I've been thinking about how democracies can defend against coordinated disinformation campaigns without creating new problems worse than the original one.

The core issue: Bot armies operate at internet speed. Traditional institutions are too slow. But we can't use the "obvious" solutions:

*AI alone?* Black-box decisions nobody trusts, and for good reason.

*Government control?* "Ministry of Truth" is the authoritarian playbook.

*Tech platforms?* Zero democratic accountability, profit motives over public interest.

*The result:* We're stuck. Each option has legitimate problems, so we end up with no solution at all. Meanwhile, coordinated bot campaigns are measurable and observable - we can literally watch the network graphs.

*Current EU proposals include mandatory digital ID verification for social media and weakening encryption. These kill anonymity/privacy or create massive bureaucratic overhead. There has to be a middle path.*

## The Proposal: AI Detection + Random Citizen Panels

*How it works:*

1. *AI does pattern detection* - Coordinated posting behavior (10k accounts, similar content, suspicious timing) - Network anomalies (new accounts all interacting only with each other) - Cross-platform coordination - Unnatural amplification patterns

2. *Random citizens review evidence* (like jury duty) - Shown network graphs, posting patterns, account metadata - Simple question: "Does this look like coordinated inauthentic behavior?" - Vote yes/no, majority rule

3. *Temporary quarantine if flagged* - 48-hour distribution pause - Transparent logging of decision + evidence - Appeals process with independent review - Auto-expires unless extended

*Key structural elements:* - Independent body (not government-controlled) - 3-6 month rotation (prevents capture) - Judges behavior patterns, not content truth - Temporary actions, not bans - Public logging of all decisions

## Why This Structure?

*Democratic legitimacy:* If regular citizens - randomly selected, rotating frequently - make the decisions, you solve the trust problem. Not faceless algorithms, not government diktat, not corporate interests.

*Speed:* AI handles scale, humans provide democratic check.

*Proportional:* Targets coordinated manipulation, not individual speech.

*Preserves privacy:* No mandatory identity verification, no killing anonymity.

## The AI's Role:

Good at: Network analysis, pattern detection, temporal correlation

NOT doing: Judging truth, making final decisions, operating autonomously

## Obvious Problems:

- Legitimate activism can look coordinated - False positives during breaking news - Who decides AI training parameters? - Corporate resistance to implementation - Resource costs - Mission creep risk

## The Question:

Is this better than the status quo (platforms deciding opaquely + bot armies unchecked)? Better than mandatory identity verification or weakened encryption?

What am I missing? How would you improve it?

Particularly interested in: - Technical feasibility of detection - Better safeguards against false positives - Distinguishing authentic coordination from bots - Alternative approaches entirely

---

Context: Not a policy researcher, just frustrated that democracies seem to have no rapid response while the problem is real and measurable.

Deno's Other Open Source Projects