Ripple and bitshares rated the most insecure blockchain software by China CERT

[JoelKatz]

I think it's difficult to draw a line between centralization and decentralization.

My test is usually this one -- is there a head that, if you cut it off, the thing will die even if the folks who are left want it to continue?

So, for example, eBay is somewhat decentralized in that the items are sold by individual people. eBay doesn't have a centralized warehouse. But, of course, if the company stopped operations, the platform would die no matter how much everyone else wanted to continue selling. The company has the secret sauce ... the database, the licenses, the server software, and so on.

Even back when everyone ran the Bitcoin software Gavin told them to, Bitcoin was still decentralized. If Gavin went away, someone else could replace him. He had no legal means to force people to do what he wanted. He held no patent, no restrictive license, no secret sauce. People just ran the software Gavin told them to because a benevolent dictator that everyone chooses to follow is a damn good form of governance, particularly when there aren't that many stakeholders who really care about the platform.

Similarly, Ripple holds no secret sauce. Others can run validators, and do. Others can trust whatever validators they wish to, and do. Generally, people tend to do what we tell them to because we make good decisions and we care about the network more than pretty much anyone else does. But we hold no stick, no legal powers, no secret sauce, no real authority. People follow us because it's easier than making their own decisions and we haven't screwed up badly enough yet for people to justify the effort of doing the work that we do for them.

If we're lucky and the platform becomes more valuable and important, it will be harder and harder to remain the benevolent dictator who keeps everyone happy. Ask Gavin.

[Sukrim]

If Ripple Inc. stopped (for whatever reason) validating RCL for 1 hour and if (that's a BIG if, since it is not a recommended setting for UNLs right now) consensus moved forward regardless, would you re-join the network at the then current stage or when you left it one hour ago? If it forked, would that influence your decision and if you decided to choose a fork, how would you choose?

[JoelKatz]

If Ripple Inc. stopped (for whatever reason) validating RCL for 1 hour and if (that's a BIG if, since it is not a recommended setting for UNLs right now) consensus moved forward regardless, would you re-join the network at the then current stage or when you left it one hour ago? If it forked, would that influence your decision and if you decided to choose a fork, how would you choose?

Currently, we would accept whichever valid ledger (that is, all transactions follow the correct rules) had the highest weight by rippled's current algorithm. That would mean it would have to pass these tests (oversimplified a tiny bit):
1) It would have to have as a prior ledger whatever the last fully validated ledger was.
2) Each step from that last fully validated ledger to the candidate ledger would have to be replayable. That is, the transactions couldn't violate any gross rules.
Of those valid ledgers, we would accept the one with the highest weight. Factors that influence the weight include trusted validations for that ledger and nodes observed to be running that ledger.
So, effectively, we would take the "majority" ledger of those that don't violate sanity rules.

This code is also the code that helps the network recover from a very rare, but always theoretically possible, potential failure mode. There's a trivial proof that consensus will always have some risk of failing. Typically, only a small percentage of nodes fail each round and the split nodes rejoin the pack quickly and cause no harm. It's like a school of fish where 2% of the fish leave every few seconds and the 2% that split off a while ago rejoin. There's always more than 90% of the fish in the school.

But it is theoretically possible for consensus to fragment very badly. You can reduce the probability of this happening, but the cost is reducing the best case speed, so it's better to tolerate it than to avoid it. Imagine this happens and the network splits into ten groups each with 10% of the network and each in their own consensus round. This is effectively the same situation as the one you are hypothesizing, just over a smaller period of time. You need some avalanche of ledgers to get the network back to being able to fully validate a ledger.

Interestingly, this doesn't hurt the transaction rate. Ledgers still close at the same speed. But it does hurt confirmation latency, since you could go several ledgers without any ledger fully validating.

[Spoetnik]

It's a pity a clearly smart mind is put to waste on such a horrible coin project idea.
Not sure how you as a dev can justify the glaring issues commonly complained about.
Dev guy.. you wonder why people do not like Ripple ?
Did that occur to you when you guys started making it ?

I do bet you love the climate now though.
You know as well as i do when you launched it, it was laughed off the Forum here.
Why ?

Why did Cryptsy refuse to add it for a year as it sat on the add-a-coin voting list ?
..even though it had more paid votes then every other coin on the massive list combined.

