Thing is, time spent on an article doesn't really mean the user actually read
the article, nor was "touched" by it.
Maybe he just... did nothing (browsed other tabs, etc).
That's why +Justin Cutroni
solution looks much more interesting (at least to me) : by tracking if the user scrolled or not + the time he took to scroll, you get a better indication of his level of interest towards your content.
But I guess it all depends on what you consider a bounce, considering your website and your goals.
AFAIC, i consider a bounce any "new guy" who leaves without reading "rather carefully" my content, because it means I failed to catch his attention, which is what I want to do in the fist place.
That's why I used a script based on +Justin Cutroni
's solution (i actually use the Savio version + some minor tweaks to fit my specific needs).
Here : http://www.savio.no/blogg/a/114/tracking-content-scrollers-scanners-og-readers-in-google-analytics
+ this Adjusted bounce rate ( > 5 secs)
=> Result : i track "fast bounces", which gives me a better bounce rate AND i get to know how my visitors actually read (or not) my content.
Anyway, glad to see GA teams are aware of the need to give more meaning to defaut metrics GA provides :)