people are saying why wouldn't webmasters try and improve user metrics, whether they are part of panda or not, and okay that is true. but how do you measure an improvement in bounce rate? for a site like wikipedia, success could be a 100% bounce rate -- as the user found the info on the page they were sent to. but on another site success could be a 0% bounce rate, because they followed a purchase all the way through a shopping cart. its not like an exam paper, when everyone knows what is good and what is bad. the truth is, that each site could have a completely different bounce rate and each would be measured as a success. its the same with other user metrics like time spent on site. if you are looking for a fact on wikipedia then you could be in and out in 30 secs, which would be a "success". but on a shopping site 30 secs would be a disaster.
if these things are now part of the algo, then how do we know for sure what google considers a success? working your bounce rate up to 100% a la wikipedia, or down to 0%, might do you more harm than good.