Social Media

The algorithms of social media are suspected of contributing to many of our modern social problems and of being poor proxies for natural social interaction—but they are still essential to modern business.

Foremost on the minds of social media executives is how to preserve the quality and utility of business, social, cultural, and political interactions, but the science of how social media algorithms affect the social fabric is poorly developed.

When artificial social environments are constructed in digital space, their rules and algorithms are a proxy or stand-in for the rules that govern social interaction in the real world. Osiris reputation system, for example, is an algorithmic proxy for how people determine who is authoritative and worthy of attention.

We do not assume that a painful direction of technology is inevitable, but rather seek to explore how pain could be avoided by improving the social proxy, especially to identify the qualities of natural social interactions that protect people while helping them to know each other and learn from each other.

We simulate natural social interaction using insights from social science and compare it to multiple social media and social proxy algorithms. We create measures for social values, such as democratic meritocracy and economic growth, and test them against social media social proxy algorithms.

In particular, we test popular crowdsourcing algorithms for their effect on the emerging oligarchy and explore alternatives for a way to protect democracy. Osiris researcher Dr. Duong’s history of award-winning social science policy testing is used at Osiris to test our reputation system.

We want to extend these tests to include measures of Osiris values, such as fairness; i.e., the ability to give all software a chance to get chosen in proportion to its merit. We will exploring creating these tests by combining Kaggle-type verification with crowdsourcing, and we will explore how the reputation system should change at different stages of Osiris growth.

We extend the same model of oligopoly and related dynamics in social networks to tests and measures of the ability of social media algorithms to fill social needs in general, starting with Osiris values. In particular, we seek to demonstrate that algorithmically promoting democracy and meritocracy creates better products.