It took a minute for me to realize: since the y axis is on a log scale it is not possible to include Vista, XP, NT and the like.
NT = nT = nano-tera-(dimensionless) = 10^-9 * 10^9 = 1
Edit: Ok I looked it up. Looks like π actually is a greek numeral (who knew?) with a value of 80. X would be 600. So it would actually be 680.
Also: "The expected return of a given policy (π) in reinforcement learning, denoted ρ^π"
...Unicode perversely has some greek letter superscripts but not π.
x axis.