For a long time now, Google has weighted behavioral signals similar to what you describe. "Bounce Rate" is the percentage of users who quickly leave your site after clicking. "Dwell Time" is the amount of time a user spends on a page.
There's even a cottage industry around gaming these signals. See SerpClix and the like.
So are you suggesting that GOOG is using bounce rate on the target site (assuming it uses gAnalytics I take it - otherwise how would they know?) to alter the ranking of search results (like "90% of users bounce in the first two seconds on scam.copycat.com/seemed-useful so we're demoting its rank by 20%"?) That would be interesting, I was musing more about if GOOG does any session tracking on their end to try and infer how happy/unhappy users are with a given set of results. Of course they could do both at the same time.
Why not? Pinterest is a popular site, with lots of content all linked to each other. Many people probably spend a long time there after clicking a result.
The inordinate amount of time to click away all of the dark ui login screens just to see the content before making the decision its not what you wanted already increased the dwell time to longer than other sites.
There's even a cottage industry around gaming these signals. See SerpClix and the like.