I completely agree. A solid typeface evaporates between the page and the retina. Its shapes ease reading and allow the reader's vision to soften around the text and concentrate on the signal rather than the delivery modality. This is especially clear in mathematical typesetting and no discussion of type on a site with hacker in the name would be complete without a nod to the godfather (Donald Knuth) of both digital type and typography (both micro and macro see:

http://books.google.com/books/about/The_elements_of_typograp...

for a discussion of LOD in typography from a master designer)

To properly design his multi-volume set of CS/mathematics books:

http://www-cs-faculty.stanford.edu/~uno/taocp.html

Knuth developed an entire software toolchain that spans type face creation (metafont) to the document layout programming language TeX (yup, it's an actual Turing complete language!):

http://en.m.wikipedia.org/wiki/TeX http://dl.acm.org/citation.cfm?id=67555

He is also responsible for the creation of "literate programming"... All this in his spare time, of course, while he wasn't occupied with his responsibilities as a CS professor at Stanford, teaching, doing research (on things other than digital typography, e.g. the content of his books, combinatorics, etc. (...). Truly an inspiration!

Sure it can be quantified using the same instruments growth hackers use daily. You can measure time on page, click through rate, conversion rate and anything else you wish. I am sure that the difference in the metrics for the A/B test version that use fonts from GP post will be negligeble.