First, I like the kid's methodology. You can eyeball the various differences between font X and font Y, and see that the same passage printed in one is going to take more ink than the other, but how do you quantify the difference? He came up with a clever hack to relate an easily measurable attribute to a not-so-easily measured one.

Second, intentionally or otherwise, he managed to divorce the savings ratio from the type of ink being used--whether you laser-print, inkjet-print, or press-print your text on paper, you're going to use x% less ink or toner with one font versus another.

However, the selection of a font should take things into consideration besides the relative amount of ink needed to produce a body of text. Human and machine readability should also be significant concerns. And I would like to point out that a cost savings of $136 million represents less than two seconds worth of spending at the US governments current spend rate of $3.5 trillion per year. I don't know about anyone else, but I can't even imagine that level of spending!