LLenny's Podcast
← All frameworks
InfluenceKeith Coleman (VP of Product) and Jay Baxter (ML Lead)

Pseudonymity for Honest Contribution

Let contributors stay anonymous — people cross partisan lines far more readily when their name isn't attached.

Difficulty
Easy
Time to result
~weeks to results
Steps
3
Confidence
88%

A counterintuitive design choice: allowing contributors to participate pseudonymously produces more honest, more cross-partisan input than requiring real names. Real identity makes people fear harassment and fear being seen breaking with 'their side,' so they self-censor. Anonymity frees them to agree with notes critical of their own tribe — provided you have enough quality mechanisms to keep the bar high.

Origin

A key learning from the Community Notes pilot (Jay Baxter and team). The team originally assumed real names were essential for trust — the first prototypes used them — and the pilot data proved the opposite. Coleman notes the same principle later justified making X likes private.

Core principles

  • 01Real-name systems suppress the very cross-boundary honesty you're trying to capture.
  • 02Anonymity reduces fear of harassment, expanding the pool of good contributions.
  • 03Privacy allows freedom for honesty — people engage with what they truly find helpful.
  • 04You can only afford pseudonymity if independent quality mechanisms hold the line.

How to run it

  1. 1

    Notice who is self-censoring

    In your pilot, watch for good contributions that aren't happening because contributors fear being attacked or being seen siding against their own group.

  2. 2

    Switch to pseudonymous participation

    Let people contribute under a pseudonym rather than a real name so they can engage honestly without reputational fear.

    Pro tip Research shows anonymity makes people markedly more willing to cross partisan boundaries and agree with the other side.

    Watch out Don't assume real names build trust — for this task they suppressed honest cross-partisan input.

  3. 3

    Backstop honesty with quality mechanisms

    Because pseudonymity could lower the quality contributors put out, rely on strong independent quality checks (bridging agreement, reputation filtering) so openness doesn't cost accuracy.

    Watch out Without robust quality mechanisms, pseudonymity's honesty gain can be offset by lower-quality contributions.

In the wild

The pilot reversed the team's assumption

The team's first prototypes depicted contributors posting under real handles, believing that was needed for the note to be trusted. Pilot feedback showed the best option was the opposite: pseudonymity drew out more, and more honest, notes.

Anonymous contribution became a permanent design choice; contributors write and cross-partisan-rate notes they'd have avoided under their real names.

Private likes as the same principle

Coleman connects the finding to X making likes private — people 'like' things they wouldn't have liked publicly.

More honest signal about what people actually find valuable, freed from social performance.

Common mistakes

Requiring real names to build trust

The intuitive assumption that real identity makes notes more trustworthy was flatly wrong; it shrank the pool of good notes and discouraged cross-partisan honesty.

Opening up honesty without quality guardrails

Pseudonymity can invite lower-effort contributions; it only works because the system has many independent quality mechanisms to keep the bar high.

Is it for you?

Best for

Designers of contribution, feedback, or rating systems on polarizing topics where honest cross-group input is the goal.

Not ideal for

Contexts where accountability or verified identity is legally or reputationally required, or where you lack strong quality-control mechanisms.

From the transcript

people are actually more willing to cross partisan boundaries when they are Anonymous or pseudonymous than when they are under their real name

1:27:00

by allowing people to just to be student Anonymous you actually get more honest answers about what they really think

1:28:00

we have so many quality mechanisms in the system that that wasn't an issue

1:28:30

From the episode

An inside look at X’s Community Notes

Keith Coleman (VP of Product) and Jay Baxter (ML Lead)