Nxt Forum

Nxt Discussion => Nxt General Discussion => Topic started by: delulo on July 24, 2014, 10:16:47 pm

Title: NXT POS and Bitshares DPOS
Post by: delulo on July 24, 2014, 10:16:47 pm
Hey guys!
I would like to discuss the ad- and disadvantages of both systems with you.

In my summary below I listed the disadvantages of the two security model ideal types "delegation" vs. "non-delegation". Discussing NXT's vs. Bitshares' block production algorithms as they are is more complicated because NXT allows for delegation (=leasing) as well and Bitshares does "approval voting" when delegating the stake as opposed to NXT which uses "direct" delegation. This makes it more complicated to compare.

Information: What is DPOS? Essentially a proof of stake variant where stakeholders can vote for delegates that then find blocks and secure the network. Here is an explanation to "delegated proof of stake" http://wiki.bitshares.org/index.php/DPOS

So my assessment up to now is this. I adapted it based on the feedback I got here. Thanks for the constructive discussion!! Please let me know if you disagree with anything.

-----------------------------------------------------------------------------------------------------------------------------------------------------------
Normal POS (no delegation possible):
Contra:
1.1) With competitive tx fees: A non delegating POS system gets centralized with those that have a big stake, more centralized than DPOS in the end. See https://bitsharestalk.org/index.php?topic=5564.0;all (don't take the thread subject too serious ;) ) --- Status: Open question: See https://nxtforum.org/general-discussion/nxt-pos-vs-bitshares-dpos/msg71586/#msg71586
1.2) If delegation is not an option, a lot of stake (!) has to be online (passwords used regularly etc. no offline tx signing) which is pretty insecure. Status: Can be crossed out because leasing stake to your own account is possible. Contradiction (see https://nxtforum.org/general-discussion/nxt-pos-vs-bitshares-dpos/msg71578/#msg71578): But this then would allow big pools which was said NXT has limited now to 0.5% of the total stake: https://nxtforum.org/general-discussion/nxt-pos-vs-bitshares-dpos/msg70340/#msg70340
1.3) Stakeholders that want to mess up the network can do that every time it is their turn. Only solution is a fork.
1.4) NXT stakeholders = forgers might not be specialized in running forging servers. [relevant if this becomes a mass market service]

DPOS (stakeholders can delegate their forging power):
Contra:
2.1) Delegates do not have the same interest in the success of the system as shareholders that are forgers when delegation is not possible.
2.2) An attacker could put up delegates and deceive shareholders. If the attacker manages to put up 50+% of the delegates by letting them gain the trust of shareholders over a long time and then he can perform one big attack after which he would be voted out. This would be cheaper than buying 51 % of the stake but maybe deceiving shareholders on such a scale is not possible at all.

Further pros and cons?

Scales better / lower tx fees under conditions of higher tx volume  is not an argument for Bitshares as I see it because the few forgers with non delegated POS could also have expensive, reliable forging servers like DPOS delegates and with transparent forging it is known in advance who produces the next block like with DPOS.

-------------------------------------------------------------------------------------------------------------------------------

CONCLUSION is not that DPOS is better only because POS (no delegation) has more contra points! Different points might be of different importance. Also it is not entirely a discussion of NXT vs. Bitshares block production because NXT also allows for leasing.

Reply of the bitshares main dev (bytemaster, Daniel Larimer):
Quote
Regular POS:   Bad Share Holders can still mess with the system *some* of the time.
DPOS:  Bad Shareholders have very little influence over operation of the network assuming the majority are not deceived.

DPOS: Can be attacked by deception of majority, but easily fixed with a quick hard-fork to new delegates.
POS: Can be attacked by deception of majority (leased forging) or by one large actor. 

Assuming large actors are benevolent I guess the systems are equal... but DPOS provides an easier way to reach consensus on change where as large owners effectively control regular POS.   

POS: Controlled by large minority share holders
DPOS: Controlled by majority of shareholders, large minorities have influence but are not sufficient to control.

POS: effectively turns into DPOS in practice, DPOS just streamlines the system
I couldn't post a link? So you can find the discussion at BitShares Forum » Discussion » General Discussion » POS vs. DPOS

I started the topic on the bitshares forum not to promote any of the two systems but to get as close to a realistic/neutral perspective as possible. There is no system without disadvantages ;)

So let's use our different perspectives to get closer to the truth and discuss it as neutral as possible. I would appreciate if we could do that!
Would be great if you could read the wiki doc so we can have a quality discussion.

My original post was this https://bitsharestalk.org/index.php?topic=6172.0 But it changed quite a bit.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Damelon on July 24, 2014, 10:24:13 pm
https://bitsharestalk.org/index.php?topic=6172.0

Found the link for you, to make it easier :)

Linking below 10 posts is disabled due to spamming issues. Sorry for the inconvenience and welcome to the forums!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 24, 2014, 10:26:32 pm
xxxlinkxxx [can't answer if there is a link in here]
Found the link for you, to make it easier :)

Linking below 10 posts is disabled due to spamming issues. Sorry for the inconvenience and welcome to the forums!
Thanks for the link and the warm welcome!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Damelon on July 24, 2014, 10:30:06 pm
And here is the wiki link: http://wiki.bitshares.org/index.php/DPOS

I just also found this link on the Bitshares forum with some extra info: https://bitsharestalk.org/index.php?topic=6102.0
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 24, 2014, 10:49:40 pm
So do you see it the same way as me in my first Quote? [getting closer to my 10 posts...]
Title: Re: NXT POS vs. Bitshares DPOS
Post by: mczarnek on July 24, 2014, 10:56:10 pm
I don't know enough about DPOS, would be nice to have a short description within the first post, as I understand it basic idea is that shareholders point their forging power at forging machines and the 100 machines with the most forging power to the forging.  Correct?

