http://news.nationalpost.com/news/canada/christie-blatchford...
http://news.nationalpost.com/full-comment/a-good-day-for-jus...
http://news.nationalpost.com/news/canada/christie-blatchford...
http://news.nationalpost.com/full-comment/a-good-day-for-jus...
If you are familiar with hypothesis testing in statistics, it is the exact same idea. Your null hypothesis (innocent) is accepted by default unless you have overwhelming evidence (a very small p-value) that your hypothesis is not true. In that case, you reject the null hypothesis (and convict). But with a lukewarm evidence (not tiny p-value), you just "fail to reject". You never actively accept the null hypothesis.
This is why "not guilty" is not the same thing as "innocent". A court does not attempt to establish someone's innocence or not. A court is biased towards convicting only in the face of undeniable evidence. And that's why things such as whether the victims contacted Ghomeshi after the assault or not are "reasonable doubt" and thus he's acquited. Not because the court proved that the assault didn't happen, but because the court failed to obtain undeniable evidence that it did.
There is a fairly dramatic, well-known example of this (outside of the area of sexual assault) in the OJ Simpson case where the high (beyond a reasonable doubt) standard of criminal law failed to be met, resulting in a criminal acquittal for murder, but the lower (preponderance of the evidence -- or, in layman's terms, whatever the jury finds is most likely to be true) standard of civil liability resulted in a finding of civil liability for wrongful death.
A criminal acquittal means innocence only from the perspective of the criminal justice system, it doesn't mean the act charged was not committed by the person charged.