Once, I had a job at a consulting firm, with a contract for a communication system, in a noisy environment.
We proposed, for the channel, a BCH error correction scheme. Well known, textbook, reliable... but complex. The client, understandably concerned about software defects, wanted extensive testing. Actually, they wanted PROOF it would never fail.
Of course, that's theoretically and practically impossible. Some rough calculations indicated a complete simulation - every possible combination of errors - would take 10 years.
I proposed a simple Monte Carlo simulation: generate random errors, and see if the code patched them correctly. A software simulation to test the software,
But no, the client insisted on a working prototype, hardware, generating REAL NOISE errors, in the actual environment, for this test! Excuse me, dipshit, er I mean Mr. Client Who Knows Best, we're talking LOGIC here, error correction code! oh well, more billable hours, who's complaining?
Incredible! How do such nimrods get into management positions?