Following on from the sudden explosion of social networking tools in the late 00’s – a moment in time when Facebook, Twitter and MySpace all seemed shiny, new and wonderful – researchers began to examine what they could learn from these collections of self-selected connections. People connected through social media because they shared some sort of affiliation: blood or friendship or career or shared interest or political persuasion or kink or…well, humans connected in communities almost infinite in their diversity. We had diversity – lots of communities – and concentration – folks with very similar interests/beliefs/heritage gathered together. By virtue of being gathered together, they could recognise one another, share with one another – and reinforce each other’s beliefs and actions.
At Harvard University, Dr. Nicholas Christakis took a deep look at these networks of affiliation and quickly realised that some things we thought of as being separate from social connection, were, in fact, intimately related.
We had diversity, and concentration.
In 2008, he co-authored a paper that laid out the first of these findings. “The collective dynamics of smoking in a large social network” showed that social networks influence people’s relationship to tobacco: if a lot of friends in your social network smoke, you’re more likely to smoke — if a large number of friends in your social network decide to quit, you’re more likely to quit. Although this seems almost obvious, this finding represented the first of an ever-broadening set of correlations. Obesity and the desire to diet were also correlated by social connections: if you went on a diet you were more likely to inspire your friends to do so. And perhaps most controversially, it was also shown that divorce also seemed to spread via social connections: if your friends divorced, it was more likely you would also divorce.
These social connections also highlighted personal qualities that an individual might never state publicly – deeply personal matters such as sexual orientation, gender identity, or political alignment. These would become visible simply by looking at the first-order connections of an individual. A person who might be among the LGBTIQ+ would be revealed in the nature of their connections to others, despite any attempts to keep it private. Connections speak louder than words, and reveal more about ourselves than we might ever admit – even to ourselves.
These social connections also highlighted personal qualities that an individual might never state publicly.
By 2010 it became clear to me that Facebook knew more about its hundreds of millions of users than they knew about themselves — and wasn’t sharing what they knew. That asymmetry of information made me feel very uncomfortable: I never knew how what Facebook knew about me and my hundreds of first-order connections was being used, and there seemed no chance that this knowledge would ever be shared. So, at the end of May in 2010, I quit Facebook and deleted my profile.
At the time most folks couldn’t see the logic of my decision. I pointed to the research coming out of Christakis’ lab at Harvard as the tip of a very big iceberg. Beneath the waterline, our profiles were becoming more and more accurate. Facebook’s precise knowledge of us meant it knew more about us than we knew about ourselves and I wasn’t about to feed the social network even more personal data.
More on networks and influence: Under the (social) influence
The better part of a decade later people are more aware of how Facebook use their profile information to influence them. That came partly on the other side of the Cambridge Analytica scandal – when a political advertising firm used that profile information to precisely spread political disinformation where it could swing the most votes.
Harsh publicity also forced Facebook to open its personal data collection to researchers who found themselves embarrassed by riches of deep data gathered over a decade of the social media firm’s continuous, detailed surveillance of its users.
Beneath the waterline, our profiles were becoming more and more accurate.
One of the first fruits of that data sharing came to light in an article published in Nature in August 2022. “Social Capital and Economic Mobility” looks at the relationship between the rich and the poor in the United States, coming to the conclusion that economic mobility is strongly influenced by social connectivity.
It’s not just smoking behaviors, dieting and divorce that spread through our social connections – they also appear to be strong determinants of our economic success. The more connections a ‘poor’ individual has to ‘rich’ individuals, the more likely they are to have upward economic mobility. Conversely, ‘poor’ individuals with fewer connections to ‘rich’ individuals have generally dimmer economic prospects over the course of their lives. The researchers put all of their work into a beautifully visualised website: ‘The Social Capital Atlas’ illustrating how this connectivity affects socioeconomic outcomes throughout the whole of the United States.
Our social connections appear to be strong determinants of our economic success.
This research provides a strong argument for fostering connections between people of differing economic circumstances; the way out of poverty is, at least in part, a ladder constructed by connections to the more affluent. It makes sense – but can this research become the bedrock of policy and of social practice? We seem to have the power to lift others out of poverty – all we need do is connect. But will we?
If we will not connect to help others who will benefit from those connections, there may be another way. Researchers, again writing in Nature, reported their efforts in developing a ‘Democratic AI,’ a bit of artificial intelligence software to oversee the distribution of economic resources in a fair manner. The researchers created an investment ‘game’ with a large number of human players, and let those players select amongst a range of options for what to do with their ‘gains’. Players could choose to keep those gains, or contribute to range of collective funds – where their gains would be redistributed. One of those redistribution mechanisms was designed by artificial intelligence which, through interactions with the human players, continuously evolved and improved its operation, until it won over the majority of human players, who preferred its ‘Democratic AI’ as the most equitable.
We seem to have the power to lift others out of poverty – but will we?
Does this mean we’ll be seeing computers running the tax office, calculating how to redistribute our incomes for greatest social benefit? That’s a political question. It feels as though we’re far from ready to accept any such form of technocratic management of the body politic, however benign and well-intentioned. Yet it’s far easier to imagine that same Democratic AI wading into the thicket of our social networks, helping the poor to make the connections they need for economic mobility. That form of equity would be nearly invisible – and almost as effective. It’s something that Facebook (now renamed Meta) could implement – behind the scenes – as a way to make amends for a decade of abuse. It wouldn’t fix everything, but it would help make our world more equal.