcing our ethical decisions to machines, we should not allow them to make cultural ones for us either. Encouraging us to make choices about cultural consumption based on path dependency, as the Netflix algorithm does, is not a neutral phenomenon.

Maciej Cegłowski has written about how by clicking on recommended links in YouTube, you can end up watching videos about conspiracy theories surprisingly quickly. Using machine learning to keep people hooked on a platform can animate our desire to either escape our emotions, or fuel them – these are two sides of the same coin.

Outrage, like escapism, can be healthy and appropriate, but coding our online life in ways that are slavish to these emotional states in subtle ways is hardly ideal. It might make platforms a lot of money, but it also degrades our sense of what is possible, and the commonality of human experiences. “By emphasising the individual to an extreme,” Joseph Turow writes in his book Niche Envy, “the new niche-making forces are encouraging values that diminish the sense of belonging that is necessary to a healthy civic life.”

Machines make mistakes and engender specific values in ways that are hard to detect when we rely on them to make decisions or recommendations. Now more than ever we must embrace the process of defining our values publicly and collectively, to invite scrutiny into machine learning, to find ways to keep these algorithms accountable and make them more transparent. Until then, I’m steering clear of anything featuring a Ryan or a Jennifer, no matter what Netflix recommends.

See original: https://www.theguardian.com/commentisfree/2018/apr/16/netflix-favourites-algorithm-recommendations