One of the things that immediately struck me was that the conservative judges seemed to be more conservative than the liberal judges were liberal, based on the vividness of the color. Aha, I thought, I knew the liberals were more moderate!

But actually I think that's an artifact of the way the coloring was done. Look at Rehnquist as an associate and see the vivid red that his first year shows as, which is 3.98. Now look at Thurgood Marshall, below him, and find a -3.95. Those should look pretty similar in terms of intensity, but the blue looks much closer to white than the red does.

What I think is happening is this. They are color coding not on absolutes like that, but based on the distance between 0 and the most conservative or liberal number. But the most liberal justice is at -6, which the most conservative one is only at -5. So if you get a 4, that's 80% of the way to being the most conservative, but someone who is equivalently liberal at -4 is only 66% of the way to being the most liberal. So they get a color that looks like they are 66 points away from moderate as opposed to 80 points for the conservatives.

Well that's misleading. I think the color gradation changes need to be symmetrical across the graph or it's going to be super confusing. Maybe just call -5 the most liberal you can be and don't worry about shading Douglas more? Or make 6 the most conservative you can be and give up the super deep red color for now.