A synthetic clinical-AI study on when sharp end-of-life risk estimates lose their reference-class basis.
research.
papers, preprints,
artifacts.
A focused archive of published and preprint work across database systems, cybersecurity, systems security, clinical AI, cloud systems, education, organization theory, physics, and notation.
grep -r papers/
Journal articles and public preprints, reconciled with the current public ORCID record.
Undecidability, software opacity, and the residual uncertainty that probabilistic cyber-risk models cannot price away.
A retry-aware authenticated-encryption design that closes the gap between link reliability and AEAD verification.
Tests why sharply confident mortality predictions can remain invalid when a case has no transportable reference class.
Sets out a low-interruption framework that keeps a young child’s self-directed activity at the centre of prepared environments.
Treats HTTP caching as a database primitive with cell-interned MVCC, dependency-derived ETags, and query-scoped deltas.
Explains why undecidability leaves a residual class of cyber-risk uncertainty beyond ordinary probabilistic estimates.
Aligns retry handling and authenticated encryption so chiplet receive state advances only after verified delivery.
Maps the parameter regimes in which the dual-polarity preon model can sustain an emergent acoustic metric.
Shows how claims about simulated consciousness depend on the theory of mind used to evaluate the simulation.
Proposes composite neutral pairs whose long-wavelength dynamics yield an effective metric in a toy model.
Develops a privacy-preserving federated intrusion-detection approach for resource-constrained IoT networks.
Models organizational hierarchies as threshold circuits to make latency, accuracy, and headcount trade-offs explicit.