Logarithmic scale would better show the percent change of population in the early years. Using linear scale hides the movement of the early years because the 20th century's magnitude is so great that the previous centuries look identical to each other. It's more dramatic this way, but perhaps misleading.

But that is the point of the graph - to show the ratio of history that has happened has happened recently. Or in other words, the point is not to show the growth patterns, but to show the nominal value.

I think that's where it's wrong, though; imo, "history" is better identified with the percent change, not with the nominal value. If someone improved some technology by 10% in 1850, and someone else improved it by another 10% in 1950, I'd call those historically equivalent contributions.