Policy & Accountability

Evidence for what audiences are exposed to.

Platform accountability requires data on what users actually see. Lurk provides structured, reproducible evidence of algorithmic content distribution.

What's broken

No citable methodology

Researchers and policymakers know algorithmic feeds shape public discourse. But there's no standard method for documenting what specific audiences actually see.

Can't compare demographics

"Different people see different things" is obvious. What's missing is structured data showing exactly how content distribution differs across age, location, and interest.

No longitudinal data

Algorithmic feeds shift constantly. A single snapshot tells you what someone saw today — but you need time-series data to show trends, amplification, and suppression.

What Lurk unlocks

Transparent Methodology

Every persona is defined. Every session is logged. Every data point is traceable. The methodology is designed to be published, cited, and scrutinized.

Demographic Comparison

Run identical time windows across different personas and see exactly how the algorithm distributes content differently by age, location, interest, and political lean.

Persona Scale

Build hundreds of synthetic personas to model diverse populations. Capture feeds at scale to move from anecdote to statistical evidence of algorithmic behavior.

Example: Investigating youth exposure to health misinformation

A research team is studying what health-related content the algorithm serves to teenagers versus adults. Anecdotally, teens report seeing more extreme diet content — but there's no structured data to quantify it.

Using Lurk, they build 20 personas across four age brackets and run week-long training sessions. The Demographic Comparison shows teen personas see 4x more restrictive-diet content than adult personas — and the ratio increases over time as the algorithm reinforces engagement patterns.

The Transparent Methodology section of their paper details exact persona definitions, training protocols, and data capture specs. Reviewers can replicate the study with the same persona parameters.

The finding moves from "teens say they see a lot of diet content" to "the algorithm serves teen personas 4x more restrictive-diet videos than identical adult personas, escalating over a 7-day window."

See what they see.

We only work with aligned organizations.
To gain access, please use your organization email and we'll be in touch.