Vanity, Vanity, all is Vanity

Anyone can make a technical error like this one
But the real problem is more serious
What really went wrong here?
1. People using Benchmark.pm have a tendency to disengage their brains
  - The author of the benchmark took the obviously nonsensical results at face value
  - He wrote up a detailed analysis of these nonsensical results
2. Benchmark.pm is complex
  - Here there was a scope problem that was obscured by the use of Benchmark.pm
  - The code wasn't doing what it appeared to be doing
3. Benchmark.pm's internals are obscure
  - This tends to inhibit understanding of the absolute numbers that it emits
  - You tend to compare the relative quantities only