Looks like this was thrown together without any proofreading whatsoever, and he didn't tell us how many total respondents there were in the survey. But, to his credit, he got rid of the pie charts and is back to the significantly more useful bar charts.
I still think that a database, allowing people (maybe just paying people, but whatever) to cross-reference and more deeply analyze the data, would be very useful. For example, answering a question like, "out of the people who wanted better 3d / OpenGL performance, what percentage of those were also interested in licensing/open source?" You cannot answer this question with the data the way it is presented.
Basically, even with the bar charts (which are an improvement, nevertheless), this valuable data is largely going to waste, because the potential to draw conclusions from it is rather limited.