I find this an interesting perspective because I prefer reading linear graphs. Show me a linear graph and explain the axes and I'll easily see what kind of dist

I find this an interesting perspective because I prefer reading linear graphs. Show me a linear graph and explain the axes and I'll easily see what kind of distribution it is. Show me something that grows non-linearly and I would have no idea if it's exponential or polynomial or what. > We all know how much a thousand dollars is (as opposed to an income centile), and how hard it is to get a $1000 raise as opposed to how hard it is to get one that will push us up by one percentile. The point here is that wealth is not linear. Getting a $1000 raise depends entirely on what you're earning. What you're earning depends on where you live and so on. With linear axes I can easily compare the numbers around my own context, but it doesn't help me extrapolate beyond that. No matter how you choose your axes will fix this. Ultimately there's no right answer to how to choose a scale. Different scales will show different patterns. The log-scale is useful here because it also represents the distribution of the wealth (to some extent). A log-scale will basically stretch out the data points at the lower values, giving more space to present the data for those who earn less. And there are more people who earn less! I do wish they presented the dollar-amount for the log-scale though. EDIT: Wait, the income-rank is actually a percentile of how much people earn? That's perfect! Then I can look at Rank=50 and know that 50% of the population is represented by the left side of the chart. That's actually very useful.