Consensus...mechanisim UNL

Google [Bot] like(s) this thread.

Re: Consensus...mechanisim UNL

Postby GryphonArk » Tue Apr 08, 2014 12:48 am

Ok let me restate it this way.

whatever % attack you need and it seem the hardware requirements are not high,

you set up x100 + virtual machines, and it seems there are not many ripple servers are out there right now and start running ripple, but with a modified source code.

Over time and maybe a day is enough you get all the transactions an verify them, with all other nodes, but and your UNLS look all ok. You then exclude the other 20% of servers from the network you are now the ripple network, running modified code, that you guessed it allow you to make more XRP or less if you wish (probably more).

All of a sudden Coinlab/Ripple has no way to access their system or use their XRP, as you could reject his as well or dilute them at least, and also ripple/coin labs has lost its user base as they would be now logging into our network.


It leads me to believe that more XRP can be created and the system taken over.
GryphonArk
 
Posts: 91
Joined: Sat Feb 23, 2013 3:13 am

Re: Consensus...mechanisim UNL

Postby moocowpong1 » Tue Apr 08, 2014 12:59 am

You're describing a scenario where you have a bunch of validators which exclude the "honest" validators, and the honest validators exclude you. This is the same as forking the network. In order to something malicious, at some point you need to get real people to trust you in particular ways. (Either validators putting you on their UNL's, or users directing their clients to talk to your servers.) Otherwise your attack is solipsistic.
rKdtD8YQXCJAnXoXQfmn645FybiaFWVghs
I accept GDW.
moocowpong1
 
Posts: 182
Joined: Wed Feb 20, 2013 5:58 pm

Re: Consensus...mechanisim UNL

Postby JoelKatz » Tue Apr 08, 2014 1:34 am

GryphonArk wrote:Over time and maybe a day is enough you get all the transactions an verify them, with all other nodes, but and your UNLS look all ok. You then exclude the other 20% of servers from the network you are now the ripple network, running modified code, that you guessed it allow you to make more XRP or less if you wish (probably more).
How would you get people to listen to you?
User avatar
JoelKatz
Ripple
Ripple
 
Posts: 1859
Joined: Sun Dec 23, 2012 3:45 pm
Location: Oakland, CA

Re: Consensus...mechanisim UNL

Postby GryphonArk » Tue Apr 08, 2014 3:07 am

JoelKatz wrote:
GryphonArk wrote:Over time and maybe a day is enough you get all the transactions an verify them, with all other nodes, but and your UNLS look all ok. You then exclude the other 20% of servers from the network you are now the ripple network, running modified code, that you guessed it allow you to make more XRP or less if you wish (probably more).
How would you get people to listen to you?


If ripple is decentralized, then any processor can join the network and you just start processing, there is no restriction on processing.

If the UNL becomes a mechanism for excluding nodes that want to join, and the UNL is controlled by say coin labs, then ripple is not de-centralized.

so the argument is is ripple decentralized or not?

In a decentralized system any one is aloud to join in whatever capacity they can manage. Eg BTC, you can join is with as may processors as you can muster, and do 51% 60% whatever, its just at that point self interest kicks in and you probably don't want to destroy the system.



Maybe the same for Ripple.

So are processors free to join ripple or not bolis down to a test if Ripple is decentralized or not, or their is tacit control by coinlabs via UNL

So if I can put it this way you should not have a way of seeking permission for people to listen to you. The proposition of the need to "get people to listen", may indicate some sort of centralized gateway mechanism.
GryphonArk
 
Posts: 91
Joined: Sat Feb 23, 2013 3:13 am

Re: Consensus...mechanisim UNL

Postby dchapes » Tue Apr 08, 2014 6:50 am

GryphonArk wrote:If ripple is decentralized, then any processor can join the network and you just start processing, there is no restriction on processing.

If the UNL becomes a mechanism for excluding nodes that want to join, and the UNL is controlled by say coin labs, then ripple is not de-centralized.

It's in these statements where your logic is flawed.

Decentralized does not imply anyone can do anything they want. Otherwise what is there to stop someone running a validator that generates new XRP or sends their own account(s) all the funds? Although anyone run whatever modified software they want on their own machines the rest of the network will (and should) ignore them if they aren't playing by the same rules. This is the same as with Bitcoin and any distributed system (decentralized or not). The fact that a bitcoin miner can't choose to award themselves >25 BTC as a block reward (and have anyone else accept that) doesn't change its distributed or decentralized nature.

