There's a difference between running and running as well as on the 4S. The demo of noise reduction is impressive.
http://www.audience.com/demos/transmit-noise-en.php [audience.com]
It's easy to see why with that noise reduction, Siri would be much more accurate than without it, in real scenarios.
Apple obviously wants Siri to be judged on it's best performance. They have a reputation for quality to maintain.
However, I do agree that Google needs to sort the 'update problem'.