Digital Cities: Social Networks, Who Cares?

9 min readJun 21, 2017

This is the fourth post in a series of excerpts from my graduate research at Cornell University; each has been adapted for the purposes of this format. To read the full report, in all its technical glory, please visit my website.

Previous Topic: Scaling Information within Social Networks

tl;dr

We observe social networks because they are shiny and new, we join them because someone we know advocates us to. We post to preserve as an outlet of creative identity archival, seeking to build upon our interests, or because we feel we must (#addicted).

Social media platforms can forecast our entire involvement on the platform, from how much we post to when we post (up to 6 hours of certainty). They can use this knowledge to make sure our connections like/comment/share our content, scratching that #addicted itch.

If a user who posts bad content gets negative feedback, then they will only increase their posting activity — this follow on activity will only get worse in quality. Counter-intuitive to the assumption we have of moderation-correction of content.

Pick your poison, but in the end does it even really matter?

Here’s the thing, I’m an urbanist specializing in social networks and civic tech product development, and yet I don’t particularly use or find personal value in the social media platforms we’ve come to know and love — what gives? While their origins lay in web 1.0 sites (i.e. The Well, SixDegrees, and Friendster), it wasn’t until the proliferation and dominance of next-generation platforms like Facebook, Instagram, Twitter, and Snapchat have their effects truly begun to shape our built environment. If these platforms are shaping our cultural, historical, and physical identities, then urbanists should not only seek to understand them, but learn to strategize how to leverage their reach in collaborating with our urban populations.

“…why do we join, why do we reveal ourselves, and what are we aiming to get out of social media?”

Questions for Social Media

Previously we established a precedent of status and constructural theory, both of which we’ll continue to use as a frame of reference in examining the human-to-human communication choices we make on social media platforms everyday. As a general guideline we’ll seek to resolve three queries: why do we join, why do we reveal ourselves, and what are we aiming to get our of social media? We’ll need to combine our understanding of social networking theory with the growing field of human-computer interaction (HCI) and information science.

Our primary question, the impetus to join an social network, is tied to two evolving characteristics of the network in question: the number of off-network connections we currently have relative to the network, and how those within the network are already connected. Taken together, these characteristics serve as the foundation to a preconceived trust advantage a network will carry if an outside observer can view known friends already connected with the network.

Snapchat’s Early Days

We’ve all heard the complaints — its so confusing, its just for sexting, its only for teenagers, etc. — and yet, until the fall of last year Snapchat continued to gain increasing traction amongst older generations and corporate entities. By all accounts the principles of HCI theory should have made the application hit the app store dead-on-arrival (and for some time it did). There have been many articles written on how Snapchat’s clumsy UI might have actually encouraged early adopters; however, it would be ignorant to ignore how the application’s trust advantage attracted new users.

Snapchat started gaining traction within LA high schools as a way to chat with their fellow students on school tablets (the application wasn’t blocked and messages would disappear leaving no digital traces). A rather humble beginning for such a large company today; however, if it wasn’t for early users encouraging their friends to use the platform I would be hard pressed to say the application would have sustained any sort of longevity. Perhaps more than any social media platform in the past, the trust advantage of a social network was critical to Snapchat’s rise.

The goal is getting from Early Adopters over the chasm into the Early Majority. Credit goes to Mark McDonald for the clean graphic on this one.

Motivations of Early Adopters

Early adopters of new social networks often display alternative motivations for joining new networks. Whether extrinsic (improvement of skills), intrinsic (leisure, sense of obligation), or archival (digital identity), many will join an infant network seeking to establish a sense of centrality with an ever-expanding physical and digital world.

Showing Our Digital Selves

User studies have shown the archival motivation to be an enduring characteristic throughout the user lifecycle, and the only one that increases over time. Every interaction on a social media platform requires a deposit of a small digital possession of the user; this social data gives the platform owner insights into meaningful events, people and interests of the user. When deployed, this data has been shown to provide users with an artful conduit to stimulate and reinforce self-reflection and reminisce on past life events — a sort of cubist portrait of the user’s own digital identity.

