The Deathblow to Proof of Stake - page 3.

lynn_402

sr. member

Activity: 462

Merit: 253

Quote from: XbladeX on July 15, 2014, 09:57:00 AM

POS in general is secure as they stake holders are if they are stupid and keep 1/3 of all coin at one place they can suffer like VRC..
If they are smarter like PPC they wont have such problem.

Here even didn't see any blow in POS security becouse attacked have to know how attack coin...

Here was panic fear of dumping 8m coin on market nothing more...
if they wanted secure network they could do it other way.

PPC also has the advantage that, because it is older, distribution is considerably better. Which means that it's very unlikely that a single exchange would have a big percentage of all coins.

illodin

hero member

Activity: 966

Merit: 1003

Quote from: Propulsion on July 15, 2014, 10:03:05 AM

A lot of discussion is specifically about NXT.

Is there any difference in NXT's PoS implementation vs most of these new altcoins with PoS?

BlackCoin has PoS 2.0:

Coin Age will no longer be a factor to generate Proof-of-Stake interest. There is a need for more nodes online and staking. More staking nodes will also speed up the BlackCoin network and transaction time. Do keep in mind the staking rewards are not affected by this change.
The stake modifier will be changed at every modifier interval
Further blockchain timestamp changes will be made to avoid Proof-of-Stake blocks from orphaning.
BlackCoin will no longer be using the Scrypt algorithm for its Proof-of-Stake phase, and will revert back to SHA-256D. Using Scrypt offers no real advantages , and is in fact slower compared to some of the other alternatives.

devphp

sr. member

Activity: 336

Merit: 260

Quote from: Propulsion on July 15, 2014, 10:03:05 AM

A lot of discussion is specifically about NXT.

Is there any difference in NXT's PoS implementation vs most of these new altcoins with PoS?

NXT's implementation is unique, except if those new altcoins are clones of NXT (NFD, NAS).

Propulsion

hero member

Activity: 658

Merit: 500

The Buck Stops Here.

A lot of discussion is specifically about NXT.

Is there any difference in NXT's PoS implementation vs most of these new altcoins with PoS?

XbladeX

legendary

Activity: 1302

Merit: 1002

POS in general is secure as they stake holders are if they are stupid and keep 1/3 of all coin at one place they can suffer like VRC..
If they are smarter like PPC they wont have such problem.

Here even didn't see any blow in POS security becouse attacked have to know how attack coin...

Here was panic fear of dumping 8m coin on market nothing more...
if they wanted secure network they could do it other way.

Istanbul34

full member

Activity: 232

Merit: 100

Quote from: DeathAndTaxes on July 15, 2014, 09:13:14 AM

Quote from: devphp on July 15, 2014, 06:06:39 AM

As price action and market volume observations show, buying 51% of NXT would be at least two orders of magnitude more costly ($50+ bln) than buying 51% of hash power for Bitcoin ($500 mln), because each time 1% of NXTs is purchased the price goes up 25%, that would be exponential growth of capital required to buy 51% of all NXTs in existance.

You don't need 51% of the coins just 51% of the active stake. In no currency can 100% of the money supply be used for minting. If it was then the currency couldn't be used for anything else.

The NXT community discussed this matter with Vitalik Buterin from Ethereum. He was surprised about the solution of NXT and said there is no fatal flaw in NXT after he understood the solution.

https://nxtforum.org/general-discussion/bounty-for-successful-nothing-at-stake-attack/msg60166/#msg60166

Read his analysis and come back. If you still can describe how you can perform a succesful Nothing at Stake attack against NXT, I will believe you and I will bow for you.

farl4web

legendary

Activity: 1205

Merit: 1000

It's simpler to buy a lot of ASIC miners than to buy 51% of NXT for example. Good luck with that! Grin

devphp

sr. member

Activity: 336

Merit: 260

Quote from: DeathAndTaxes on July 15, 2014, 09:13:14 AM

Quote from: devphp on July 15, 2014, 06:06:39 AM

As price action and market volume observations show, buying 51% of NXT would be at least two orders of magnitude more costly ($50+ bln) than buying 51% of hash power for Bitcoin ($500 mln), because each time 1% of NXTs is purchased the price goes up 25%, that would be exponential growth of capital required to buy 51% of all NXTs in existance.

You don't need 51% of the coins just 51% of the active stake. In no currency can 100% of the money supply be used for minting. If it was then the currency couldn't be used for anything else.