One other thing to throw into the mix, the current plan is that Nxt is going to not allow leasors to lease to anyone with more than 0.5% of the forging power.  Meaning that it should be harder for anyone person to get a large percentage of the forging power.  Basically the same thing, right?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 24, 2014, 11:31:40 pm
I don't know enough about DPOS, would be nice to have a short description within the first post, as I understand it basic idea is that shareholders point their forging power at forging machines and the 100 machines with the most forging power to the forging.  Correct?

One other thing to throw into the mix, Nxt is going to not allow leasors to lease to anyone with more than 0.5% of the forging power.  Meaning that it should be harder for anyone person to get a large percentage of the forging power.  Basically the same thing, right?
No DPOS has changed since the original whitepaper.  It is more like an election now. 
Instead of everyone competing to forge on their own with their own stake, people just use their stake to vote for the most trusted people to do the forging for the whole network.  Right now the top 101 "delegates" with the most votes are elected sign the blocks.  Dividends can be paid equally to all shareholders by simply destroying transaction fees (your percent ownership of network goes up so this is the same as equally distributing fees).  You can see what percent of transaction fees each delegate has decided to keep as a fee for service and what percent is destroyed to pay shareholders.  You can vote for cheap delegates that destroy most of the fee if you like.

It's arguably more scalable because the "delegates" can eventually invest in the hardware needed to process very high transaction volume.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 24, 2014, 11:39:05 pm
I don't know enough about DPOS, would be nice to have a short description within the first post, as I understand it basic idea is that shareholders point their forging power at forging machines and the 100 machines with the most forging power to the forging.  Correct?
Let me try: Shareholders delegate/lease their forging power to delegates that earn tx fees when producing blocks. Block production is predictable like with NXT transparent forging. Everyone knows in advance (for the next 101 blocks) who's where in the queue. Delegates are competing for votes and are campaigning (on the forum) with different strategies. Some advertise their delegate with high up-time, some burn the tx fees (=pay dividends), some promise to do good marketing with the tx fees they earned etc. One person can set up and campaign with several delegates. The chance of being a delegate should increase if it is a well known forum member (which is a guess).
The voting algo is "approval voting" (as opposed to delegation, which I guess NXT/leasing is based on, correct?): A shareholder can approve up to 101 delegates = everyone he trusts. The top 101 delegates, based on how many approvals they got, get to produce blocks and earn tx fees.   
Probably forgot s.t. but that should be roughly it.

Quote
One other thing to throw into the mix, Nxt is going to not allow leasors to lease to anyone with more than 0.5% of the forging power.  Meaning that it should be harder for anyone person to get a large percentage of the forging power.  Basically the same thing, right?
Yes, delegating and leasing are two different words for the same thing. With DPOS you can not earn tx if you dont delegate / lease. You can be a delegate yourself and delegate the whole forging power you have based on the size of your stake to your own delegate.

I also have a question: Is the following possible? I have a lot of nxt (say 5 percent of the whole stake). I then lease my forging power to one of my own accounts that only has a small balance. This would mitigate the insecurity involved with having to have my bunch of next online all the time. In other words: Can I put my NXT in cold storage and be safe and at the same time have my forging power leased to an account I control and earn tx fees there?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Damelon on July 24, 2014, 11:45:24 pm
I also have a question: Is the following possible? I have a lot of nxt (say 5 percent of the whole stake). I then lease my forging power to one of my own accounts that only has a small balance. This would mitigate the insecurity involved with having to have my bunch of next online all the time. In other words: Can I put my NXT in cold storage and be safe and at the same time have my forging power leased to an account I control and earn tx fees there?

Short answer: yes.

Except that Nxt doesn't have 'cold storage' in the sense that bitcoin has, of course. 'Cold storage' in Nxt terms would mean "having removed all possible online passphrase information", as we use a brainwallet.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 24, 2014, 11:52:42 pm
I also have a question: Is the following possible? I have a lot of nxt (say 5 percent of the whole stake). I then lease my forging power to one of my own accounts that only has a small balance. This would mitigate the insecurity involved with having to have my bunch of next online all the time. In other words: Can I put my NXT in cold storage and be safe and at the same time have my forging power leased to an account I control and earn tx fees there?

Short answer: yes.

Except that Nxt doesn't have 'cold storage' in the sense that bitcoin has, of course. 'Cold storage' in Nxt terms would mean "having removed all possible online passphrase information", as we use a brainwallet.
Ok. Thanks. I crossed out that disadvantage above and on the bitshares forum....
Do people actually use this option?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Damelon on July 24, 2014, 11:55:58 pm
Ok. Thanks. I crossed out that disadvantage on the bitshares forum....
Do people actually use this option?

Most people tend to use leasing for making it easier to forge with smaller accounts in leasing pools: https://nxtforum.org/leasing-pools/

http://wiki.nxtcrypto.org/wiki/Account_Leasing

It's altogether possible that people do this to protect their large accounts, though. Should be checkable on the blockchain :)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 25, 2014, 12:02:54 am
Aren't forging pools a threat to NXT/centralization?  For instance, I say "lease your forging power to me and I will send back 90% of the fees relative to the stake you leased so you don't need to bother with it."  Even if you limit it by address, I can set up multiple addresses to accept leased forging.  I can also set up pools that don't appear to be under the same ownership but actually are.

The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.  You have to rely on people being altruistic to avert this attack.  People have to reject the short term financial gain for the greater good of the NXT network... That's asking A LOT!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Damelon on July 25, 2014, 12:18:11 am
There are some interesting discussions on this in here: https://nxtforum.org/account-control/

Leasing is part of a larger project called "Account Control": http://wiki.nxtcrypto.org/wiki/Account_Control
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 25, 2014, 12:19:03 am
Aren't forging pools a threat to NXT/centralization?  For instance, I say "lease your forging power to me and I will send back 90% of the fees relative to the stake you leased so you don't need to bother with it."  Even if you limit it by address, I can set up multiple addresses to accept leased forging.  I can also set up pools that don't appear to be under the same ownership but actually are.
They dont have approval voting I think (regarding leasing; correct?) so putting up many pools and asking for their "vote" wouldn't make sense.
I also guess that the nxt stakeholders that join a pool expect to have the tx fees pay back almost completely anyway.
I would agree that this is a problem if there are not enough big nxt holders that are big enough so that they dont need a pool. That would keep the system relatively decentralized.

