Technology adoption, both axes on matching power scales (fit is a straight line)

x = (years before 2031)^0.22, labelled in calendar years; y = (years to majority)^0.233, labelled in time. Same fit as the other charts.

source: Same estimates and fit as adoption-one-curve.pages.dev and adoption-years-before-2031.pages.dev (Our World in Data/ITU, World Bank, WHO/UNICEF, GSMA, archaeology refs); AI point = ChatGPT weekly users (OpenAI posts) extended on the fastest trend • transform: x axis: years from each launch to 2031 (the fitted end year), raised to the power 0.22, so it sits between a plain timeline (power 1, old points squashed together) and a log scale (power near 0, recent points squashed together). 0.22 is the power that puts the 22 points closest to evenly spaced along the axis. Labels are calendar years. y axis: years to majority raised to the power p/k (the x power divided by the fit's exponent), so the fitted curve becomes exactly a straight line ending at 0 (instant) in 2031; that straightness is the form of the fit, not a finding. Same fitted curve: years = A x (2031 - year)^k. Dots = 21 estimates (car and plane removed: ownership is a poor measure) (teal: share of all people; green: share of each year's newborns vaccinated). Bars = range of intro year and of years to majority. Purple diamond = projection, NOT data (it is in the fit). Scroll to zoom. • drawn 2026-10-21 03:39 UTC