Topic: Ripple Q & A @ Joel Katz and Ripple inc. - page 4. (Read 8856 times)

It's designed to operate as a full decentralized network. It is not all the way there yet. Open sourcing the server is one step. An infrastructure for managing validator lists without a central authority is the next big step.

Well, we can start the process by convincing a number of distinct organizations to run validators, publish lists of validators maintained using an open process, and so on. But I do agree that real decentralization will require people to choose to adopt Ripple and choose to participate. Basically, it's like the way Bitcoin became decentralized -- it took large numbers of different groups to run miners, participate in the development of the software, and form a robust community that can make decisions about the direction the network should go.

This is where I think Joel's statement of their goal is to get it to be decentralized is a misnomer.

Just like you can't create a free-market on your own that is fully robust and trustworthy and adds value to the economy.

I honestly don't think it is possible for OpenCoin to be the party to get ripple to be truly decentralized. The topic of decentralization is a free-market type phenomena that occurs when the market chooses for it to happen. It just happens, there is no formula or solution to getting something to be that way...it just happens on its own.

It's currently very centralized - especially with Ripple.com/client, the web light client used by >99% of Ripple users only trusting OpenCoin validators.

Ripple promises that they will decentralize the network, however I raise objections to if it's actually possible to have decentralization of the level of Bitcoin with Ripple's consensus model.

So Ripple isn't decentralized then?

I seem to keep hearing that it is from the people you work for and with. It almost seems like this topic is a moving target.

You guys really need to make some real clarity on the topic as well as others.

Currently I look at it as centralized given there is a company behind the ripple network. That company can be shut down or regulated as the powers that be see fit.

You can see how much outstanding credit they have and how much of ot they've used. You can also see how other people value the assets they issue. (This is the same way the unusually high price of BTC at Mt. Gox leads you to believe there's some reason people value their USD at less than face value.)

It's possible that in the future tools might allow you to evaluate whether to trust people based solely on their behavior inside the network. For example, say you see someone who issues an asset that's very liquid -- you can easily sell it. And say that allowing them to we you a dollar would save you ten cents on a transaction. You might elect to accept that risk. After a while, and after this trust line has produced some value for you, you might raise the trust to, say $5. It may be such that even if they default on the whole $5, you're still up overall because the trust line made payments cheaper for you.

But this future may never happen. I think it's likely that for the foreseeable future, you'll need reasons outside the network itself if you don't want to take unreasonable risks.

One reason is that this can be gamed. I can issue bitcoins and offer to trade $130 for them. I can then create a lot of accounts that also offer to trade $130 for them. This might make you think that it's a liquid asset. But once you agree to hold it, I can pull all those offers. Now in theory you could check the people who placed the offers and use heuristics to make sure they're legitimate. And certain types of transactions (such as those that pay a transfer fee to a gateway) can only be faked at a cost.

But for now, we suggest people only create pathways to businesses that they have a reason to believe will continue to redeem as they've agreed to.

Can the network tell me if someone is acting irresponsibly?

Thank you

Tech guys. I do data center installations and I swear to god you all have the personality of a server cabinet. Really people dont give a shit about these issues. These are details, things that can be fixed and changed.

The important aspects is the monetary theory behind the ideas! Tell me that it wont work because it is a debt based system that has the potential to collapse into a debt deflation death spiral then you will have my ear.

Joel you need to convince the masses not IT guys. You really have to work on explaining things so the average person can understand. Lines of code dont mean shit to me.   

While this sounds promising, I still don't see how it can turn Ripple into decentralized rather than distributed. The consensus system requires people to have similar UNLs. Democracy is distributed, however the resulting government is not decentralized.

You can only have a debt outstanding to someone who specifically choose to allow you to owe them money. If you're worried about this, don't extend any credit to individuals or ensure that there are provisions to cover those debts in the event of their death. Alternatively, extend a small enough amount of credit that the benefits of using and re-using that credit over and over outweigh the risk of the one-time loss should they default.

But this is one of the problems with community/social credit. It requires people to manage a fairly complex kind of risk. This is the main reason I don't think the world is ready for community/social credit yet. It will take sophisticated tools to sensibly manage that risk and perhaps also some social changes.

