Three ways to design social networks that make our lives better

[First published on UX Collective]

We have more data than ever describing our social lives. Digital artefacts, in particular, are modifying our social connections in unprecedented ways. How can we make the best of these phenomena?

[This is the second in a series of three posts, the first describes how technology is changing our social connections, and designers should take responsibility. The third is on sources of social network data.]

This post describes three theories that can guide designers when they have to make decisions that will impact our social connections.

The common-pool resources and social capital sections describe how social network structures can lead to beneficial social outcomes. Costly signalling is a tool for understanding people’s intentions from their social network behaviours. Understanding people’s true intentions can indicate their needs and capacities, and help to allocate resources fairly.

These three strategies are diverse and expansive domains of research; they suggest directions for exploration rather than a fully mapped territory.

Common-pool resources

Törbel, Switzerland. Robert McC. Netting’s book Balancing on an Alp describes how the villagers share pastures and timber harvesting rights without a ‘tragedy of the commons’ occurring.

Nobel winning economist Elinor Ostrom devoted her life to understanding common-pool resources. Much of her work concerned combatting the idea of the ‘tragedy of the commons’, where selfish individuals will inevitably exploit and destroy resources unless those resources are either regulated by the government or made into private property and regulated by market forces. Her research demonstrates a third way is possible. She describes an emergent approach to resource sharing that relies on social norms to regulate it — the ‘common-pool resource’, or commons.

She developed a set of design principles from her empirical research into successful commons. Two of her principles particularly relate to social networks: Clearly defined boundaries and Effective monitoring.

Non-digital example: the village of Törbel, Switzerland has pastures and forests organised as a commons. Legal documents indicate the system goes back to at least 1224. The steep land and low-quality soil mean that farmers need to graze their cows over large areas. Fencing in such large areas would be prohibitively expensive. Instead, the village relies on a clearly defined social network to police land use. Buying property in Törbel does not automatically give the owner a right to graze cows, instead, the village must vote to give rights to new owners. Herders, as they tend their herds, monitor the alp to detect those disobeying the rules. A tight social structure is used to ensure compliance without physical fences.

Digital example: The Arxiv preprint is a ‘publishing commons’, where anyone can submit a scientific paper free of charge. There is no peer review process. The publishing commons would, however, become polluted if there were no clearly defined boundaries. Low-quality submissions from non-experts would swamp the high-quality submissions and devalue the service. As with Törbel, the answer is a clearly defined social boundary. Submitting to Arxiv requires either membership of a recognised research institution or a sponsor with a track record on the site. Endorsers can give negative endorsement and get an author kicked off. Again, social structures ensure compliance.

Further development: Social network models of common-pool resources are being developed. They may suggest refinements that can be made to social networks in order to better support common-pool resources. Ostrom’s Institutional Analysis and Development framework provides a more rigorous approach to the anecdotal sketch given above.

Social capital

Bonding social capital describes the tight social connections that foster trust. Trust, in turn, enables group-based credit programs, which are popular in developing countries. (Image via Technoserve)

Social capital is an enormous and contested domain. However, some ideas from social capital research are well established and translate very directly to designing for social networks. We can think of social capital as the benefit that accrues to an individual through the set of social connections they have. Two network features are often said to give benefits to individuals:

  1. Brokerage, or bridging social capital. When a single person is the only connection between two groups of individuals they have privileged access to information from both groups. This can be highly advantageous.
  2. Closure, or bonding social capital, describes the trust that can occur within a tightly connected group. If you and all your friends know someone, you can lend them money and be confident that they only way they can abscond with your money is to lose all their friends, so you are more likely to trust them.

Non-digital example: One of the most famous studies in social capital is Granovetter’s Strength of Weak Ties, which lends empirical support to the idea of bridging social capital. By surveying 282 people in a Boston suburb, Granovetter showed that job opportunities disproportionately occur through ‘weak ties’ — connections that an individual has but their friends do not. If all your friends know about a job opportunity, likely, one of them will take it before you. Conversely, if you are the only person in your close social group who knows a contact offering a job, you have access to unique information, so you are more likely to get the job.

Bonding social capital matters too. The image shows a group-based credit organisation. Such credit organisations are especially relevant in developing countries where they can help people access funds for large items of expenditure, for example to cover health care costs. They can only occur in high-trust networks (bonding networks), where participants can be reasonably sure debtors will not abscond with the money.

Digital Example: Alex Pentland, a professor at MIT’s Media Lab, has investigated how tight social connections (roughly, bonding capital) and diverse social connections (roughly, bridging capital) correlate with performance on a social stock trading website called eToro. He finds that a balance between the two works best.

Further development: In a research project in Bretton, a parish in the city of Peterborough, we identified eight individuals active in the local community from Twitter data. We invited them to a focus group and discovered that the parish council wanted to translate some literature into Polish for the large immigrant community in the area, but could not afford professional translation. The school was able to volunteer bilingual students for the task. This kind of resource mobilisation can be enabled by improved social network structures guided by social capital theories. This is a nascent area: a recent meta-analysis identified 58 papers linking social capital and digital social network activity. A follow up post will consider how better data will improve research in this area.

Costly signalling

Springbok ‘stotting’, performatively leaping high into the air, possibly to signal to predators that they are fit and in good health — so will be hard to chase down. (Credit)

Costly signalling differs from the two areas of research described above. Rather than suggesting that a particular social network structure will lead to a particular outcome, costly signalling describes a way of inferring intentions from network behaviour.

If you want to know people’s true intentions, as opposed to what they want you to believe, look at actions which are difficult to fake — costly signals (sometimes called honest signals).

Non-digital example: Peacock’s tails are a classic example of costly signalling. Peacocks want to convince peahens that they have the best genetic material. Having an enormous tail shows that the peacock can survive despite spending calories growing spectacular feathers, an encumbrance that also makes it harder to escape predators. A peacock with lower quality genetic material might try and fake the tail to woo peahens, but a massive and vibrant tail is a hard thing to fake, which is what makes the peacock’s tail a good signalling strategy. Deer’s huge antlers function in the same way.

Digital Example: Twitter allows at least two distinct kinds of interactions between users. Users can follow one another, and they can also “@ mention” one another. Following someone is cheap, you just click the follow button. Sustained @ mention behaviour is expensive, you have to spend time typing out messages. In my research I found that @ mentioning behaviour between Twitter accounts that discuss local issues reflected local resident’s perceptions of the community. That is, the most mentioned Twitter accounts belonged to people or organisations that were considered the most recognisable, even when asking locals who were not on Twitter. Less mentioned Twitter accounts belonged to people or organisations that were less recognisable to locals, again regardless of whether the person was themselves on Twitter or not.

If we look at the data, most Twitter users follow many more people than they message, perhaps because of the effort involved. The idea here is that the more costly activity of messaging will reflect more accurately reflect the way they experience their community. The wider theory of signalling on digital social networks is developed by Judith Donath, currently a fellow at Harvard.

Further development: Costly signalling is a valuable tool because it ties into economic theories of understanding behaviour — the idea of revealed preference. Cory Doctorow describes the value of this idea in his post on platform socialism. In short — the better we understand individual’s needs and capacities, the better we can share resources.


A third and final post will discuss the politics of social media data.