How Did Software Get So Reliable Without Proof?

From AcaWiki
Jump to: navigation, search

Citation: C. A. R. Hoare How Did Software Get So Reliable Without Proof?.
Internet Archive Scholar (search for fulltext): How Did Software Get So Reliable Without Proof?
Wikidata (metadata): Q65951249
Download: https://www.gwern.net/docs/math/1996-hoare.pdf
Tagged:

Summary

Crude estimates suggest between 1 and 10 independent correctable errors per 1000 lines of professionally written software code. Any one error could have disastrous effects. However, only 10 of several thousand deaths attributed to dependence on computers can be explained by software errors, while real-time mission critical software systems with tens of millions of lines of code each work. Why have prophecies of doom been unfulfilled?

  • Management
  • Testing
  • Debugging
  • Over-engineering
  • Programming Methodology

Recommends both improving practice and advancing theory, and for researchers to study practice based on available data. Also for researchers to close gaps between theories; specialization presents opportunities for unification.

Theoretical and Practical Relevance

Video summary: https://www.youtube.com/watch?v=eY8AyCZ5uUg

Retrieved from "https://acawiki.org/index.php?title=How_Did_Software_Get_So_Reliable_Without_Proof%3F&oldid=11531"