I smell an air of deniability.
And i think that you Ripple guys are going for it NOW because you see how corrupt all of crypto is.
You seen the scene change from anti-IPO shitcoin to now ICO's are somehow legit.
And now you Ripple guys are trying slide right in and pull an Ethereum after waiting a couple years.

I guess in a couple more years the bar will be lowered so much MORE that...
LEO Coin will be considered legit and you all will be buying them up eh ?

[Sukrim]

If Ripple Inc. stopped (for whatever reason) validating RCL for 1 hour and if (that's a BIG if, since it is not a recommended setting for UNLs right now) consensus moved forward regardless, would you re-join the network at the then current stage or when you left it one hour ago? If it forked, would that influence your decision and if you decided to choose a fork, how would you choose?

Currently, we would accept whichever valid ledger (that is, all transactions follow the correct rules) had the highest weight by rippled's current algorithm. That would mean it would have to pass these tests (oversimplified a tiny bit):
1) It would have to have as a prior ledger whatever the last fully validated ledger was.
2) Each step from that last fully validated ledger to the candidate ledger would have to be replayable. That is, the transactions couldn't violate any gross rules.
Of those valid ledgers, we would accept the one with the highest weight. Factors that influence the weight include trusted validations for that ledger and nodes observed to be running that ledger.
So, effectively, we would take the "majority" ledger of those that don't violate sanity rules.

Great to hear. This is quite theoretical anyways, since probably a lot of validators would not reach quorum once your nodes are down, so manual intervention would likely be necessary while no ledgers are being validated.

@troll above: You're getting boring.

[Spoetnik]

If Ripple Inc. stopped (for whatever reason) validating RCL for 1 hour and if (that's a BIG if, since it is not a recommended setting for UNLs right now) consensus moved forward regardless, would you re-join the network at the then current stage or when you left it one hour ago? If it forked, would that influence your decision and if you decided to choose a fork, how would you choose?

Currently, we would accept whichever valid ledger (that is, all transactions follow the correct rules) had the highest weight by rippled's current algorithm. That would mean it would have to pass these tests (oversimplified a tiny bit):
1) It would have to have as a prior ledger whatever the last fully validated ledger was.
2) Each step from that last fully validated ledger to the candidate ledger would have to be replayable. That is, the transactions couldn't violate any gross rules.
Of those valid ledgers, we would accept the one with the highest weight. Factors that influence the weight include trusted validations for that ledger and nodes observed to be running that ledger.
So, effectively, we would take the "majority" ledger of those that don't violate sanity rules.

Great to hear. This is quite theoretical anyways, since probably a lot of validators would not reach quorum once your nodes are down, so manual intervention would likely be necessary while no ledgers are being validated.

@troll above: You're getting boring.

I simply echo history.. i don't see anyone disputing what i said.
Want to challenge me ? If i HAVE TO i will go dig up the early topics on Ripple/Bitshares and quote them proving my point. (that both had a majority verbal reaction and market reaction that they were indeed unfair and unwanted)

That is called being honest ..not Trolling.
And since i am one of the few people who did not leave the Altcoin scene in disgust as it sinks in scammyness, then i was here to witness a lot of older ALT drama bs.
And you guys hate that and wish i would just go away so you can lower the bar more for more profits and more ICO's and more ROI'z  

Point here is almost no one ever dares challenge anything i say.
For example i point out how ICO's are scammy and they simply ignore it and make more.
Are they a better way to distrib coins over the BTC model ?
It can't be the same so it's either better or worse.. so which is it.. ?
Show your mouthy face's and back up your mouth and defend your stance !
I contend EARTH and the billions of people ignore ICO coins because.. they are scammy bullshit.
Which is why ADOPTION is doomed with any ICO.

PS:
Who ever did explain the great Ripple giveaway fraud shenanigans ?
Remember that ? i sure as fucking hell do.. i was here LOL

[JoelKatz]

Who ever did explain the great Ripple giveaway fraud shenanigans ?
Remember that ? i sure as fucking hell do.. i was here LOL

That's definitely a fair criticism. We tried, but we did make an awful lot of mistakes. It turns out that it's *really* hard. We got better as time went on, but honestly not that much better. I'm convinced now that giveaways of the types that we were initially trying are just a mistake, even if you could do them well.