The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.  You have to rely on people being altruistic to avert this attack.  People have to reject the short term financial gain for the greater good of the NXT network... That's asking A LOT!
POW, POS and DPOS has that same problem. Stakeholders would have to come to the conclusion that they gain more by not making their stake worthless (which would happen if they followed such an overpaying pool) than by taking in a few more tx fees. There is a freerider problem though! It is dangerous. In DPOS stakeholders could come still come to the consensus (social consensus) that pay back pools/delegates are not acceptable in the first place which would make it much harder to set up an overpaying delegate compared to when pay back delegates are established / accepted.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 25, 2014, 12:40:10 am
Aren't forging pools a threat to NXT/centralization?  For instance, I say "lease your forging power to me and I will send back 90% of the fees relative to the stake you leased so you don't need to bother with it."  Even if you limit it by address, I can set up multiple addresses to accept leased forging.  I can also set up pools that don't appear to be under the same ownership but actually are.
They dont have approval voting I think (regarding leasing; correct?) so putting up many pools and asking for their "vote" wouldn't make sense.
I also guess that the nxt stakeholders that join a pool expect to have the tx fees pay back almost completely anyway.
I would agree that this is a problem if there are not enough big nxt holders that are big enough so that they dont need a pool. That would keep the system relatively decentralized.
delulo, you are wrong.  there is a very obvious incentive for pools.  The pool operator makes money (% of fees) and the leaser gets a consistent income without running their computer 24/7.
The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.  You have to rely on people being altruistic to avert this attack.  People have to reject the short term financial gain for the greater good of the NXT network... That's asking A LOT!
POW, POS and DPOS has that same problem. Stakeholders would have to come to the conclusion that they gain more by not making their stake worthless (which would happen if they followed such an overpaying pool) than by taking in a few more tx fees. There is a freerider problem though! It is dangerous. In DPOS stakeholders could come still come to the consensus (social consensus) that pay back pools/delegates are not acceptable in the first place which would make it much harder to set up an overpaying delegate compared to when pay back delegates are established / accepted.
Again delulo, you are wrong.  Approval voting solves this problem in DPOS; it is not the same as leasing.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 25, 2014, 12:54:10 am
Aren't forging pools a threat to NXT/centralization?  For instance, I say "lease your forging power to me and I will send back 90% of the fees relative to the stake you leased so you don't need to bother with it."  Even if you limit it by address, I can set up multiple addresses to accept leased forging.  I can also set up pools that don't appear to be under the same ownership but actually are.
They dont have approval voting I think (regarding leasing; correct?) so putting up many pools and asking for their "vote" wouldn't make sense.
I also guess that the nxt stakeholders that join a pool expect to have the tx fees pay back almost completely anyway.
I would agree that this is a problem if there are not enough big nxt holders that are big enough so that they dont need a pool. That would keep the system relatively decentralized.
delulo, you are wrong.  there is a very obvious incentive for pools.  The pool operator makes money (% of fees) and the leaser gets a consistent income without running their computer 24/7.
I didn't say it is not incentivizing pools. I just said it must not lead to a centralization (which you proposed if I understood you right), at least not a total centralization, if there are enough stakeholders that are big enough to not need to join a pool. If that is the case you will have the pools with the lowest fees plus the few stakeholders that don't need a pool.
How would it make sense to put up several pools in parallel and ask for votes without approval voting? (if that is what you suggested)

The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.  You have to rely on people being altruistic to avert this attack.  People have to reject the short term financial gain for the greater good of the NXT network... That's asking A LOT!
POW, POS and DPOS has that same problem. Stakeholders would have to come to the conclusion that they gain more by not making their stake worthless (which would happen if they followed such an overpaying pool) than by taking in a few more tx fees. There is a freerider problem though! It is dangerous. In DPOS stakeholders could come still come to the consensus (social consensus) that pay back pools/delegates are not acceptable in the first place which would make it much harder to set up an overpaying delegate compared to when pay back delegates are established / accepted.
Again delulo, you are wrong.  Approval voting solves this problem in DPOS; it is not the same as leasing.
How does it solve it?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 25, 2014, 01:03:46 am
How would it make sense to put up several pools in parallel and ask for votes without approval voting? (if that is what you suggested)
It makes sense because they have proposed limiting the size of pools as some kind of solution to their centralization problem.  (Leased stake above a certain % doesn't add to your forging power).  I'm just pointing out there is an easy way around their solution where pool operators will just use multiple addresses or multiple pools to maximize their profits and get around the limit.

Quote
Again delulo, you are wrong.  Approval voting solves this problem in DPOS; it is not the same as leasing.
How does it solve it?
Because with approval voting, voting for an attacker doesn't stop you from voting for all the good delegates.  Also the attacker almost certainly won't get enough support to compete to be a delegate in the first place.  So all the idiots voting for someone trying to attack their own network won't get paid anything for their ignorance.  (as long as he isn't a delegate he has no power or profits to share and will be ignored)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ThomasVeil on July 25, 2014, 01:18:05 am
Reply of the bitshares main dev (bytemaster, Daniel Larimer):
Quote
DPOS: Can be attacked by deception of majority, but easily fixed with a quick hard-fork to new delegates.

I'm not 100% clear on how hard-forks work. But do I see it correct here: If the votes are in, but the developer doesn't agree to the outcome, he just invalidates them?

I slowly get the idea behind DPOS, but stick to my critique: It's unnecessarily complicated. Adding social dynamics that are hard to predict.
Besides the higher barrier of entry, there is actually quite some "cognitive computation" needed, that Nick Szabo often talks about. Do small shareholders get enough incentive to pay the cognitive cost to vote? They get no extra payoff.

