I'm interested in the ability to measure bounce rate by amount of time spent on the page. A bounce that occurs in less than 5 seconds after the page load is a lot different than a bounce that occurs 30 seconds after the page loads. Both visitors may not have found the content/page engaging and decided to leave without interacting with the page, but their experiences and intent could be entirely different. With the less than 5 seconds segment of visitors, perhaps they got to the page by mistake and immediately hit the back button. Whereas the group that lands on the page and doesn't immediately bounce may have found something interesting but just not enough to engage them to click. By segmenting the different groups of "bouncers", we'd get a better sense of performance.
I know that the measurement of web behavior may not be there yet, but when it is, I hope this type of tracking/reporting becomes available!