Is More Always Merrier? The Hidden Privacy Cost of Your Multi-Platform Social Life

TL;DR

Think your private life is safe because you only share a "little bit" on Facebook, a "little bit" on LinkedIn, and a "little bit" on Twitter? Think again. This research proves that by aggregating these snippets, a third party can reconstruct a highly granular identity. Using data from 35,000+ users, the authors show that cross-platform aggregation can increase the success of identifying your home address and phone number by up to 500%.

Background: The Illusion of Siloed Privacy

In the early 2010s, the "social footprint" emerged as a critical metric for digital privacy. However, the fundamental question remained: Is a user with five OSN profiles five times as exposed, or is the exposure exponential? This paper positions itself as a deep dive into the complementarity and consistency of our digital twins.

The Core Motivation: Why Aggregation Matters

The researchers identified a psychological gap: users often provide different details to different platforms based on the platform's purpose (e.g., job history on LinkedIn vs. favorite movies on Facebook).

• Prior Work mostly counted the number of attributes.
• This Work explores the quality and overlap of those attributes, asking if patterns of "what we hide" and "what we show" are consistent enough to serve as a secondary fingerprint.

Methodology: Quantifying the Footprint

The authors crawled 179,188 profiles across 10 OSNs (Google, Facebook, Twitter, LinkedIn, etc.). They moved beyond simple counting by introducing two key metrics:

1. Complementarity (Jaccard Distance): Measuring how much new info is gained by adding a second profile.
2. Consistency (Sokal & Michener): Measuring how reliably a user follows their own "privacy policy" across sites.

Figure 1: Comparison of available vs. actually populated attributes across different OSNs.

The study reveals a startling "Behavioral Correlation": Users who own more profiles are inherently more inclined to share more information on each individual profile.

Experimental Results: The Linking Attack

The most chilling part of the paper is the "Australian Phonebook Case Study." By taking location data (often just a city name) from multiple profiles, the researchers could narrow down a search in a public residential database.

Figure 2: How aggregating profiles dramatically increases location granularity.

Key Findings:

• Identity Unmasking: Using aggregated OSN data made it 5x easier to uniquely identify a person’s residential address and phone number compared to using a single source.
• Value Consistency: 85% of users were found to have over 50% matching values across their profiles, meaning we rarely lie—we just provide different pieces of the same truth.
• Occupation Bias: People in customer-facing roles (Real Estate, Marketing) have significantly larger social footprints than educators or students, making them prime targets for social engineering.

Figure 3: Distribution of attribute value consistency; note the high peak around 0.6-0.8.

Critical Insight & Summary

The paper concludes that we are more predictable than we think. We have "preferred patterns" for revealing information. This consistency is a goldmine for attackers.

Takeaways for the Industry:

• For Users: Using pseudonyms on "fun" OSNs (like YouTube) significantly reduces your footprint compared to using your real name.
• For Platforms: OSNs should alert users when their "aggregate" profile across the web becomes too identifiable.
• Limitations: The study is historical (2012) and biased toward users who have Google profiles, but the mathematical logic of "cross-linking attacks" remains a foundational threat in the age of AI-driven OSINT.

The "Social Footprint" isn't just a list of what you've shared—it's a puzzle that, once assembled, reveals the parts of your life you never intended to put online.