liveBook · Manning

Measurement

As explained in Hidden Influences

The process of assigning values to properties of objects or systems according to defined procedures and reference standards.

To compare cars, you might measure their maximum speed. To describe the weather, temperature alone is not enough: rain, humidity, and changes in pressure matter too. What you measure determines which differences you can see. section 4.1 section 5.1

Measurement assigns values to properties so that objects or systems can be compared. It requires a unit, such as degrees Celsius for temperature, and a baseline against which to interpret the result. Some quantities are verifiable in principle; others, such as a wine’s quality, depend partly on personal judgment. Price may stand in for quality without reliably capturing it. section 4.1

Algorithmic amplification—the extra exposure that content or its creators receive through recommendations—is harder to capture than speed or temperature. User preferences and the different ways people use a platform complicate comparisons. A useful measure should reveal something meaningful about the recommender system while acknowledging those limits. section 5.1 section 5.2

Measuring a complex system calls for choices about:

  • Perspective: Use multiple measures to illuminate different aspects; a single number cannot fully describe amplification. section 5.1
  • Comparability: Consider how a measure treats users with very different habits, from people who rarely post to professional creators. section 5.1
  • Practicality: More metrics can add detail, but too many can make decisions difficult. Computing them also depends on system implementation and technical expertise. section 5.3
  • Timing: A measurement taken every three months can miss daily changes on a platform, including events between measurements. section 5.3

Measurement is also a question of power. If an impact is not measured, it may be easier to claim that it has no impact. Platforms may prioritize business outcomes, so independent researchers and other stakeholders can help define and validate metrics. The goal is not measurement for its own sake, but a clearer basis for questioning platform behavior and demanding transparency. section 5.3 section 5.2

How Hidden Influences covers it

How deep it goes
Explains units, baselines, multiple measures, and practical metric-design trade-offs without specifying a complete amplification formula.
Examples
Compares car speed and wine quality, then uses weather measurements to illustrate why complex phenomena need multiple measures.
What you can do after
Evaluate measurement choices by considering comparability, practical constraints, timing, and whose interests metrics serve.
How it connects
Connects measurement with algorithmic amplification, recommender systems, proxies, and platform transparency.

Also explained in

Compare how 6 books explain measurement →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Hidden Influences. The section links point to the parts of the book it draws on.