What We Actually Know About Software Development, and Why We Believe It’s True

Notes from talk

  • have high standards when evaluating new methodologies / look for citations
  • 10-100x developer study
    • based on a small number of developers in the 60’s
    • newer study looked at ~50 and each person for less than 1 hour
  • Lutz Prechelt has done many studies here
  • Study has shown errors correlated to lines of code
  • most errors introduced at requirements analysis and design stage
    • how does agile vs intensive design compare?
  • limiting features greatly reduces complexity
    • maybe this is why agile is good?
  • if rewriting more than 25% of code, better to start from scratch
    • for aviation use case does it apply for other areas?
  • 1970’s study showed best way to remove bugs is to read the code
    • one person for ~1 hour
    • not much extra benefit adding another person
    • that could inform how we make pull requests
  • post release fault rate
    • no correlation based on geography
    • high correlation to distance in org chart
  • science can make mistakes
    • some metrics identified in the 1980’s as predictive of errors was actually due to lines of code
    • when lines of code was taken in to account the metrics didn’t do predict anything
  • Learn to think and ask the important questions



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s