Putting the Social Back into Social Media | Mikkeline Thomsen | TEDxLinz

Mikkeline Thomsen opens by describing her 2019 research into threats to digital democracy in Denmark, a country where 91% of citizens use social media and 82% are on Facebook. Using an AI tool capable of detecting verbal attacks, her team analyzed roughly 63 million posts and comments on Danish media and politician Facebook pages. The results were alarming: one in 20 comments contained offensive language or hate speech, the attacks were unevenly distributed targeting specific victim groups, and the most effective growth strategy on Facebook was through verbal attacks — a finding corroborated by whistleblower Frances Haugen. Furthermore, 65% of Danes avoided online debate due to the toxic culture, with just 1.5% of users responsible for 100% of the attacks — a dynamic she calls 'imbalanced freedom of speech.'

However, Thomsen's concurrent fieldwork in digital communities told a very different story. She encountered thriving groups such as an angling group for wheelchair users, a mental health safe space, a bonus grandparents network, and political protest communities. This contradiction prompted her team to look for nuance in the data. They found that verbal attack rates varied dramatically by page type: 6% on politician pages, 4% on media pages, but only 1% in citizen-led groups. Critically, the engagement distribution was the inverse: politicians held 3%, media 7%, and citizen communities a remarkable 90% of all online engagement.

Thomsen then presents what she calls the 'digital civil society' of Denmark: 48,000 active Facebook groups with over 100 members each, 70% of Danes belonging to at least one group, and 450,000 posts and comments published daily — equivalent to over 70,000 letters to the editor. Globally, Meta reports tens of millions of active groups with 1.8 billion monthly active users.

She outlines six reasons why these citizen communities succeed. First, they are membership-based rather than follower-based, giving all members equal access to the floor and agenda-setting. Second, they are built on common ground — shared diagnoses, hobbies, professions, localities, or identities. Third, they host a wide variety of social activities beyond debate, including mourning, organizing, joking, education, and mutual aid — what she calls 'social glue' that makes sensitive discussions easier. Fourth, they are established and led by citizens: Denmark's 48,000 groups are managed by 100,000 volunteer administrators and moderators who collectively contribute the equivalent of 11,000 full-time positions annually. Fifth, these human moderators act as 'bouncers,' enforcing community rules and screening for bots. Sixth, there is 'democratic abundance' — multiple competing groups for the same locality or interest allow members to leave and find alternatives if one group's leadership feels too restrictive.

Thomsen also argues these communities deliver on the original promise of the internet by including marginalized groups, pooling knowledge, connecting like-minded individuals, and creating belonging for those who are isolated — such as the only queer person in a rural area. She acknowledges bad groups exist (roughly 1% of the 48,000 are categorized as antisocial) and that bad advice circulates, but argues the public, visible nature of community boards allows for rapid contradiction.

She closes with a call to action: while this digital civil society thrives in defiance of platforms still optimized for retention and polarization, she challenges her audience to imagine the potential if digital infrastructure were deliberately designed around common ground and community dynamics, taking the lead from the millions of volunteers already investing time and social glue into these spaces.