I've constitute an interesting article inwards MSDN magazine. It describes several paradoxes that tin tumble out during software testing process.
1. Simpson's paradox demonstrates that conclusions of several information sets tin differ from conclusions of a combined information set. Thus, the exam lawsuit summary does non necessarily represents the same every bit the master uncombined data.
2. Braess's paradox may arise during network functioning testing in addition to states that adding capacity to a network tin genuinely cut the performance.
3. Parrondo's paradox assumes that when combined, ii completely independent but faulty systems tin brand upwardly a right system.
The article is costless to read here:
Enjoy reading!