Each individual validator can choose whatever UNL they want with whatever policy they want; if you want to run a validator that auto-adds any other validator it sees to it's UNL (bad idea), you're free to do so. There is no one making any hard rules about this whatsoever (there are suggestions on how to to it in a way that keeps the network healthy and secure). There is no central control of anyone's UNL.

As an example, in the (hopefully near) future when the majority of validators are not run by Ripple Labs, if Ripple Labs rolls out a new feature that no one wants, the other validators are free to configure their systems to disable the feature and vote via the change process to not enable it. If the majority of validators do so there is nothing Ripple Labs can do to force the feature into existence. This is the same as if the Bitcoin foundation or Gavin released a new change in bitcoind that no one wanted, they have no power to force any of the mining pools to accept blocks using the new feature(s). Similarly, if the community rolls out a change that Ripple Labs is against, as long as a majority of validators accept the feature there is nothing Ripple Labs can do to stop it.
OpenPGP KeyID: france short hawaii
Although I moderate and post a lot on the Ripple forum, I am not associated with RippleLabs; I'm just a Ripple user. I speak only for myself.
I am somewhat associated with RippleFederation and a minor financial supporter of RippleUnion.
dchapes
 
Posts: 1445
Joined: Thu Mar 14, 2013 1:08 pm
Location: Ontario, Canada

Re: Consensus...mechanisim UNL

Postby donch » Tue Apr 08, 2014 7:49 am

dchapes wrote:As an example, in the (hopefully near) future when the majority of validators are not run by Ripple Labs, if Ripple Labs rolls out a new feature that no one wants, the other validators are free to configure their systems to disable the feature and vote via the change process to not enable it. If the majority of validators do so there is nothing Ripple Labs can do to force the feature into existence. This is the same as if the Bitcoin foundation or Gavin released a new change in bitcoind that no one wanted, they have no power to force any of the mining pools to accept blocks using the new feature(s). Similarly, if the community rolls out a change that Ripple Labs is against, as long as a majority of validators accept the feature there is nothing Ripple Labs can do to stop it.


The "hopefully near" parenthesis tells a story, unfortunately. Due to the fact that Ripple Labs' five validators don't trust any of the other nodes on the 60 strong network, of which only about 30 I can see after the peer code refactor, and also the fact that they control the source code for the Ripple client, which defaults to their servers, there is no decentralisation at all. There is redundancy, but that is a different concept.

The "enable feature" process is untested and no transaction of that type exists as of yet in the ledger. The DeliveredAmount and Memo types bypassed that process when they were added to the protocol. There is no test network to test new features and I see no public discussion of how contracts are to be implemented.

If Ripple Labs want to promote this system as open and decentralised they need to trust someone other than themselves.
donch
 
Posts: 796
Joined: Mon Nov 18, 2013 8:07 pm

Re: Consensus...mechanisim UNL

Postby GryphonArk » Tue Apr 08, 2014 8:11 am

dchapes wrote:
GryphonArk wrote:If ripple is decentralized, then any processor can join the network and you just start processing, there is no restriction on processing.

If the UNL becomes a mechanism for excluding nodes that want to join, and the UNL is controlled by say coin labs, then ripple is not de-centralized.

It's in these statements where your logic is flawed.

au contraire my logic is not flawed, your reply confirmed my logic, you agree with in parts me but do not realize it, and in parts you state you don't like a particular outcome lets go through this part by part.

dchapes wrote:Decentralized does not imply anyone can do anything they want.

Yes, yes it does or it could. From where does your proposition spring from?



dchapes wrote:Otherwise what is there to stop someone running a validator that generates new XRP or sends their own account(s) all the funds?

that's not a problem of a definition, rather and outcome that you don't like. My argument is agnositc, if you can show more XRP can be generated, then good, ripple needs to be fixed or at least the parameters acknowledged.



dchapes wrote:Although anyone run whatever modified software they want on their own machines the rest of the network will (and should) ignore them if they aren't playing by the same rules.

Why, if the network consits of more nodes that like a particular form and you don't and then thats bad luck for your version of the game.



dchapes wrote:This is the same as with Bitcoin and any distributed system (decentralized or not). The fact that a bitcoin miner can't choose to award themselves >25 BTC as a block reward (and have anyone else accept that) doesn't change its distributed or decentralized nature.

