Aggregating Confidence: A Problem and A Solution

Stella Cohen; Rose Gotlieb; C Michael Holloway; Sarah M Lehman

Home/Archive/NASA NTRS

NASA NTRS · 20230012813

Aggregating Confidence: A Problem and A Solution

Stella Cohen·Rose Gotlieb·C Michael Holloway·Sarah M Lehman

View original source ↗Export metadata ↓

Abstract

This paper has two main purposes: to articulate a concern about assessing arguments based on confidence; and to provide an assessment procedure which takes this concern into account. The concern is that confidence is a tricky notion for non-deductive Arguments: doubts inherent in Reasoning coalesces with doubts inherent in Premises. And, within a Compound Argument, this may happen many times over, thus accumulating in unpredictable ways. We explain two procedures we have developed to help address the concern in a methodical way.

Explore related subjects

language argument philosophy safety reasoning

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Stella Cohen, Rose Gotlieb, C Michael Holloway, Sarah M Lehman. 2023-11-01. Aggregating Confidence: A Problem and A Solution. https://ntrs.nasa.gov/citations/20230012813

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

NASA NTRS2025

Detangling Safety Knowledge with Contextualism

We have contradictory intuitions about knowing that things are safe: on the one hand, we claim to have knowledge that many systems (even complex ones) are safe in ordinary language. On the other hand, when challenged by an expert, we easily admit that we do not, in fact, possess very much safety knowledge at all. This Technical Memorandum (TM) attempts to reconcile these two competing intuitions using a philosophical view called epistemic contextualism. We can use epistemic contextualism to explain the importance of safety regulations and the goal in implementing them: first, to differentiate types of disagreement and establish a common basis upon which regulators and regulatees can know that a system is safe; second, to raise ordinary consumers’ standards for safety knowledge (given adequate knowledge of the comprehensiveness of the regulation). This TM also uses the contextualist framework of relevant alternatives to elucidate types of failure possibilities and explain complications in attempts to prevent common mode failures.

language↗

NASA NTRS2023

The Friendly Argument Notation (FAN): 2023 Version

This document constitutes the official description of the current iteration of the Friendly Argument Notation (FAN). This new version provides several enhancements to the original 2020 instantiation, while maintaining essential compatibility with it. Specifically, the new version enables distinguishing between deductive and non-deductive arguments, removes the requirement for always providing an explicit statement of reasoning, and relaxes the rules for when labels may be used. The primary intended use of FAN is unchanged: creating and evaluating arguments about safety-critical systems, specifically the types of arguments common within safety and assurance cases.

language↗