An algorithm is a defined procedure for processing inputs and producing an output. Recommendation systems may use past viewing, similarity among users, time of day, device, language, and creator history to rank content. The word algorithm does not name a magical independent actor. People choose the goal, training data, variables, thresholds, and measures of success. Organizations decide whether the system should maximize watch time, purchases, safety, diversity, or another outcome.

Algorithms can reproduce social patterns even without a variable labeled race, gender, or class. Postal code, device type, purchase history, school, and social connections may act as proxies for socially patterned conditions. Historical data can also carry past decisions into a model. A hiring system trained on prior promotions may learn an earlier organization’s unequal pathways. Evidence of unequal output raises a question, but researchers still need to trace data, design, use, and appeal procedures before naming the cause.

Datafication is the conversion of action and experience into records that can be stored, compared, and acted upon. A platform can record pauses, clicks, routes, contacts, purchases, and location. These traces support convenience and personalization, but they can also create new forms of monitoring and classification.

Surveillance is the systematic observation or collection of information about people or groups. Surveillance can be conducted by states, employers, schools, companies, families, and peers. Its meaning depends on power and purpose. A patient monitoring a heart condition and an employer tracking every keystroke both collect data, but the relationship, consent, stakes, and control differ.

Sociologists ask who can see whom, who knows observation is occurring, who can challenge a record, and what consequence follows. A false fraud flag matters more when it blocks rent money than when it changes an advertisement. The same technical error therefore carries unequal social weight across institutions.

Users also produce value. Posting, reviewing products, tagging photos, moderating communities, and training recommendation systems through clicks can be understood as digital labor. Some work is paid, some voluntary, and some woven into ordinary participation. Calling all online activity labor would be too broad. The concept becomes useful when user activity creates measurable value for an organization while control and compensation remain uneven.

Quick review: Open the black box with five questions: What outcome is the system built to produce? Which data enter? Who chose the rule? Who can appeal? Which opportunity or penalty follows?

Watch the lesson connection

Media and Social Media gives you a second explanation of the ideas surrounding this lesson. As you watch, pause when the lesson concept appears and explain how the example fits.

Try the idea yourself

Write one original example, one close nonexample, and one observation that would help you tell them apart. That small exercise turns a definition into a sociological tool you can use in daily life.