Since January 2025, I’ve been utilizing an innovative automated system designed to investigate a lingering question: when detrimental events to democracy occur, is the media adequately covering them, or are we being sidetracked?
This system, known as [The Distraction Index] (https://distractionindex.org), has been operational for over 59 weeks, revealing some notable trends.
How It Functions
The system processes articles from three distinct news sources every four hours, evaluating each political event along two independent criteria:
Score A — Constitutional Damage: This score assesses the level of institutional harm inflicted by an event based on seven key governance factors: judicial independence, press freedom, voting rights, environmental policy, civil liberties, international norms, and fiscal governance.
Score B — Media Distraction: This score gauges the amount of media attention an event garners, including metrics related to hype and strategic manipulation. It considers factors such as the timing of the event in relation to damaging occurrences, coordinated messaging, and distraction patterns.
When significant damage occurs but receives minimal media attention, it is flagged as undercovered. Conversely, when a high-distraction event aligns with a high-damage event that is ignored, it is categorized as a smokescreen.
Insights from 59 Weeks of Data
After examining over 1,500 political events, the findings are compelling:
- 210+ smokescreen pairs identified: Instances where attention-grabbing media events coincided with neglected institutional damage.
- Certain types of harm are consistently underreported, particularly regarding judicial independence and environmental regulatory rollbacks, irrespective of prevailing distractions in the news cycle.
- The disparity between constitutional damage and media coverage has been widening over time.
The Importance of Transparency
Implementation of AI to assess political events brings with it a credibility challenge. Many may assume bias. To counter this, my solution is to ensure everything is auditable.
- All scoring formulas, weightings, and AI prompts are published at [/methodology](https://distractionindex.org/methodology).
- Weekly data snapshots are immutable—once locked, scores cannot be altered quietly.
- The entire codebase is open source: (https://github.com/sgharlow/distraction).
- Corrections made post-freeze are append-only and timestamped.
If there is inherent bias within the system, it can be traced back to specific weightings or framing decisions, allowing for constructive dialogue on potential adjustments—this is the core intent.
The Broader Perspective
In today’s landscape, the overwhelming number of political events makes comprehensive tracking unfeasible. Significant institutional changes can occur while we remain engrossed in the latest tweets or media spectacles. This tool does not dictate opinions; rather, it highlights what might otherwise go unnoticed.
Launched in February 2025, the site has continuously operated since then, with data dating back to January 2025.
Explore it here: (https://distractionindex.org)
View source code: (https://github.com/sgharlow/distraction)
Read the methodology: (https://distractionindex.org/methodology)
I invite feedback from this community—what trends do you observe? Are there any scores you would assess differently?