It also in a way is the idea everyone has at first: Give everyone equal power to every person instead of every coin. No one has a counter to the sybil attack problem though. So how does DPOS solve that?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 25, 2014, 01:46:32 am
Reply of the bitshares main dev (bytemaster, Daniel Larimer):
Quote
DPOS: Can be attacked by deception of majority, but easily fixed with a quick hard-fork to new delegates.
I'm not 100% clear on how hard-forks work. But do I see it correct here: If the votes are in, but the developer doesn't agree to the outcome, he just invalidates them?
Hard-fork is really just the theoretical ultimate protection but in practice highly unlikely to be needed.  In order to understand this, you have to understand that what gives a crypto-currency value is ultimately the broad market consensus around it.  An example of hard-fork protection would be if the US gov't decided to buy up over 51% of a POS crypto such as NXT in order to now be able to control it or 51% attack it (BTW for anyone who still believes NXT is proof against "90%" attack, tl:dr it's not.)  If they did this, at least with DPOS, it would be easy to identify which shares they own and the rest of the community could agree to just completely remove that stake via hard fork so now the US gov't doesn't own anything.  For an example of this, look how Vericoin hard forked to remove the stake of a hacker who stole a huge portion of total coins from an exchange.  It's not unique to bitshares, it also isn't up to the developer, it's up to the market to accept the corrected distribution.

Quote
I slowly get the idea behind DPOS, but stick to my critique: It's unnecessarily complicated. Adding social dynamics that are hard to predict.
Besides the higher barrier of entry, there is actually quite some "cognitive computation" needed, that Nick Szabo often talks about. Do small shareholders get enough incentive to pay the cognitive cost to vote? They get no extra payoff.

It also in a way is the idea everyone has at first: Give everyone equal power to every person instead of every coin. No one has a counter to the sybil attack problem though. So how does DPOS solve that?
BitShares gives equal power to shares, not people, so sybil attacks have nothing to do with it.  It's also not really complicated to vote, you just vote for various people who seem to be trustworthy (not sockpuppets) and capable of running a computer.  It's easy to check if delegates are doing their job, so it's not a big deal if you vote in a bad one, you'll see soon enough and he'll be removed.  If small shareholders don't vote it's not a real big deal because their stake doesn't matter much.  Large shareholders will take a couple seconds to do this.  Real companies work this same way, big shareholders care enough to vote.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 25, 2014, 11:05:23 am
How would it make sense to put up several pools in parallel and ask for votes without approval voting? (if that is what you suggested)
It makes sense because they have proposed limiting the size of pools as some kind of solution to their centralization problem.  (Leased stake above a certain % doesn't add to your forging power).  I'm just pointing out there is an easy way around their solution where pool operators will just use multiple addresses or multiple pools to maximize their profits and get around the limit.
It all depends on what you call centralized/decentralized. I assumed that there might even just be one pool as a result of economies of scale + big stakeholders that don't need a pool and I would have called that not totally centralized.
Does anybody know about how many forgers there are that are big enough to not forge in a pool? That would be interesting.

Quote
Again delulo, you are wrong.  Approval voting solves this problem in DPOS; it is not the same as leasing.
How does it solve it?
Because with approval voting, voting for an attacker doesn't stop you from voting for all the good delegates.  Also the attacker almost certainly won't get enough support to compete to be a delegate in the first place.  So all the idiots voting for someone trying to attack their own network won't get paid anything for their ignorance.  (as long as he isn't a delegate he has no power or profits to share and will be ignored)
I am not sure whether you are still talking about the attack you described in the first place of which I said POW, POS and DPOS all suffer from:
Quote
The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.
In DPOS an attacker would have to set up 52 "100% + x pay back delegates". If everyone voted for such a scheme because it is seen as legitimate or because people are ignoring the long term effects on the share price of this his 52 delegates will be in the top 101. In addition to everyone voting on these 52 delegates each shareholder would vote on different additional delegates but as everyone voted on the 52 delegates but on different other delegates the 52 would definitely be in. Now one can argue that it is more effort to vote for 52 delegates than to join one pool but that is not enough of an argument. In the end it all depends on the perception of shareholder who calculate their benefits and risks plus might have some social constraints which is why I tried to establish a non "pay back delegate culture" in the first place. If pay back delegates do not exist in the first place it would be harder to run with a pay back delegate or even with an 100%+x pay back delegate later.
POS and POW have the same problem though.
Shareholder perception matters.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Agent86 on July 25, 2014, 11:35:15 am
We'll have to agree to disagree delulo.  You are not understanding the difference in incentives.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 25, 2014, 11:37:30 am
Show me a thread / a post of yours that proves me wrong.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: mczarnek on July 25, 2014, 06:44:40 pm
So, let's say that Nxt caps the leasing power per forging pool.  In the simplest implementation, this means you can't lease to an account if your forging power plus theirs would push it over some amount.  I believe the number we came up with was 0.5%

Then the differences I see are:
DPOS essentially allows choosing multiple forgers to lease your forging power to
DPOS only allows the top 101 forgers in terms of leased forging power to forger

First of all, seem like simple enough changes and if the market decides it prefers that form.. easy enough for Nxt to switch over.  Just as much incentive for the big guys to have powerful computers to attractive leasers.  However, I personally prefer allowing the little guy to forge on his own if he would prefer to.

Regarding allowing leasing your forging power to multiple forgers.. I don't see a difference one way or another.  Because you give people more votes and it seems to me your average person will be more free with their votes.  Attacker still gets to forge every once in a while.

And Nxt could hardfork too and allow people to switch up their leasing power.. just doesn't seem like a good idea to even consider it to be a solution.  I'm pretty sure the attitude around here is that if you trust your Nxt to an exchange and the exchange gets hacked.. tough luck.

btw Nxt is not 90% attack resistant in it's current form but with upcoming changes, it should be before the end of the year... I think it's been planned for sometime in November.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: 2Kool4Skewl on July 25, 2014, 06:54:50 pm
I personally prefer allowing the little guy to forge on his own if he would prefer to.

+1  This allows the spirit of true decentralization.

