Michael Bolton has gone off like a volcano in Iceland, writing a series about what exploratory testing isn’t:
Another thing I would add to this:
Exploratory testing is not defined by any specific example of exploratory testing.
Just as tap dancing does not characterize ballroom dancing, you can’t take any one example of exploratory testing and treat that as representative of the entire concept of ET.
If you were to hear me singing an aria by Mozart, that would be an example of opera singing. It would be an example of BAD opera singing, but it would truly be an example of the style. Similarly, I regularly talk to testers who go “oh yeah I’ve seen that exploratory testing stuff but it’s not structured… not documented… not x… not y… not whatever.” And my reply is “you probably haven’t seen skilled exploratory testing. Would you like to hear me sing an opera now? OR, I could show you a good example of ET in practice.”
Exploratory testing can be done in an unskilled, slapdash, silly way. Just as a unskilled driver behind the wheel of a car is still a driver who is driving a car, a poor tester can still be doing ET– albeit probably not very well.
The cool thing about ET is that, even done badly, it’s still a great way to find some bugs. Michael and I try to help you do much, much better than that.
The core idea of ET remains as it always has been. It’s been expressed in many different ways, but boils down to this: test design and test execution and learning mixed together in a mutually supportive way. Whenever you see that, and to the degree that you see that, you are seeing exploratory testing.