1. double-spend. It seems ripple uses trust instead of mining to prevent double spend. Even if an IOU is issued by a trustworthy entity, people spending it may not. In contrast, in a colored-bitcoin scheme you only need to trust the IOU issuer, and the rest is protected by mining. If there is a trust-less and decentralized solution without any kind of proof-of-work or proof-or-stake, it would definitly be a bitcoin-killer (but I don't think this could happen).
Think of a room full of people who all agree with each other. To enter the room, you must agree with them. To disagree with them, you must leave the room. They all sit in this room maintaining continuous agreement on everything. Each of them who is honest puts their first priority on enforcing the rules of the room, their second priority on maintaining agreement with everyone who is also willing to follow the rules, and their third priority on accepting legitimate transactions provided they don't violate the first two rules. The rules of the room make it infeasible to agree to a transaction once a conflicting transaction has been agreed to -- such an agreement cannot be formed and be valid according to the rules.
So this is basically a proof-of-stake based on number of nodes and web-of-trust? i.e., all of my honest friends think this is not a double-spend, so this is not a double-spend. I seriously doubt if this could be at least as safe as proof-of-work of bitcoin. If it were, Satoshi would not have to implement proof-of-work. Obviously, an attacker could split the network (if not overwriting transactions) by controlling >51% of nodes and declaring the rest as dishonest. The only reason for abolishing mining I could think of is to allow the developers to collect transaction tax through selling Ripple Credit
How about my second point? I can illustrate by the following table:
| Payment method | Anonymity | Traceable by unrelated parties |
| Gold, cash, pre-paid gift card | Yes | No |
| Credit card | No | No |
| Bitcoin | Yes | Yes |
| Ripple | No | Yes |
In a creditor-debtor relationship, only the creditor could be anonymous. If the debtor is anonymous, the creditor will very likely get burnt (e.g. TorWallet). For cash and gift card, the holders are debtor so they could be anonymous. Gold and bitcoin are commodity and not debt, so they could be anonymous for anyone. On the other hand, credit card and ripple are based on credit so they could not be anonymous. (theoretically the debtor in ripple could be anonymous, but this seems not intended in the design)
Gold and cash are basically untraceable. There is transaction record for gift card and credit card but they are inaccessible to unrelated parties and are good enough for daily legal use. Bitcoin is decentralized and the record has to be traceable for everyone to prevent double-spend. This is not a major problem because bitcoin is anonymous. Ripple is the worst choice among all. Who would like to show their credit card bill to everyone, related and unrelated? Not me.