New communities add variance to the outcomes from social interaction. The probability of a big positive interaction goes up. So does the probability of a big negative one.

Progress seems to come both from increases in this variance and from social systems that trim the lower tail.

Together, they mean that we get the benefits of a big upper tail without the costs of the lower tail.

In the history of physical communities, the most dangerous negative interactions came from infectious disease. It took centuries before clean water and rules about sanitation brought life expectancy in cities back up to rural levels. In the United States, it is only in recent decades that most urban areas have found ways to limit criminal interactions in cities.

The Internet has created new communities with a vastly increased set of possible interactions that humans can undertake. It is now possible for two people located literally anywhere in the world to interact directly with each other.

This community is only beginning to grapple with the problem of trimming the lower tail. It doesn’t help that the Internet was started by academics who thought that trust and concern for reputation would be sufficient to prevent negative interactions. These factors were enough when a handful of people who knew each other were the only users, but it should come as no surprise that trust does not scale to billions of users. The technical potential for social acceptance of anonymity has also undermined the potential of reputation to scale.

In parallel with posts on the city as the unit of analysis (TAG – CUA), I’ll also be posting on the general process of trimming the lower tail (TAG – TLT).

In my TLT posts, I’ll focus primarily on crime, but digress occasionally into related areas such as public health.

This post originally appeared on the NYU Stern Urbanization Project’s blog. To read the original post, click here.