This does not touch on the position where there is more combined hash that does allow 25 BTC to themsleves, so a non issue.

dchapes wrote:Each individual validator can choose whatever UNL they want with whatever policy they want; if you want to run a validator that auto-adds any other validator it sees to it's UNL (bad idea), you're free to do so. There is no one making any hard rules about this whatsoever (there are suggestions on how to to it in a way that keeps the network healthy and secure). There is no central control of anyone's UNL.

That's my point so if more UNLS's say 80% plus swamp out the exisitng UNLs game over for status quo, your agreeing with me here

dchapes wrote:As an example, in the (hopefully near) future when the majority of validators are not run by Ripple Labs, if Ripple Labs rolls out a new feature that no one wants, the other validators are free to configure their systems to disable the feature and vote via the change process to not enable it. If the majority of validators do so there is nothing Ripple Labs can do to force the feature into existence. This is the same as if the Bitcoin foundation or Gavin released a new change in bitcoind that no one wanted, they have no power to force any of the mining pools to accept blocks using the new feature(s). Similarly, if the community rolls out a change that Ripple Labs is against, as long as a majority of validators accept the feature there is nothing Ripple Labs can do to stop it.

and your agreeing with me here as per last point.

Your argument boils down to, your don't like an outcome, so it should not happen. That's a different position to what defines who makes the rules in a particular system
GryphonArk
 
Posts: 91
Joined: Sat Feb 23, 2013 3:13 am

Re: Consensus...mechanisim UNL

Postby BeRichLiveFree » Tue Apr 08, 2014 2:01 pm

GryphonArk wrote:
JoelKatz wrote:Over time and maybe a day is enough you get all the transactions an verify them, with all other nodes, but and your UNLS look all ok. You then exclude the other 20% of servers from the network you are now the ripple network, running modified code, that you guessed it allow you to make more XRP or less if you wish (probably more).
How would you get people to listen to you?


GryphonArk,
You do realize you did not answer the above question? You do understand this is a security mechanism?

GryphonArk wrote:If ripple is decentralized, then any processor can join the network and you just start processing, there is no restriction on processing.


This statement is based on your narrow definition and most likely your background with BTC. People need to understand Ripple is not anything like the BTC or any other blockchain coin. Sometimes, I wish I could go back in time when BTC was running like 20 or 40 servers and listen to people complain about BTC not being decentralized. LOL

GryphonArk wrote:If the UNL becomes a mechanism for excluding nodes that want to join, and the UNL is controlled by say coin labs, then ripple is not de-centralized.

so the argument is is ripple decentralized or not?


Ripple is decentralized processing; but, controlled for security reasons thru the UNL. As, for me, I am comfortable with how much the UNL is currently decentralized and it will become more decentralized over time.

However, one of the reasons everyone should be shaking in their boots is BTC 51%--which by the way does not have to be 51%--45% would still allow me to reek some havoc. Not to speak of the fact 3 large mining groups control 52% hashrate distribution (Ghash.IO 24%, Discus Fish 15%, BTC Guild 13%). http://blockchain.info/pools

When it comes to protecting people's money (personal and business) who do you think I would trust more? A publicly exposed company or a group of miners spread out all across the world not in the same legal jurisdiction. Again, Ripple is not BTC--Ripple labs is focused on a top down approach.

GryphonArk wrote: In a decentralized system any one is aloud to join in whatever capacity they can manage. Eg BTC, you can join is with as may processors as you can muster, and do 51% 60% whatever, its just at that point self interest kicks in and you probably don't want to destroy the system.


Again, your narrow definition of a decentralized system based on your BTC experience.
And, of course you do not want to destroy the system--not if you are smart--you want to manipulate the system to your advantage (read three mining groups). Why because your self interest (greed) kicks in.


GryphonArk wrote:Maybe the same for Ripple.

So are processors free to join ripple or not bolis down to a test if Ripple is decentralized or not, or their is tacit control by coinlabs via UNL

So if I can put it this way you should not have a way of seeking permission for people to listen to you. The proposition of the need to "get people to listen", may indicate some sort of centralized gateway mechanism.


