Where to start.<p>First off, he's comparing proportions of OS use from two totally different datasets--one is proportions of web visitors to ycombinator.com, and one is systems described in interviews by usesthis.com. You can't meaningfully compare two totally different populations like this, can you?<p>But even if you could, you wouldn't want to represent the results as some type of off-axis skewed 3D pie chart, because those <i>only distort the data</i>.<p>So you have dishonest data, represented dishonestly. If you want to make a blog post about this, ask PG for his ycombinator.com visitor stats for this past...whatever period is equivalent to the period he got his numbers from in 2005. And visualize it using honest two-dimensional pie charts. And then you might have something useful to say.