Resilient Data Futures

A SciOS whitepaper

Resilient Data Futures

The SciOS Resilient Data Futures whitepaper argues that research data loss is architectural rather than operational — and that the accumulated loss carries roughly $1.1 billion per year in latent liability at a representative R1. The same architecture that hedges that liability also produces the infrastructure AI-ready data requires.

73–93%

of published research can't produce its data on request

Citation: C-0002

$1.1B / yr

latent liability at a representative R1 ($200M annual research)

Citation: C-0005

191

research data repositories closed since 2012

Citation: E-0006

The form

We publish the argument as a discourse graph

Every claim, evidence item, question, method, and source is its own addressable node. You cite a claim by ID, contradict it with a counter-claim, or support it with a single new piece of evidence — without writing a paper around it. Narratives composed from the graph regenerate as new nodes accumulate.

Loading graph…

349 nodes · 644 edges

View full graph

Where to start

Engaging with a discourse graph

A discourse graph isn't read like a paper. There's no fixed reading order — readers choose where to enter and what to follow. See the structure at a glance, read the argument end-to-end, or open a single node and follow its edges from there. Each path covers the same set of claims, evidence, questions, methods, and sources. Over time, as the graph grows, the seams between papers begin to dissolve: a Claim, an Evidence item, a Method belongs to the graph of human discourse — becomes a part of whatever uses it — not to any single publication that happened to introduce it.