That's an outcome that I was only able to derive after giving the graph the benefit of the doubt, eyeballing some of the numeric data, and doing some calculations in IPython. But it isn't a result that I feel one can accurately derive from the visualization presented by the graph. (If both had increased by 50%, the shape of the graph would be very similar.)
It is not a contradiction to say the presentation doesn't help show the conclusion, and that the conclusion is (nonetheless) correct.
Fair enough, I guess I just think about things a little differently. I didn't feel the need to do any calculations. You can just eyeball it and see one went from roughly $600 to $900 and that is obviously a 50% increase and one went from a little above $1000 to a little above $1800 and that is obviously around a 80% increase. Those are all nice and easy to process numbers even if they aren't exact. There is a big enough gap in the results to be confident that whatever margin of error existed in my estimates it is clear that the developer salaries grew at a faster rate.
It is not a contradiction to say the presentation doesn't help show the conclusion, and that the conclusion is (nonetheless) correct.