/. should not be a forum for perpetrating common ignorance, such as the comment,
"What's more, the number of Flash users is based on a questionable internet survey of just 4,600 people â" around 0.0005% of the suggested 956,000,000 total. Is it really possible that 99% penetration could have been reached?"
They really needed to survey just 1,000 people to get a statistically meaningful survey.
It does not pass the smell test because it leaves out a number of important devices we know to exist on the Internet (for example, the iPhone).
The problem is almost certainly sample bias. 1,000 data points is significantly relevant if your sample is truly random and not skewed towards a particular subgroup. Sample bias means that your mechanism for picking who you sampled would be more likely to pull data points from a specific subgroup. For example, a methodology that discouraged responses from people on mobile devices.