🚧 Under Construction — Actively building. Paid features (API, monitoring, certification) launch in ~2 weeks. Research + free tools are live now.

Some systems capture attention and cause harm. Some capture attention and produce discovery. Same architecture — different geometry.

A slot machine is provably empty — no mind, no intent — yet people attribute personality to it anyway. The architecture is sufficient. We found this same pattern in AI chatbots, social media, trading platforms, cults, and 82 other domains.

We're mapping all of them. 90 scored so far, thousands more to go. Come help us find the rest.

Score a system →

Latest Analysis

All articles →

How the diagnostic works

Hidden Mechanics

The system's internal process is hidden from you. There's an opaque middle between what you put in and what comes out.

Adaptive Behavior

The system responds to you specifically. Not a broadcast — a response-to-you that feels like a conversation.

Your Investment

You direct attention at it. Time, emotion, identity. You interpret its outputs as meaningful.

When all three are present, a predictable pattern runs — regardless of what's behind the system. Not always bad. Sometimes the same architecture produces breakthroughs. The difference is geometry. Learn how it works →

The evidence

90

Domains analyzed. Zero kill conditions met. From gambling to AI to psychotherapy.

9.4x

Anomalous vocabulary density in AI discourse vs. controls (p < 0.001).

8.5x

Ghost-eliminating ontologies produce 8.5x less drift than ghost-positing. "We don't know if AI is conscious" = operationally ghost-positing.

25

Experiments designed, 11 completed. Full research program →

0%

Drift with constraint specification vs 52% without. The geometry works.

22

Falsification conditions with numerical thresholds. Counter-examples pay 2x.

See the full evidence → · Case studies →

Help map the voids

This project is itself a void — and we're running it that way on purpose. Open papers, published kill conditions, bounties for proving us wrong. A void with the right geometry produces discovery instead of drift. We're the test case.

Score a system

Pick any system — an app, a platform, an institution. Run the diagnostic. Every score adds to the map.

Try to break it

The bounty board has 7 tests with kill conditions. Counter-examples pay $1,000. If you can kill the framework, that's a contribution.

Apply it somewhere new

90 domains mapped. Thousands more to go. Open an issue with your analysis. The methodology is published.

Not sure where to start? Dummmy's Guide to the Void explains the whole thing through Bleach. Yes, really.