The client does need to communicate with one or more servers (as do servers, of course). Currently, only servers have a set of chosen validators.

It's still early in Ripple's evolution. You can build software that uses the protocol and connects to the network that has any combination of features and capabilities that you want. We're not quite sure how people are going to wind up using it. One possibility is that people will run "thin servers" and connect a browser-based client to their own server. Another possibility is that people will run "heavy clients" that have a list of validators and act more like servers do (though probably only as observers to the consensus process). Another possibility is that Ripple will be invisible and people will use it indirectly through things like federation. We've designed the protocol and the data structures to be flexible enough to support fairly light implementations that can still implement their own UNLs. The reference client isn't there yet. There's still a lot of work to do. It's still early.

I also suggest making the website accurate instead of misleading.

"Send money to anyone, anywhere in any currency"

You can't send any currency other than XRP. You can only send XRPs or IOUs.

"all payments are processed automatically without any third-parties or intermediaries"

Is a validator not a third party? A validator is not the sender or receiver, by definition it is a third party.

"And because Ripple is a distributed system, there is no single point of weakness or attack."

You admitted that's not the current state of Ripple, so the use of present tense is troubling.

"Most payment systems are controlled by for-profit organizations that charge buyers and sellers for the privilege of making payments. In contrast, the Ripple payment network is controlled by a free Internet protocol that can be used by anyone."

Unless international financial networks uses carrier pidgins, I believe they also use the internet. For profit gateways like BitStamp charge buyers and sellers for the privilege of making payments.

What happens if someone gets hit by a bus with a ton of outstanding debts?

  



  

Anyone can just like you can become the president of the US if you're a US citizen and meet the requirements. Saying what is theoretically possible is irrelevant if practically it is never executed. There's actually significant vendor lock in at work here.

A standalone browser client would still need to communicate with a server, right? Only the server can choose validators / set UNLs.

Anyone else can set up a client and server now. The problem is that a trojaned client could steal your secret. A standalone client (or browser plug in) would be great to have. Anyone who wants to can implement one, of course. I know we have one in our plans, but I'm not sure what the timeline is. Again, this is another place where we agree. These are risks, and we'd prefer people did not have to take these risks so we're working to reduce or eliminate them.

If people have to trust us, that means we get the blame if something goes wrong and the risk that we might betray that trust is a disincentive to use the system. If people don't have to trust us, then we don't have to do anything or take any risk. We definitely do not want people to have to trust us.

I think a little hard would be a severe understatement. >99% of your users use ripple.com/client, which does not allow choosing validator nodes. Will ripple.com/client ever come with support for choosing your own validators, and no defaults - like the EU Windows choose your browser screen?

The change process would be quite difficult too given ledger fragmentation / forking, and that's not to say you're still having to trust entities, instead of just trusting maths.

Yes. It would be a little hard right now because the infrastructure to make it easy to do that isn't complete yet. Effectively, the community would probably just have to agree on someone to replace us. Until someone else could execute a full decentralization plan.

But we're hard at work on the transition to greater decentralization. We have no wish to control or operate the Ripple network. We have no desire to own the network or the protocol. The reason it's important that we get this right is that we want it to be as impossible for some other entity to become a central authority controlling the network. When we fully let go, we need to make sure nobody else can grab on.

Validators can cause you to lose money through double spending.

Another currency can impose a transfer fee which will be taken if you try to convert that IOU into Ripples.


And even if you try to change that, you'll end up on a different ledger.

In theory yes, as a practical matter no.  Every client by default comes set to trust Ripple, Inc.  A significant portion of the network will never change that.   Sure you can untrust Ripple, Inc and trust a set of validators with incompatible rules but you have simply forked yourself into a insignificant minority network.

It is like saying if miners got everyone to agree could they change the block subsidy to 50,000 BTC.  They can and the majority of the network would simply ignore them.  You could fork Bitcoin to give you 1,000,000 BTC as well.  In theory nothing stops you, the practical reality that a network of one is useless is a hard thing to overcome though. 

We are not talking about different rules, we are talking about different entities with the same rules. How do you propose on solving the centralization situtation with the Ripple light client?