There is no need for individuals to use pools except they get payouts faster.  Right now NXT is worth so little it doesn't really matter.  I can guarantee there will be more than 101 forgers in the NXT system.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 25, 2014, 09:46:45 pm
I also have a question: Is the following possible? I have a lot of nxt (say 5 percent of the whole stake). I then lease my forging power to one of my own accounts that only has a small balance. This would mitigate the insecurity involved with having to have my bunch of next online all the time. In other words: Can I put my NXT in cold storage and be safe and at the same time have my forging power leased to an account I control and earn tx fees there?

Short answer: yes.

Except that Nxt doesn't have 'cold storage' in the sense that bitcoin has, of course. 'Cold storage' in Nxt terms would mean "having removed all possible online passphrase information", as we use a brainwallet.
Ok. Thanks. I crossed out that disadvantage above and on the bitshares forum....
Do people actually use this option?

Indeed, they do.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 26, 2014, 01:52:03 am
Thanks for all the responses.

btw Nxt is not 90% attack resistant in it's current form but with upcoming changes, it should be before the end of the year... I think it's been planned for sometime in November.
Can anyone explain how that (the 90% attack resistance) would work?

Quote
I personally prefer allowing the little guy to forge on his own if he would prefer to.

+1  This allows the spirit of true decentralization.

There is no need for individuals to use pools except they get payouts faster.  Right now NXT is worth so little it doesn't really matter.  I can guarantee there will be more than 101 forgers in the NXT system.
How would you define "true decentralization"?

I think every consensus network (Bitcoin, NXT , Bitshares) has to centralize in the end under the conditions of market competition for low fees / customers.

Bitcoin/POW centralizes because miners migrate to the pool with the lowest fees. Atm this is limited only by a commitment of of Ghash to not take any new miners anymore.

For NXT: If crypto currencies have gotten beyond the speculation phase and enter a competition phase where different systems compete for users with low and lower fees all consensus networks are forced to offer similarly low fees (especially if micro payments and/or machine to machine payments matter). Let's assume in addition that in order to forge a forger needs professional equipment and a reliable internet connection so that the consensus network can guarantee high up time and many tx / second. These costs any forger / delegate has to pay are equal for every system. The consequence is that only those with large stakes can mine on their own, the others will have to join a pool that is equally big, because the costs for equipment are the same for everyone. This will result in centralization (to a certain degree) with NXT.  Here is a more detailed description https://bitsharestalk.org/index.php?topic=5564.0;all of this economies of scale phenomenon.

A System where delegation is possible would be more decentralized. But in practice because NXT allows pools it is essentially the same in terms of the potential to stay decentralized as DPOS. Based on the above paragraph the pools in NXT would, under harsh competition for the lowerst fees between different networks, have to be as big as the biggest stakeholders or a bit bigger because they have to charge some fees.

I also updated my assessment in the OP (Quote 1). Looking forward to your feedback. 
Title: Re: NXT POS vs. Bitshares DPOS
Post by: 2Kool4Skewl on July 26, 2014, 04:29:15 am
Thanks for all the responses.

btw Nxt is not 90% attack resistant in it's current form but with upcoming changes, it should be before the end of the year... I think it's been planned for sometime in November.
Can anyone explain how that (the 90% attack resistance) would work?

Quote
I personally prefer allowing the little guy to forge on his own if he would prefer to.

+1  This allows the spirit of true decentralization.

There is no need for individuals to use pools except they get payouts faster.  Right now NXT is worth so little it doesn't really matter.  I can guarantee there will be more than 101 forgers in the NXT system.
How would you define "true decentralization"?

I think every consensus network (Bitcoin, NXT , Bitshares) has to centralize in the end under the conditions of market competition for low fees / customers.

Bitcoin/POW centralizes because miners migrate to the pool with the lowest fees. Atm this is limited only by a commitment of of Ghash to not take any new miners anymore.

For NXT: If crypto currencies have gotten beyond the speculation phase and enter a competition phase where different systems compete for users with low and lower fees all consensus networks are forced to offer similarly low fees (especially if micro payments and/or machine to machine payments matter). Let's assume in addition that in order to forge a forger needs professional equipment and a reliable internet connection so that the consensus network can guarantee high up time and many tx / second. These costs any forger / delegate has to pay are equal for every system. The consequence is that only those with large stakes can mine on their own, the others will have to join a pool that is equally big, because the costs for equipment are the same for everyone. This will result in centralization (to a certain degree) with NXT.  Here is a more detailed description https://bitsharestalk.org/index.php?topic=5564.0;all of this economies of scale phenomenon.

A System where delegation is possible would be more decentralized. But in practice because NXT allows pools it is essentially the same in terms of the potential to stay decentralized as DPOS. Based on the above paragraph the pools in NXT would, under harsh competition for the lowerst fees between different networks, have to be as big as the biggest stakeholders or a bit bigger because they have to charge some fees.

I also updated my assessment in the OP (Quote 1). Looking forward to your feedback.

Let's take the example of a user running a business on NXT.  This business is going to need a "high speed" (NXT TF requires 1.5 Mbps, I'm not sure if that qualifies as "high speed") internet connection anyway to operate online.  NXT PoS requires very minimal processing power to sign transactions.  You can do this on extremely cheap hardware that the business would already have or need to acquire for normal operations anyway (web server).  You don't need "professional" hardware.  The only other cost is power which is negligible considering their hardware is going to be running 24/7 anyway (web server).  There is no reason for these businesses, who are incurring these costs from normal business operations anyway, to consolidate their power in a pool unless they want to reap their forging fees in a steady payout instead of one lump sum.  There is no other pressure to force individuals into pools in NXT.  Unlike bitcoin where the longer you solo mine the less likely you are to find a block because of the continual increase in hashpower on the network.  In NXT, their is no additional hashpower increase over time.

