No, they don't all agree...
You only need some of the current core nodes in your unl to be sure to be part of the main chain, as compared to bitcoin where the Genesis block is simply compiled into the main client.
As ripple uses ledgers, this wouldn't work (you only need about 40 mb to get going, not 10+ GB of data). They could put some core nodes in their code, but then people would again complain that this is too centralized etc.
Your own server isn't useless, it is just a fork if you don't use the main ledger chain.
"some of the current core nodes". OK. So I put 30% "current core nodes" and 70% nodes that have forked ledgers. How would that work out? Could there end up a proliferation of forked ledgers across the internet? I doubt it. What incentive is there to running a forked ledger, when there is no core of services supporting that fork?
It seems to me that there is little incentive for any individual to run servers in Ripple. Whether you run a server or not, you are dependent on "core nodes"; so you might as well just use a client. I predict that Ripple will simply amount to a core set of services that people connect to through clients; whether you are a selling or buying. In order to knock out the entire Ripple network, it would just take the shutting down of a handful of core services.
In bitcoin, I'm forced to trust the chain with the most combined proof of work. In Ripple I'm forced to trust the ledger built by some particular core providers. I would say Bitcoin is a little more flexible. Right now the majority of the proof of work comes from a few pools, but the network could quickly adapt if those pools proved to be unreliable, and then the source of this proof of work could rapidly change. With Ripple, it would take the network much longer to adjust if the core providers were knocked out ie. due to legal or financial reasons. There would be far more users in the Ripple network who rely on using it via clients that connect to a particular server.