There are not many reasons for people not to forge/stake in NXT, either solo (bigger stakes) or through pools (smaller ones). That's why most of the coins are forging, hence, yes, you need 51% or close to that and the capital two orders of magnitude larger to buy all those coins than the capital needed to buy hardware to hashrate attack Bitcoin.

farl4web

legendary

Activity: 1205

Merit: 1000

The difference is this. When people do a 51% attack on some mid-altcoin and kill it, they can later point there miners to the next victim. In Proof-of-Stake, you will lose your own stake. Then you not happy, you're done. Grin

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: devphp on July 15, 2014, 06:06:39 AM

As price action and market volume observations show, buying 51% of NXT would be at least two orders of magnitude more costly ($50+ bln) than buying 51% of hash power for Bitcoin ($500 mln), because each time 1% of NXTs is purchased the price goes up 25%, that would be exponential growth of capital required to buy 51% of all NXTs in existance.

You don't need 51% of the coins just 51% of the active stake. In no currency can 100% of the money supply be used for minting. If it was then the currency couldn't be used for anything else.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: masterOfDisaster on July 15, 2014, 04:59:07 AM

For PoS attacks you need currency units of the network you intend to attack. If you succeed, you diminish the value of the owned currency units - estimatively by vast amounts; there's no reuse for different networks.

No you need to have currency units in the PAST. That is the basis for the nothing at stake problem.

Say the active stake is 10% of the money supply.
In block X I have >5% of the money supply.
In block X+1 I sell my coins.
By x+10 the transaction is confirmed the new owner(s) have the coins.

I now have NOTHING as in nothing at stake.
I can still re-org the network by building an alternate chain back at block x when I did have the majority of the stake. It doesn't cost me anything to try, there is nothing I can lose in the process. I am using not coins but the history of coins I once had to perform the attack.

superresistant

legendary

Activity: 2156

Merit: 1131

Quote from: Brangdon on July 15, 2014, 06:03:08 AM

Quote from: illodin on July 15, 2014, 04:56:19 AM

Sorry my ignorance regarding NXT, but someone could just start as many nodes as he wants and have majority of them?

Yes.

Nxt also has the notion of "hallmarking" a node, which means it is associated with an account and therefore a stake. Other nodes tend to trust hallmarked nodes more.

Quote from: newuser01 on July 15, 2014, 04:45:29 AM

I would still rather have big centralized pools than someone having the power to control the network without owning any hashing power or even 51% of the coins (even if he had them at one point, he can sell them off and then attack the network - attacking it at no cost).

How does your PoS deal with that, fork it to an earlier stage? haha

Nxt deals with it by not allowing block-chain re-organisations past 720 blocks. That means the attacker has a narrow window for making the attack. That is in addition to the usual difficulty of acquiring a large enough stake to make the attack feasible. For comparison, in a PoW currency it would go:

Buy hashpower.
Attack PoW.
Sell hashpower.

for a near-zero cost attack. In Nxt it would go:

Buy stake.
Wait 1440 blocks so that stake can forge.
Sell stake.
Attack PoS (within 720 blocks).

In both cases, the hard part is step 1. The difference in the ordering of last two steps is just a few hours. The PoW attacker has the advantage that they can sell off their hashpower at a gradual pace, without crashing the market. With Nxt, they'd have to carry out their attack within 12 hours of selling, so they'd have to sell quickly. Basically, buying 51% of Nxt is going to cost a fortune, and dumping 51% of Nxt would itself crash the price never mind the attack; and the attacker would lose a lot of money from the price crashing before they could sell their entire stake. So the notion that this attack has no cost is ludicrous.

+1
Great explanation.

Please everyone, read that quote before commenting on PoW vs PoS.
No system is perfect but don't let people spread FUD.

micax1

hero member

Activity: 708

Merit: 502

also I think we need a lot of time to get to pure PoS as mainstream
maybe 5-10 years

micax1

hero member

Activity: 708

Merit: 502

extreme attacks requires extreme measures... so rollback is justified
however i consider mintpal story bullshit.

Brangdon

sr. member

Activity: 365

Merit: 251

Quote from: illodin on July 15, 2014, 06:04:26 AM

Quote from: Brangdon on July 15, 2014, 05:41:06 AM

Quote

Or all IPO+PoS coins for that matter, because in the beginning someone had 100% of the coins.

