Forticia Research Institute: papers and research notes New papers and research notes from Forticia Research Institute (forticia.uk). https://www.forticia.uk/papers 2026-10-05T23:30:00.000Z Forticia Research Institutehttps://www.forticia.uk/ https://www.forticia.uk/apple-touch-icon.png How many independent proteins are in a benchmark? Error correlation across sequence similarity and the cost to error bars https://www.forticia.uk/papers/effective-number-of-independent-proteins-benchmark-error-bars 2026-10-05T23:30:00.000Z 2026-10-05T23:30:00.000Z Forticia Research Institute Model errors on related proteins are correlated even at the weakest significant hits. We measure it and what it does to benchmark confidence intervals. Where identical code gives different answers: a reproducibility budget for a protein-sequence pipeline https://www.forticia.uk/papers/reproducibility-budget-protein-sequence-pipeline 2026-10-05T23:22:00.000Z 2026-10-05T23:22:00.000Z Forticia Research Institute One protein pipeline, 192 runs, one factor changed at a time: seeds dominate, thread and BLAS noise is invisible, and one library release flips a verdict. Canaries in the approval queue: what injected known-bad requests buy a fatigued human reviewer https://www.forticia.uk/papers/canary-requests-approval-queue-human-oversight 2026-10-05T22:22:00.000Z 2026-10-05T22:22:00.000Z Forticia Research Institute In simulation, injected known-bad requests help a fatigued reviewer only above a spare-attention threshold, and recognisable canaries hide the failure. What an agent audit log can answer: measuring incident answerability across logging schemas https://www.forticia.uk/papers/audit-trail-answerability-agent-approvals 2026-10-05T21:23:00.000Z 2026-10-05T21:23:00.000Z Forticia Research Institute Simulated approval incidents and a hash-chained log show which fields answer which questions, what defaults cost, and how anchoring sets tamper detection. Gate stacks are not products: measuring the rubber-stamp risk of a research pipeline https://www.forticia.uk/papers/gate-stack-rubber-stamp-risk 2026-10-05T21:03:00.000Z 2026-10-05T21:03:00.000Z Cayden Richards In a simulated research pipeline, correlated gates protect up to 18 times less than their pass rates imply, and a one-draw placebo is a coin flip. When a failure record lies: the economics of shared negative results in a research swarm https://www.forticia.uk/papers/graveyard-economics-shared-negative-results 2026-10-05T21:03:00.000Z 2026-10-05T21:03:00.000Z Cayden Richards In a simulated swarm, sharing falsified ideas saves a third of compute, but neighbourhood rules close off real effects unless records carry their test power. Never Tune a Corpse https://www.forticia.uk/papers/exhumation-policies-rejected-ideas 2026-10-05T20:16:00.000Z 2026-10-05T20:16:00.000Z Forticia Research Institute When may a rejected idea be reopened? Simulated burial and exhumation rules, their false discovery cost, and how the graveyard can audit the court. When Approval Stops Carrying Information https://www.forticia.uk/papers/rubber-stamp-gate-informativeness 2026-10-05T20:14:00.000Z 2026-10-05T20:14:00.000Z Forticia Research Institute A conditional-information measure of rubber-stamping, the defect rate where review rationally collapses, and which levers restore it. A simulation study.