When the NXT network grows to have more than 101 different businesses operating on it, each forging individually because they are incurring the forging costs anyway from basic operations, then NXT will be more decentralized than Bitshares.  This isn't including all the other individuals who decided to forge on their own for other reasons, ideological or otherwise.  The more business that run on the NXT network the more decentralized it becomes.  This is not true with Bitshares.  Bitshares will only ever have 101 forgers regardless of the number of individuals or business running on its network.  It is "decentralize" impaired or restricted.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 26, 2014, 09:49:42 am
Aren't forging pools a threat to NXT/centralization?  For instance, I say "lease your forging power to me and I will send back 90% of the fees relative to the stake you leased so you don't need to bother with it."  Even if you limit it by address, I can set up multiple addresses to accept leased forging.  I can also set up pools that don't appear to be under the same ownership but actually are.

The even more sinister attack is I set up a pool that actually returns even more than 100% of fees.  I basically pay people to give me power over the network.  You have to rely on people being altruistic to avert this attack.  People have to reject the short term financial gain for the greater good of the NXT network... That's asking A LOT!

That is an interesting process you described.

There is something similar possible in PoW: Proof of Idle (you pay others to reduce their mining power) - http://www.peercointalk.org/index.php?topic=3086

If it could work, I do not know. Maybe it does. However, the timeframe is extremely narrow and expensive.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Come-from-Beyond on July 26, 2014, 09:51:06 am
Can anyone explain how that (the 90% attack resistance) would work?

I'm drawing comics to explain it.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: valarmg on July 26, 2014, 10:11:54 am

Let's take the example of a user running a business on NXT.  This business is going to need a "high speed" (NXT TF requires 1.5 Mbps, I'm not sure if that qualifies as "high speed") internet connection anyway to operate online.  NXT PoS requires very minimal processing power to sign transactions.  You can do this on extremely cheap hardware that the business would already have or need to acquire for normal operations anyway (web server).  You don't need "professional" hardware.  The only other cost is power which is negligible considering their hardware is going to be running 24/7 anyway (web server).  There is no reason for these businesses, who are incurring these costs from normal business operations anyway, to consolidate their power in a pool unless they want to reap their forging fees in a steady payout instead of one lump sum.  There is no other pressure to force individuals into pools in NXT.  Unlike bitcoin where the longer you solo mine the less likely you are to find a block because of the continual increase in hashpower on the network.  In NXT, their is no additional hashpower increase over time.

When the NXT network grows to have more than 101 different businesses operating on it, each forging individually because they are incurring the forging costs anyway from basic operations, then NXT will be more decentralized than Bitshares.  This isn't including all the other individuals who decided to forge on their own for other reasons, ideological or otherwise.  The more business that run on the NXT network the more decentralized it becomes.  This is not true with Bitshares.  Bitshares will only ever have 101 forgers regardless of the number of individuals or business running on its network.  It is "decentralize" impaired or restricted.

Exactly. Every business that uses NXT will already have a node running 24/7. The forging fees are negligible for most people, so they won't care about maximizing that by leasing (and it doesn't even maximize, just makes payouts more consistent).

For example, jl777 is designing a system have will involve a network of decentralized servers for his InstantDex/Privacy services. Each one will be a Nxt node. That's just a single Nxt business and already hundreds of nodes.

There's no reason there won't be hundreds of thousands of Nxt nodes once Nxt takes off.

What seems to have been lost in this discussion (talking about Nxt becoming centralized in large leasing pools) is that forging fees are not going to be the main incentive to run Nxt nodes. The main incentive will be the advantages that using Nxt provides. Most Nxt nodes won't be run because of fees. It'll be because traders are trading on the MGW, or a businesses are accepting Nxt payments, etc. The forging fees are a nice bonus but for most nodes, they won't be the reason the node is run, and thus they won't care about leasing to the biggest pool to get an extra few pence a day or whatever.

Title: Re: NXT POS vs. Bitshares DPOS
Post by: liondani on July 26, 2014, 10:24:46 am


  If small shareholders don't vote it's not a real big deal because their stake doesn't matter much.  Large shareholders will take a couple seconds to do this.  Real companies work this same way, big shareholders care enough to vote.

makes sense!
It is easier for big share holders to vote for security reasons than to get technically involved by them self.  It is not a big deal now because the biggest shareholders are from the crypto community  but that is going to change when mass adoption will take place.  The demographics will change. So it will be much easier for large shareholders to let third trusted party's to secure the network then by they own, since I guess the most of them will not have the IT knowledge and the willingness to setup a super server even if that is a vps. They will prefer to vote a combination of delegates that are suggested for example from future trusted Rating Agencies ...! (?)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: liondani on July 26, 2014, 12:05:06 pm
When the NXT network grows to have more than 101 different businesses operating on it, each forging individually because they are incurring the forging costs anyway from basic operations, then NXT will be more decentralized than Bitshares.  This isn't including all the other individuals who decided to forge on their own for other reasons, ideological or otherwise.  The more business that run on the NXT network the more decentralized it becomes.  This is not true with Bitshares.  Bitshares will only ever have 101 forgers regardless of the number of individuals or business running on its network.  It is "decentralize" impaired or restricted.

So you say that bitshares would get less decentralized than NXT in future because of the 101 delegates restriction...
So what would you say if in future the number of delegates on the bitshares network would dynamically change depended on business that run each time?
More business? More delegates. Less business? Less delegates (minimum 101).
If that would happen how would you rate BitSharesX compared vs NXT? Which of them  would you prefer then and why?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 26, 2014, 12:19:20 pm
Thanks for all the responses.

btw Nxt is not 90% attack resistant in it's current form but with upcoming changes, it should be before the end of the year... I think it's been planned for sometime in November.
Can anyone explain how that (the 90% attack resistance) would work?

Quote
I personally prefer allowing the little guy to forge on his own if he would prefer to.

+1  This allows the spirit of true decentralization.

There is no need for individuals to use pools except they get payouts faster.  Right now NXT is worth so little it doesn't really matter.  I can guarantee there will be more than 101 forgers in the NXT system.
How would you define "true decentralization"?

I think every consensus network (Bitcoin, NXT , Bitshares) has to centralize in the end under the conditions of market competition for low fees / customers.

