If their ads said that negative ROI are likely, they probably wouldn't get much new business. I am no longer a Prosper fan, but I am sure they are using statistics to be able to claim (and probably justify) that this is not fraud. They clearly looked at some undisclosed set of past loans (based on Prosper 3.0 criteria) and extrapolating that data to come up with this "supported" ROI projection.
I would bet strongly that the mean Prosper 3.0 ROI will be very near or below 0. Since Prosper has changed their grading criteria and put in rate floors, the ROIs have likely changed (either up or down). IMHO we won't be able to reasonably claim that we are we can predict the results 3.0 loans ROIs until we have a few months of default information to see how they compare to earlier loans. If this data indictations the same miserable results, Prosper will use the same story as RP. They will claim that you can't really know the final ROI until 3 years have passed. If they are still around in 3 years, the strategy will have worked, bonuses will have been paid, and .org will likely have many new members.