this doesn't have 'gateways', that's Ripple you're talking about. It's not easy to shut down the authorities. They can be run from a cell phone or any other kind of unreliable connection(even email is possible). It doesn't require that anyone run a server, so you could authorize transactions from anywhere, even a mobile device. Secondly you must shut down ALL the authorities- and there could be hundreds all in different countries with different connections. It's quite resilient.
The fact is if you want some kind of asset with backing, there must be some kind of authority somewhere, because ultimately you need to be able to redeem those assets somehow. This system offers the MOST flexibility for designing that authority.
Then it looks like I didn't understand your paper.
What is the difference between a ripple gateway (such as bitstamp) and your "authority"? This post seems to imply that one asset might have multiple authorities... which would be cool, but how would be that possible in practice? i.e. every single authority should be able to redeem your "coins" with the backed valuables...
First off, anyone can implement this concept. Im not attempting to profit from this, at least directly. I am interested in building p2p financial products. I would like to USE this technology and I am looking for others who want to use this, then we can build a community of people who want this functionality. I think that ANYONE should be able to build and market financial products and liquidity should be available to ANYONE. Don't you agree?

The authorities are simply some kind of entity who has identified themselves in some way. It could simply be through the use of a key pair. It could be some bank sanctioned by the government of Costa Rica who advertises on their website that they are an authority in this chain. It could be the NY branch of the Federal Reserve. These authorities, once defined, have the ability to approve the block chain. The algorithm allows for something interesting- a kind of VOTING mechanism on what goes into the chain. Now if you are familiar with Bitcoin, this applies to the problems of, eg. double spends. If one authority thinks a transaction is valid, but the others dont and they overpower that authorities WEIGHT, then it doesnt get put into the chain. This, in of itself, is quite useful because we've factored out Proof Of Work, thus we have no latency- while maintaining the integrity aspect of Bitcoin. The only drawback is you must define these authorities in the first place.
There are other interesting advantages. In the case of distributed exchanges, the chronology can determine market winners and losers. How do we decide who gets to bid on trades? What if we wanted an auction? who decides who wins the auction if it depends on TIMING? This also solves this problem for the users so long as they feel properly represented in the order of authorities. If we tried to do this with PoW, then the miners might compete to put the chain in a favorable order. In my system we make this negotiation explicit.
this order can be defined in many ways. For instance you could have one singular FIAT authority who rules all, very easy to manage, but not very easy to sell to the public. You could have a TRIUMVIRATE- or three equal authorities- which is fairly easy to coordinate and also non-biased. You could extend this principle geometrically to five or more authorities. You can also have different weight ratios for instance 1 primary authority but a school of smaller authorities that can outweigh it if unanimous(called KING AND COURT in my terminology).
Re. Ripple, specifically how this idea differs from whatever Ripple is at the moment is anyone's guess. They have no released their source code. At this point I dont think they can drift significantly from the core concepts, and I do know that those concepts work quite differently than what I have. Confidence Chains is very familiar to Bitcoin developers. ASMOF, you can use the transactions AS IS, in the system- which is very convenient for Bitcoin users and developers.