Social Proof In Increasingly Digital Environments

Social proof applies profoundly to crowd-sourced information in the digital realm. The “wisdom of the crowd” only works if each person makes an autonomous evaluation. But we are calibrated to seek out the judgments of others and to weight those judgments heavily—this is often nudged by user interfaces.

The Asch conformity experiments demonstrate this tendency with high generalizability. I hypothesize the internet amplifies it. The internet presents us with far more options and easier access to social proof. When the number of options increases, the time we can spend evaluating any single option decreases. As a result, we place greater weight on social proof and more time seeking it out. This accelerates information cascades.

In digital-physical environments, there may also be stronger pressure for social approval due to increased visibility (pseudonymity may negate this). That said, this is complex: the nature of the task, the composition of the crowd, and the interests and competence of the individual all play important roles. I think our intuitions serve us reasonably well when we consciously account for these biases.

However, acting appropriately is harder:

People often say, “I don’t know what to trust on the internet anymore.” But we don’t use the internet less, we use it more. Some heuristic for truth is still required. If I want to start skiing, I can either find skier friends and ask them directly, or I can keep reading online reviews. The latter is far easier. I only have so much time and effort to spend, and I must exceed a certain threshold before it makes sense to go out of my way to consult real people.

Because the internet is so fast and frictionless, that threshold—the point at which I redirect my time and effort to other ways of getting information—may be set too high. As a result, people continue relying on online information and distrust only leads them to rely on it more.

Often, social proof on the internet takes a different form. It appears as an aggregate statistic, such as the number of downloads for a song or the average rating of a product. This contrasts with the physical world, where social proof is typically inferred from observing the actions of one’s peer group. Digital aggregate signals are more susceptible to lock-in effects. If early signals are biased, that bias can be amplified over time. This exacerbates inequality and variance between choices. When systems with these dynamics are run multiple times, they also exhibit greater unpredictability, with outcomes diverging despite identical starting conditions.

When individual decisions are subject to social influence, markets do not simply aggregate pre-existing individual preferences.
— Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market

And this is especially the case in the digital age where our time is spread thin, summarized information is abundant, and trust is in short supply.