Software Testing as a Multi-Armed Bandit Problem
ocram575.substack.com · 3,318 words · saved by 1 readers
approaching software testing using adaptive sampling techniques
Recently delved into the rabbit hole of software verification, after listening to Jane Street’s Signals & Threads with Ron Minsky and Antithesis CEO’s Will Wilson. I would highly recommend checking out the podcast in general, they truly have something for everyone. This episode in particular introduced Antithesis’s goal of solving software verification for any generic piece of software or software system. Someone who might be familiar with the theory of computation and Church-Turing Thesis can already see a problem with this in the form of the halting problem which is incomputable. As…
saved by
related reading
- The Project of Software is Completesubstack.com
- Software Engineering at Googleabseil.io
- Reading 2: Testingweb.mit.edu
- Catching bugs with fractional proofs | Theoremtheorem.dev
- The Multi-Armed Bandit Problem and Its Solutions | Lil'Loglilianweng.github.io
- Asymmetry of verification and verifier’s rule - Jason Weijasonwei.net
- Competing with sampling — Alignment Research Centeralignment.org
- GitHub - abhivaikar/howtheytest: A collection of public resources about how software companies test their softwaregithub.com
- What it Takes for Coding Agents to Complete Large Software Tasksfactory.ai
- Notes on software qualityanthonyhobday.com
- Stephen Wolfram Writingswritings.stephenwolfram.com
- Multi-armed bandit - Wikipediaen.wikipedia.org