Bitcoin/POW centralizes because miners migrate to the pool with the lowest fees. Atm this is limited only by a commitment of of Ghash to not take any new miners anymore.

For NXT: If crypto currencies have gotten beyond the speculation phase and enter a competition phase where different systems compete for users with low and lower fees all consensus networks are forced to offer similarly low fees (especially if micro payments and/or machine to machine payments matter). Let's assume in addition that in order to forge a forger needs professional equipment and a reliable internet connection so that the consensus network can guarantee high up time and many tx / second. These costs any forger / delegate has to pay are equal for every system. The consequence is that only those with large stakes can mine on their own, the others will have to join a pool that is equally big, because the costs for equipment are the same for everyone. This will result in centralization (to a certain degree) with NXT.  Here is a more detailed description https://bitsharestalk.org/index.php?topic=5564.0;all of this economies of scale phenomenon.

A System where delegation is possible would be more decentralized. But in practice because NXT allows pools it is essentially the same in terms of the potential to stay decentralized as DPOS. Based on the above paragraph the pools in NXT would, under harsh competition for the lowerst fees between different networks, have to be as big as the biggest stakeholders or a bit bigger because they have to charge some fees.

I also updated my assessment in the OP (Quote 1). Looking forward to your feedback.

Let's take the example of a user running a business on NXT.  This business is going to need a "high speed" (NXT TF requires 1.5 Mbps, I'm not sure if that qualifies as "high speed") internet connection anyway to operate online.  NXT PoS requires very minimal processing power to sign transactions.  You can do this on extremely cheap hardware that the business would already have or need to acquire for normal operations anyway (web server).  You don't need "professional" hardware.  The only other cost is power which is negligible considering their hardware is going to be running 24/7 anyway (web server).  There is no reason for these businesses, who are incurring these costs from normal business operations anyway, to consolidate their power in a pool unless they want to reap their forging fees in a steady payout instead of one lump sum.  There is no other pressure to force individuals into pools in NXT.  Unlike bitcoin where the longer you solo mine the less likely you are to find a block because of the continual increase in hashpower on the network.  In NXT, their is no additional hashpower increase over time.
At the moment you are right. Like I said my assumption is visa scale network sizes and hard market competition. Then the forgers in the network have to be able to guarantee high up time, high bandwidth and it also takes a lot of high end (=expensive) CPU, memory.

When the NXT network grows to have more than 101 different businesses operating on it, each forging individually because they are incurring the forging costs anyway from basic operations, then NXT will be more decentralized than Bitshares.  This isn't including all the other individuals who decided to forge on their own for other reasons, ideological or otherwise.  The more business that run on the NXT network the more decentralized it becomes.  This is not true with Bitshares.  Bitshares will only ever have 101 forgers regardless of the number of individuals or business running on its network.  It is "decentralize" impaired or restricted.
But you assume that the businesses (i guess you mean e.g. a trusted issuers of NXT assets, a lottery company based on nxt, a DNS register; what else did you think of as an example?) also have the biggest NXT stakes. Why should that be?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 26, 2014, 12:27:17 pm
Can anyone explain how that (the 90% attack resistance) would work?

I'm drawing comics to explain it.

Looking forward to your art work!

Quote
What seems to have been lost in this discussion (talking about Nxt becoming centralized in large leasing pools) is that forging fees are not going to be the main incentive to run Nxt nodes. The main incentive will be the advantages that using Nxt provides. Most Nxt nodes won't be run because of fees. It'll be because traders are trading on the MGW, or a businesses are accepting Nxt payments, etc. The forging fees are a nice bonus but for most nodes, they won't be the reason the node is run, and thus they won't care about leasing to the biggest pool to get an extra few pence a day or whatever.
As for validating your own tx that is right. But we are talking about forgers, those that find blocks. The number of nodes doesn't matter for a 51% attack if they do not find blocks often enough = have a big stake.
I always talked about centralization of forging/block producing power which is the one that matters to control the network (revise tx, just mess with it due to political reasons). 
Title: Re: NXT POS vs. Bitshares DPOS
Post by: valarmg on July 26, 2014, 01:26:19 pm
As for validating your own tx that is right. But we are talking about forgers, those that find blocks. The number of nodes doesn't matter for a 51% attack if they do not find blocks often enough = have a big stake.
I always talked about centralization of forging/block producing power which is the one that matters to control the network (revise tx, just mess with it due to political reasons). 
But all nodes are potentially a node that will forge a block (providing they haven't leased their power), no? It all depends on the amount of stake and amount of luck. What is the distinction you are drawing between forgers and nodes?


At the moment you are right. Like I said my assumption is visa scale network sizes and hard market competition. Then the forgers in the network have to be able to guarantee high up time, high bandwidth and it also takes a lot of high end (=expensive) CPU, memory.
Why is the high end CPU required? Also, I don't think always-on high bandwidth is necessarily required. If a node is down, then the network will move to the next one. That node will lose it's forging power, but it shouldn't be a big deal for the network.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 26, 2014, 06:25:12 pm
As for validating your own tx that is right. But we are talking about forgers, those that find blocks. The number of nodes doesn't matter for a 51% attack if they do not find blocks often enough = have a big stake.
I always talked about centralization of forging/block producing power which is the one that matters to control the network (revise tx, just mess with it due to political reasons). 
But all nodes are potentially a node that will forge a block (providing they haven't leased their power), no? It all depends on the amount of stake and amount of luck. What is the distinction you are drawing between forgers and nodes?
You said it right, "It all depends on the amount of stake and amount of luck". All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.

