Don't speak for me: Why Facebook Messenger's suggested 'Conversation Topics' overstep the mark

It’s one thing for Facebook to filter out the volumes of content from my friends and brands and publishers on the platform because there is just too much to see everything. And yes, that leads to a horrific filter bubble meaning: if I were to rely solely on my Facebook feed as my world view, I’d have a completely unreal perception of the culture around me. But for Facebook to start making recommendations of what I start conversations about, it feels like a step so far over the line that I look back and can’t even see the line because of the curvature of the Earth.

Facebook sits on a wealth of data about its audience – and the potential for creating wonderful, clever ways of connecting people together has yet to be realised in full. But this feels crude and invasive.

Equate this to a real world example. If Bram was sat opposite me in the pub and interrupted my conversation to say, “Matthew, I think you should talk to me about my Beyoncé experience,” I would probably wonder what on earth he’s playing at. Even worse, if Joanne suggested I ask Bram about how transcendental Bram’s musical evening with Bey was, I’d be more concerned about why Bram doesn’t feel like he can speak for himself.

This pilot, regardless of the role of Facebook in editorialising my network group, asks even bigger questions about the erosion of the concept of friendship and the weakening of connections within networks, driven by digital platforms. The lack of effort it takes me to ‘be aware’ of what my friends are up to creates an unpleasant taste in my mouth.

If I want to have a conversation with someone, I am genuinely interested in what they’ve been up to and want to engage with them. A semi-automatically generated, ‘Hey {firstname}. How was {event_meta_data}?’ would feel like some sort of response from a call-centre, rather than an honest inquiry from someone I care about. I am not looking for a CRM platform for my friends. Indeed, I already fear that many people have fallen into the trap of being broadcasters, posting status updates and hoping that people react, rather than starting conversations with a question.

Whether this is a cynical ploy to test what types of topics the platform is able to ‘tag’ from user content and then match against what resonates most with people in conversations (in order to better match advertising and insight) or a misplaced but honest attempt at ‘connecting people’, or, indeed, a prelude to its other experimentation into ‘chat rooms’ – I’m just not sure. But this is one test-and-learn which feels like it has really overstepped the mark.

As we read daily about algorithms skewing our world view, bots generating support for political parties and the use of AI for the creation of music and movie trailers, we have to ask ourselves some deep questions. Where should systems and platforms sit on the spectrum ranging from enabling connections for the purpose of sharing ideas, to shaping the conversations, to editorialising our behaviours.

Anyone who knows me is well aware that I’m a huge fan of using data to explore not just optimise. Engineering serendipity is something that seems to be a theme throughout most of my work. But sometimes, I think the networks need to just get out of the way and let people be people, rather than ‘users’ to ‘activate’.

Matthew Knight is head of strategic innovation at Carat. *Bram and Joanne are Matthew's teammates.