Are you confusing "Nothing at Stake" with "History Attack"? Nxt mitigates history attacks by not allowing block-chain re-organisations past 720 blocks. That means we don't have to worry about the founders mounting an attack with ancient coins.

Thanks for the reply.

My pleasure.

Quote

Does any other PoS coin have such a "rewind" limitation like NXT's 720?

I don't know. Nxt is open source and has its clones, and I guess the clones have the same rules, but I don't know about PoS coins which are more original. I gather Peercoin use centralised check-points instead.

devphp

sr. member

Activity: 336

Merit: 260

As price action and market volume observations show, buying 51% of NXT would be at least two orders of magnitude more costly ($50+ bln) than buying 51% of hash power for Bitcoin ($500 mln), because each time 1% of NXTs is purchased the price goes up 25%, that would be exponential growth of capital required to buy 51% of all NXTs in existance.

illodin

hero member

Activity: 966

Merit: 1003

Quote from: Brangdon on July 15, 2014, 05:41:06 AM

Quote

Or all IPO+PoS coins for that matter, because in the beginning someone had 100% of the coins.

Are you confusing "Nothing at Stake" with "History Attack"? Nxt mitigates history attacks by not allowing block-chain re-organisations past 720 blocks. That means we don't have to worry about the founders mounting an attack with ancient coins.

Thanks for the reply.

Does any other PoS coin have such a "rewind" limitation like NXT's 720?

Brangdon

sr. member

Activity: 365

Merit: 251

Quote from: illodin on July 15, 2014, 04:56:19 AM

Sorry my ignorance regarding NXT, but someone could just start as many nodes as he wants and have majority of them?

Yes.

Nxt also has the notion of "hallmarking" a node, which means it is associated with an account and therefore a stake. Other nodes tend to trust hallmarked nodes more.

Quote from: newuser01 on July 15, 2014, 04:45:29 AM

I would still rather have big centralized pools than someone having the power to control the network without owning any hashing power or even 51% of the coins (even if he had them at one point, he can sell them off and then attack the network - attacking it at no cost).

How does your PoS deal with that, fork it to an earlier stage? haha

Nxt deals with it by not allowing block-chain re-organisations past 720 blocks. That means the attacker has a narrow window for making the attack. That is in addition to the usual difficulty of acquiring a large enough stake to make the attack feasible. For comparison, in a PoW currency it would go:

Buy hashpower.
Attack PoW.
Sell hashpower.

for a near-zero cost attack. In Nxt it would go:

Buy stake.
Wait 1440 blocks so that stake can forge.
Sell stake.
Attack PoS (within 720 blocks).

In both cases, the hard part is step 1. The difference in the ordering of last two steps is just a few hours. The PoW attacker has the advantage that they can sell off their hashpower at a gradual pace, without crashing the market. With Nxt, they'd have to carry out their attack within 12 hours of selling, so they'd have to sell quickly. Basically, buying 51% of Nxt is going to cost a fortune, and dumping 51% of Nxt would itself crash the price never mind the attack; and the attacker would lose a lot of money from the price crashing before they could sell their entire stake. So the notion that this attack has no cost is ludicrous.

Chillin_with_beer

newbie

Activity: 43

Merit: 0

pos is the only sustainable solution so far. What you describe is the problem of one young coin, not pos. See https://www.youtube.com/watch?v=A2jx1TlkMBs , he does a very thorough and clear analysis of pos in the long term.

devphp

sr. member

Activity: 336

Merit: 260

Quote from: Brangdon on July 15, 2014, 05:41:06 AM

Are you confusing "Nothing at Stake" with "History Attack"? Nxt mitigates history attacks by not allowing block-chain re-organisations past 720 blocks. That means we don't have to worry about the founders mounting an attack with ancient coins.

Yes, he does confuse them, many people do, because they simply parrot others' wrong opinions.

History attack is impossible because all blocks older than 720 in the past are irreversible, there are decentralized rolling checkpoints in NXT which take care of that.

Here is how Vitalik Buterin formulated the "Nothing at stake" issue" two weeks ago:
https://nxtforum.org/general-discussion/bounty-for-successful-nothing-at-stake-attack/msg60114/#msg60114
read that thread from that post to the end to see that he was apparently satisfied with the replies he got that NXT is not vulnerable to N@S, or at least he had nothing else to retort.

Topic: The Deathblow to Proof of Stake - page 3. (Read 7929 times)