Weekly Journal : Testing Reflection

It is hard enough to find an error in your code when you are looking for it; it is even harder when you have assumed your code is error-free. — Steve McConnell

Submitting Your Work

Submit a Markdown formatted plain text file, named week08_testing.md, to the individual portfolio_NETID repository provided for you in the course's Gitlab group.

Specification

While all code likely contains bugs, it is still vitally important to minimize them because software is literally a matter of life and death. Many say that testing code as you go improves and speeds up the development process and that testing is extremely creative and intellectually challenging. Yet most programmers still do not value testing or find it a boring task they would rather put off.

Now that you have some experience trying to test complex programs and, hopefully, trying to test as you go rather than only at the end of the project, take some time to reflect on your experience and how to improve it.

• Describe your preferred ways to do testing.
• How has it been to incorporate automated testing into programming process? What has made it easy or hard to do?
• How has testing changed or improved your debugging process?
• How has testing changed or improved your willingness to refactor code?
• How do you assess the quality of a test?
• How easy or hard is it for you to come up with "sad" path tests for your code?
• How could you improve your testing process?
• How important is it to you to do testing?

Finally, try to come up with 10 tests for the Google homepage (basically just a text field :).

Ethical Essay

Cellular Automata simulations have been used to model a variety of things scientifically, such as fire, racial segregation, ant foraging, fur patterns, other patterns in nature, economic theory, city growth, or even the entire universe. While such models certainly can help better understand complex systems, the information they abstract out can affect their accuracy or introduce bias, leading to ethical concerns. Thus it is said that all models are hypotheses that must be proven.

Beyond modeling, using programming in general for science is practically a requirement now but bugs, biased data sets, and flaws in data anonymization undermine any results' validity. Will it ever be possible for programmers to build systems that can earn and maintain public trust in scientific research?

Based on your interest or knowledge about the topic, choose one of the following issues to read about and discuss:

• Researchers find bug in Python script may have affected hundreds of studies
• Duke researcher says he ‘undoubtedly made a mistake’ in false data scandal
• Facial Recognition Software Developed at Duke Linked to Chinese Surveillance Programs
• Issues using Breast Cancer AI to Help Train Duke Students, Improve “High-Stakes” Decision Making
• Why Are We Using Black Box Models in AI When We Don’t Need To? A Lesson From an Explainable AI Competition
• Deceit at Duke: How fraud at a university research lab prompted a $112M fine