We were talking about the need for UNLs and consensus, which make Ripple servers totally dependent on a configured list of trusted nodes to fend off sybil attacks. Since XRP is just a unit of quantity that is being tracked on this weakened transactional system, how could you possibly call it an improvement? In addition, they decided to use a centralized model for distribution of XRP. How is that a step forward? Ripple transactions are dependent on the very weaknesses that Bitcoin was designed to overcome and ripple distribution is completely contrary to the goals of Bitcoin.
After all the XRP have been distributed or sold, the situation for Ripple is no different than that of Bitcoin after every coin has been mined. It just happens on a different timescale. It does not affect the properties of the cryptocurrency, only its distribution. The distribution method does not change the usefulness in serving as a medium of exchange used in commerce, other than having a different volatility profile (which the market can mitigate, i.e. bitpay).
It is an open question whether or not the Ripple consensus model is "weakend." The Ripple developers claim that their method is rigorous and is guaranteed to work correctly when the total number of participating nodes exceeds a relatively small threshold. On the other hand, there is no peer reviewed research paper that "proves" the claim. And Ripple is currently closed source so there's no realistic way to audit this method.
Whether or not Ripple is an improvement upon Bitcoin remains to be seen - it is an unproven statement. It could go in either direction at this point. If, however, the consensus model proves to be workable and correct then it will be clear that Ripple performs many additional useful functions beyond Bitcoin.
If the consensus model turns out to have a fatal weakness, it would be a disaster and Ripple will die on the vine.