…our social data forms a cubist portrait of our identities…

Within social networks there exists two forms of disclosure; one in which the user self-discloses their digital possessions, and the other which represents the network’s omniscient understanding of its users’ cluster characteristics and digital identity. Much of a user’s self-disclosure is a function of their desired balance of social attention, their preferential broadcast range between nodes of strong and weak ties (good friends vs. acquaintances). Once a user’s social balance is determined, their self-curation of digital possessions can be viewed in relation to their digital identity; this insight enables a platform to forecast the lifecycle of the user, including their disclosure frequency.

A cubist portrait by Picasso — similar to the bits of ourselves social media reconstructs to depict our preferences, its us…just not quite the real us.

The second form of disclosure, the network’s understanding of a user’s digital identity (profile), allows the social media owner to view digital possessions on the network as clues of future user self-disclosure preferences. Modeling these clues demonstrates how users tend to increase network viewing 24 hours before and after posting (self-disclosure). While there are many motivating factors for the post, most of which we’ve broadly discussed, perhaps the most potent is that of enhancing social status within a network cluster.

Users post for future retrieval and expression of a their own digital identity for an audience of one (themselves) to many (friends, family, and the public).

Social Media Can See Your Future

As a user increases viewing 6 hours before a forecasted posting, social media platforms can begin to examine the user’s profile to predict their motivation for posting (archival, intrinsic, or extrinsic). From there network owners can begin deploying behavioral nudges within the target user’s network to draw more attention to that user’s predicted post. Obviously this type of capability represents a sharp detraction from the propagation usually extolled by network owners; at scale their true nature isn’t to provide a portal of communication, but to shorten edge ranges within the network to maximize relationship initiation.

While the existence of such a mechanism within the platform holds its own nefarious connotations, the tool is a product of the aggregate balance of social attention within the affected range of the network. When we post we expect a response and the platform seeks to facilitate those responses for us, both to meet our demands and slow network decay within the cluster. The network can transform delivery of the post mapped against recipient profiles to prioritize sentiment reception in a way that maximizes the resulting interaction amongst the user’s visible network.

Let the audience guide you

Known as the ballot-blind prediction, this process provides insight into a group’s evaluation of the disclosure based purely on their profiles, enabling the network to infer the cluster’s approval/disapproval of the post and projected interaction lifecycle. If audience composition can empower those viewing the network to predict interaction magnitude, then we can further examine relative status within the cluster to determine network priorities in shortening wavelengths between posts and increasing activity amplitude.

Within behavioral psychology this predictive model of audience reception rests on the foundations of operant conditioning; a framework of communicative evaluation where positive community feedback will elicit better future content from an author, while negative feedback will cause the author to refrain from contributing further. This phenomena stems from the negativity effect sourced from polar responses having deeper troughs of response than crests of contribution. However, when applied within social media there is little evidence the operant conditioning framework provides a foundational basis of community interaction evaluation.

Posters of bad content will post more

Research shows authors of negatively reviewed content on social media will increase future contribution frequency (textual quality), of which content will be both of lower quality and more harshly evaluated by the community (community bias). These resulting phenomena present a continuous negative feedback-loop; wherein authors are not driven in a direction beneficial to community-at-large, and negatively evaluated authors reciprocate their reception with equally negative reviews of others. If left undeterred by the network, an accelerated state of rapid network decay will surface.

Users that post bad content and are met with negativity will increase the amount of bad content they post, and it’ll only get worse.

What you say are clues to how long you’ll stay

If the social network is a visual forum (example: Instagram), then evidence suggests falloff of user activity is directly mapped against the user’s tenure on the platform. However, if the network is enabled by textual or audible interaction, then the prediction of a user’s lifecycle becomes more linguistic in nature. Using neural networks and voice-to-text protocols we can forecast a user’s lifecycle and overall activity by their reception to the communication norms of the community.

3 Phases of a User’s Lifecycle
Adolescence: learning the norms of the community (i.e. what to say)
Maturity: understand what/how to communicate but begin to develop a gap between their knowledge and the community’s evolving language
Conservatism: norms of the community are foreign in nature

Rarely can linguistically conservative users reinvigorate communication norm adaptation; thus, arrival in this phase signals an impending departure from the network.

End of Excerpt

Next Topic: Hybrid Placemaking in the City

I welcome your feedback; keep in mind this is only a part of a series in which we’ll fully vet the concepts proposed here. Opinions are my own.