Blog

Each post is a discovery. IRDME is the method — the finding is the story. All results link to their pre-registered hash.

theoryfinding

AI Attention vs Human Attention: Five Models Confirm the d4 Infrastructure Gap

We asked five AI systems (Claude, Grok, ChatGPT, Google AI, NotebookLM) to rank npm packages by importance using only a dependency edge list and no structural context. All five -- in both named and anonymous conditions -- recovered infrastructure centrality (d2) far better than human Stack Overflow attention does. Preliminary result: 10/10 conditions confirmed.

theoryfinding#d4#observer-layer#fpl#pre-registration#c-elegans#npm

d4: The Observer Layer -- Two Pre-registered Experiments on Where Attention Goes

We added a fourth layer to the IRDME grammar: d4, the observer layer. d4 encodes how a community, instrument, or external agent perceives a system -- not how the system is built (d1), not how it flows (d2), not how it behaves (d3). Two pre-registered experiments tested d4 formally for the first time: C. elegans neuroscience and npm software. The results are structurally surprising.