Very true, realy it is 33% of those who bothered to try it and bothered not to verify the results and bothered to tell them. I had a go, had a big fat picture splat up on my screen with some perverce advent calandar mentality going on. I concluded Bing failed at that point, and as I didn't record my input as many others then I was never counted.

That raises the point that optional feedback will gain 100% of those who migrate at the time and a far lower percentage of those who will not. So that 33%, well you can half that and half again and still be short on the number who tried it.

Now if that 33% was also backed up by access log saying we had 10,000 unique vistors to this experiment and 33% of those said they were using bing now then I'd be less sceptic.

You can also bet this 33% will get turned into 33.33 recuring % or 1/3rd and from there the lie will be complete.