Is Manual Testing Superior to Automated Testing?

Automation traditionally takes over the "three Ds"

• dull
• dirty
• dangerous with a fourth added more recently:
• dear (expensive)

with this in mind, manual testing can be dull and dear. So it fits 50%, most of the time.

Depends on how you define manual testing.

If you define manual testing as a big stack of test cases that you execute any time you release a feature, I would argue that should be automated and executed manually.

If you define manual testing as someone critically thinking about the system, it's users, risk, and exploration I would argue manual testing is superior. Both are needed for a well rounded test strategy.

Manual testing is a bottleneck at scale:

• It requires a human to execute, where as automated testing can be scheduled
• You have to manually document that you executed tests and record their results
• Manual testing can foster tribal knowledge; tribal knowledge is easily lost with turnover.
• You can't move towards a true continuous deployment environment with manual testing, it's not practical for a human to repeat a suite of tests multiple times a day at scale.

I'll bite! To me, it seems we should do both. Automated testing can do things like:

• Test very quickly
• Test all the scenarios
• Repetitively test
• Test under load

...and so on. However, what it can't do is more "ad-hoc" - more like "I wonder what happens if I click here?" A test case writer would have to anticipate that in order to test that case. A good friend of mine (and fellow exec in our venture) is just such a good tester because he likes to play around - jiggle the mouse, click randomly and drag/drop unexpected combinations. That alone identifies many issues that a test case writer wouldn't imagine to code for.

why choose between them? take both

• let users test manually
• record all user interactions in Playwright
• replay test case automatically with/in Playwrigth

github.com/microsoft/playwright