[adwf]

Yes very good point. Particularly with tablets, the use-case they are frequently replacing is that of an old-fashioned print magazine.

I did just test turning off webfonts on Medium and I still honestly can't see any huge problem with my default system typeface. This could possibly be because I have nice defaults, but at the same time, I think the good readability of Medium is far more about the overall style/layout of the page (particularly the decent large font-size), than about any particular typeface.

Example 1 - normal Medium: http://i.imgur.com/D4hYBPy.png

Example 2 - default system fonts: http://i.imgur.com/Jvnuhkm.png

Whilst I can see the difference in typeface, I can't honestly say that I'm experiencing any huge readability changes. Perhaps if I were to switch to a Sans of some sort, but that's not exactly a huge decision that needs to take a lot of time and fuss. By far and away the most important thing that makes this page readable and nice is the font size, closely followed by the line-width.

[anandgrafiti]

Good typography is often invisible. I clearly feel far more elegance with the original typefaces:) Its a subtle thing, something that cannot be quantified. Ofcource, content is the king, but in a world where everyone can create good content, how well will you present your content will be key.

[kevgnulldev]

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!

[vdaniuk]

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.