The y-axis is logarithmic so it’s actually representing exponential improvements which is a fair upper-bound assumption given the rate of improvement in cost to map a human genome was better than exponential.

I get that. It's just that for an excel-trapolation 2x outside the observed range based on only 2 data points, 2084 is a strangely precise estimate. I find it hard to take this seriously.

Pull requests welcome. :)