Introduction

Facemash was an early social application developed by Mark Zuckerberg in 2003, before Facebook came into existence. It first appeared while he was a sophomore at Harvard University. Facemash allowed users to compare photos of students placed side-by-side, and to vote on who was "hotter." This simple platform utilized an algorithm to rank people based on the choices of the users. Let's delve into how the Facemash algorithm functioned and its social and algorithmic implications.

The Facemash Algorithm Explained

The algorithm used by Facemash bears resemblance to the Elo rating system, which is used in games like chess to calculate the relative skill levels of players. The general concept can be broken down into the following steps:

1. Initialization: Each photo starts with a median number of "points," potentially 1400 to match the initial value commonly used in Elo systems.
2. Comparison and Voting: • Two photos appear side-by-side. • A user votes on which photo represents the "hotter" person.
3. Score Adjustment: • The algorithm adjusts the scores of both images based on the votes. • If the higher-ranked photo wins, both images have a slight adjustment. • If the lower-ranked photo wins, it gains more points while the other loses more.

Using a simplified form of the Elo rating calculation:

$$E\_A = \frac{1}{1 + 10^{(R\_B - R\_A) / 400}}$$

Where: • $E_A$ is the expected score of photo A. • $R_A$ and $R_B$ are the current ratings of photo A and B, respectively.

The resulting new ratings $R'_A$ and $R'_B$ are calculated as:

$$R'\_A = R\_A + K \times (S\_A - E\_A)$$

$$R'\_B = R\_B + K \times (S\_B - E\_B)$$

Here, $S_A$ and $S_B$ are the actual scores (1 for win, 0 for loss, 0.5 for tie), and $K$ is a constant that determines the weight of the result, often set to 32.

A Detailed Walkthrough

Consider two photos A and B: • Initial Ratings: $R_A = 1400$, $R_B = 1400$ • User votes for A.

Calculate expectation: • $E_A = 0.5$, $E_B = 0.5$

After A wins: • $R'_A = 1400 + 32 \times (1 - 0.5) = 1416$ • $R'_B = 1400 + 32 \times (0 - 0.5) = 1384$

The new scores reflect the users' preferences, with A gaining points and B losing points.

Key Implications and Criticisms

Facemash and its algorithm highlight several key points in both computational and ethical realms:

1. Algorithm Simplicity: The simplicity of such ranking algorithms makes them versatile and efficient.
2. Scalability: This type of algorithm can scale as more photos are added, maintaining performance and reliability.
3. Ethical Issues: Despite technical benefits, the ethical concerns in ranking people's appearances are significant. The platform reacted significantly due to privacy concerns and the objectification of individuals.
4. Data Utilization: The platform highlighted early examples of using social data for engagement and interaction maximization, a precursor to more complex social media algorithms.
5. Privacy Concerns: Facemash used images without permission, sparking discussions about privacy and consent, leading to Mark Zuckerberg facing disciplinary action from Harvard.

Summary Table of Key Points

Conclusion

Facemash's algorithm, although simple, played a critical role in the development of more advanced social interaction systems. It serves as a historical footnote in the timeline of social networking technology. Despite being a controversial project, the algorithm's principle has been utilized across various domains, pointing to the versatility but also the potential ethical dilemmas in algorithmic decision-making.