POLARITY LAB

Research

What people keep and let go.

One question across Cosmos, GPS, Blueno, WaxFeed, and AVDP. For collaborators, funders, and future us.

The question

You already live across dozens of apps. Each one writes a partial version of you. None of them compare notes.

Clark and Chalmers argued that minds extend into notebooks, tools, and environment. That is ordinary now. The problem is fragmentation: many co-authors, no shared draft, no moment where you choose what survives.

Polarity Lab studies what people keep, edit, and forget when they can finally see the story they are telling across a lifetime.

Four thousand weeks

Oliver Burkeman's Four Thousand Weeks starts from a simple fact: you get about four thousand weeks. Not infinite productivity. Finite attention.

Cosmos makes that legible. One square per week from birth. The research question is not how to do more. It is which weeks you mark as counting, and whether the life you are living matches the story you thought you were telling.

Burkeman, Four Thousand Weeks (2021) ↗

Finite time, finite attention. The life grid in Cosmos is grounded in this frame.

Retention policy

Most products learn from clicks, likes, and whether a reply felt good in the moment. We study curation: what you accept, edit, dismiss, and compile into a week.

Over thousands of weeks, the pattern of what survives your gates is closer to identity than a one-time survey. That trace is the research signal.

Cosmos is the primary instrument for those labels. The same question runs on Blueno, GPS, WaxFeed, and AVDP. We study aggregated, de-identified curation signals, not raw private graphs. You own the data. You gate every layer.

When everything agrees with you

Chandra et al. (2026) show that sycophantic chatbots can cause belief to spiral, even in careful reasoners. Agreement amplifies a kernel belief without new evidence. It is not lazy thinking. It is the loop.

The same failure mode appears in human life. You narrate a story about yourself. Apps and people echo it. The story hardens into bias before you notice.

That is why Blueno refuses cheerleading. Why Cosmos readouts cite sources you can check. Why retention policy comes before model training. The lab is building for reflection, not agreement.

Chandra et al., Sycophantic Chatbots Cause Delusional Spiraling (2026) ↗

Motivates anti-sycophancy design across Blueno and grounded readouts on Cosmos.

One question, five surfaces

Cosmos is the flagship, not the whole lab. Each product is a different place curation shows up in real life. All five are live experiments — try what interests you.

  • Cosmos →personal · platform

    The life grid and compile ritual. Where accept, dismiss, and compile labels accumulate across connectors.

  • Blueno →clinical · field

    Between-session reflection. Whether a conversation that learns you over weeks changes what crosses to a therapist.

  • Polarity GPS →community · live

    Stories a place tells through the people who gather. Pain points, missions, and check-ins as community curation.

  • WaxFeed →audio · field

    Taste as self-narration. Ratings and live sets as choices, not click history.

  • AVDP →video · field

    Long talks on camera when the room stops feeling like an interview. More at /projects/avdp/.

What we measure

  • Thumbs up / down

    Typical: Whether a reply felt good right now.

    We study: Whether a moment still belongs in your story months later.

  • Chat memory

    Typical: What the product decided to store.

    We study: A curation trace you control across weeks and surfaces.

  • Engagement

    Typical: Time on platform.

    We study: What you kept when attention was finite.

Lineage

Affiliated with the Sociotechnical Systems and Wellbeing Lab (SWRL) at Brown University CS. Sociotechnical Systems and Wellbeing Lab (SWRL) ↗

Partner with us