We call it a security mechanism. It blows my mind that people really think all people can be trusted when it comes to money and a system which controls it. When it comes to my money, I prefer a more controlled UNL, than a we love everybody, because everyone is trustworthy UNL.

I think the BTC world has already proved not everyone can be trusted! But, if thinking everyone can be trusted works for you--kumbaya!

There might be things which are not perfect yet; but, Ripple as we know it today is just over a year old as I understand it. Ripple can not be what some people want it to be right this second; but, neither was BTC the first year or two.

And, if you do not feel comfortable with using the Ripple network that is ok--it appears enough people do because it continues to grow every month.

A lot of new people like me are not vested into the BTC mindset. I see what Ripple has to offer and the power of the system. And, having now studied BTC and Ripple for a year now, I understand the weaknesses and strengths of both I believe. And, not having been vested in either a year ago, I can now tell you I would pick Ripple every time.

Example: I buy BTC to move into Ripple to buy XRP. I have to be very careful and watch how BTC is trending when I buy it thru Coinbase. Why? Because it takes 15 to 30 minutes to then send the BTC to Bitstamp. And, it is only that quick because Bitstamp only requires 3 confirmations.

All systems have strengths and weaknesses and I am sorry you are so focused on what you think is a weakness. You might be missing the forest for the trees.

BeRichLiveFree :+1: 8-) :+1:
User avatar
BeRichLiveFree
 
Posts: 641
Joined: Fri May 17, 2013 5:02 pm

Re: Consensus...mechanisim UNL

Postby freequant » Tue Apr 08, 2014 3:14 pm

GryphonArk wrote:
dchapes wrote:Each individual validator can choose whatever UNL they want with whatever policy they want; if you want to run a validator that auto-adds any other validator it sees to it's UNL (bad idea), you're free to do so. There is no one making any hard rules about this whatsoever (there are suggestions on how to to it in a way that keeps the network healthy and secure). There is no central control of anyone's UNL.

That's my point so if more UNLS's say 80% plus swamp out the exisitng UNLs game over for status quo, your agreeing with me here

You are not talking about the same thing. What you say is that one can create hundreds of validators, have them trust each other so that they make 80% of each others UNL, and then you think that this will take over the network.
But what dchapes tries to explain you is that although you can indeed create that group of renegade validators, that won't affect the existing network because the decision making about what really happened isn't something universal that will spread to the whole network , but something local and subjective to each validator. It is totally possible that a validator that elected to listen only to bad apples in his UNL will see a different version of reality than a validator who put mostly reliable and reputable sources. So in your example where a large group of malicious validators decided to vet each other and agree on a lie, what would happen is that your group of validators and only this group will be dead sure that the lie you introduced is the hard truth, but nobody else would believe that since nobody is trusting any of your malicious validators, let alone connecting to them.

This is really similar with society at large. One example is research papers. Research papers need to cite other research papers to support claims that aren't being demonstrated in the paper. A researcher will cite only research papers that he believes are sound and relevant to his study. You could indeed make a group of corrupt and unknown researchers who would agree to cite each other as 80% of the references and keep only 20% of real whitepapers so as to create the illusion that there is consensus on something that isn't actually scientifically established. But if none of these researchers has any influence and any citation outside of their phony circle, and if their results are conflicting with what the mainstream holds as valid, they don't have much changes of being effectively cited in legit research papers.
Of course some lazy legit researchers may found their phony paper on ArXiv and decide to cite them without doing much due diligence, and they may even get a few articles in low quality tabloids, but they won't make it through the peer review process of major scientific magazines or in the reference section of most trusted researchers, and will fail to spread in academic literature. What's more, it won't be long before they are called out as imposters and unanimously rejected by the scientific community.
freequant
 
Posts: 894
Joined: Sat Apr 20, 2013 12:11 pm

Re: Consensus...mechanisim UNL

Postby donch » Tue Apr 08, 2014 3:19 pm

But to extend your analogy further, the current status quo is that all accepted papers are only being authored by Stanford University and they aren't citing anybody else:-)

There's the theory and then there's the reality. All successful transactions go through and require Ripple Labs hardware. Single point of failure. Non-decentralised. Unfortunate.
donch
 
Posts: 796
Joined: Mon Nov 18, 2013 8:07 pm

PreviousNext

Return to General Discussion

Who is online

Users browsing this forum: No registered users and 9 guests
cron