At the moment you are right. Like I said my assumption is visa scale network sizes and hard market competition. Then the forgers in the network have to be able to guarantee high up time, high bandwidth and it also takes a lot of high end (=expensive) CPU, memory.
Why is the high end CPU required? Also, I don't think always-on high bandwidth is necessarily required. If a node is down, then the network will move to the next one. That node will lose it's forging power, but it shouldn't be a big deal for the network.
If you would want to process say >100 tx per second and have 10 second block times you would not only need high bandwidth, reliability / up-time (determining that someone missed his block with TF costs time and makes the network less predictable which might not be acceptable under high market competition) as well as a lot of CPU and memory although I can not assess at which point (how many tx per sec etc.) the CPU/memory begins to play a role.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 26, 2014, 08:17:43 pm
If you would want to process say >100 tx per second and have 10 second block times you would not only need high bandwidth, reliability / up-time (determining that someone missed his block with TF costs time and makes the network less predictable which might not be acceptable under high market competition) as well as a lot of CPU and memory although I can not assess at which point (how many tx per sec etc.) the CPU/memory begins to play a role.

1000 transactions are not an issue for today's processors. Even less for those to come. Within several years, that will be peanuts. I suspect that even smartphones will be able to handle that without effort.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on July 26, 2014, 10:03:31 pm
If you would want to process say >100 tx per second and have 10 second block times you would not only need high bandwidth, reliability / up-time (determining that someone missed his block with TF costs time and makes the network less predictable which might not be acceptable under high market competition) as well as a lot of CPU and memory although I can not assess at which point (how many tx per sec etc.) the CPU/memory begins to play a role.

1000 transactions are not an issue for today's processors. Even less for those to come. Within several years, that will be peanuts. I suspect that even smartphones will be able to handle that without effort.

CPU speed is not really the limiting constraint. Memory capacity (need more RAM), memory bandwidth (need more CPUs), and network bandwidth (need access to bigger cables) are the real killers.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: valarmg on July 26, 2014, 10:19:15 pm
You said it right, "It all depends on the amount of stake and amount of luck". All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.

A possible way to solve that would be to create a portal site. Those with high capacity nodes point their server at the portal, and those with Nxt to lease point their Nxt at the portal. The portal ensures that the leased Nxt is distributed throughout the network of high capacity nodes, creating a robust decentralized network.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 26, 2014, 11:38:14 pm
CPU speed is not really the limiting constraint. Memory capacity (need more RAM), memory bandwidth (need more CPUs), and network bandwidth (need access to bigger cables) are the real killers.

I do not see killers here. Memory is cheap and is getting cheaper the minute I write this. My home desktop scans (verifying + building up internal datastructures) the entire blockchain (200,000 blocks and 300,000 transactions) within 8 seconds. I consider my desktop mediocre (right now) if not low-end within 10 months. Bandwidth is the only thing I worry about a little bit.

However, considering the growth rate, we will not outpace the real-world technology and infrastructure.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Daedelus on July 27, 2014, 10:30:56 am
You said it right, "It all depends on the amount of stake and amount of luck". All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.

A possible way to solve that would be to create a portal site. Those with high capacity nodes point their server at the portal, and those with Nxt to lease point their Nxt at the portal. The portal ensures that the leased Nxt is distributed throughout the network of high capacity nodes, creating a robust decentralized network.

+1, I saw you mention this in another thread. I like the idea but sounds technically difficult in practice given the 1440 wait time for moving forging power. The portal won't be able to re-assign forging power on the fly in real time. And the portal can't just forge with the whole stake and send fees out to the nodes in proportion, that woul be cetralisation into one pool.

Am I  understanding correctly? How could this be solved?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: valarmg on July 27, 2014, 11:09:32 am
You said it right, "It all depends on the amount of stake and amount of luck". All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.

A possible way to solve that would be to create a portal site. Those with high capacity nodes point their server at the portal, and those with Nxt to lease point their Nxt at the portal. The portal ensures that the leased Nxt is distributed throughout the network of high capacity nodes, creating a robust decentralized network.

+1, I saw you mention this in another thread. I like the idea but sounds technically difficult in practice given the 1440 wait time for moving forging power. The portal won't be able to re-assign forging power on the fly in real time. And the portal can't just forge with the whole stake and send fees out to the nodes in proportion, that woul be cetralisation into one pool.

Am I  understanding correctly? How could this be solved?
I see what you mean. There are definitely problems with this idea.

Regarding the assigning of forging power.
One solution is that the portal assigns the lease directly to one server's address. User clicks on leasing portal, says how much it wants to lease and for how long, portal uses an algorithm to determine which server address should be used and tells user to lease to that address.

Another problem is ensuring that the servers pay back their forged Nxt to the leasee (or wherever the leasee wants the forged Nxt to go). Might mean that servers have to leave a deposit with the portal. Or maybe servers only receive a small amount of leased power until they prove that they are trustworthy. It's not like each individual block receives a lot of Nxt.


Title: Re: NXT POS vs. Bitshares DPOS
Post by: Brangdon on July 27, 2014, 11:28:46 am
I also have a question: Is the following possible? I have a lot of nxt (say 5 percent of the whole stake). I then lease my forging power to one of my own accounts that only has a small balance. This would mitigate the insecurity involved with having to have my bunch of next online all the time. In other words: Can I put my NXT in cold storage and be safe and at the same time have my forging power leased to an account I control and earn tx fees there?
That's roughly what I do. I have three accounts. One is a savings account, one a current account and one a forging account. The savings account is most secure; it's not actually offline yet (the default client doesn't support offline transactions well today) but the password is more buried and rarely used. That one contains the bulk of my coins. The forging account is least secure because it runs on AWS cloud and I don't trust Amazon sysops not to steal my NXT. It only holds a few coins. Both the savings account and the current account lease to the forging account.

This works pretty well. Leasing is a general purpose tool for managing how you arrange your accounts and how you forge.

Does anybody know about how many forgers there are that are big enough to not forge in a pool? That would be interesting.
Kind-of everyone is? Forging in a pool actually makes less money, because there are transaction fees first to lease and second to withdraw funds. If your income from forging is likely to be small compared to the fees, but you happen to have a computer online 24/7, it's probably better not to lease. So it isn't really to do with the size of the account, but how cheap bandwidth is for you.

All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.
Everyone can be a node, and everyone can be a forger, but not every forger is individually significant.

I added the "individually", because if we end up with a Pareto distribution (http://en.wikipedia.org/wiki/Pareto_distribution) with, say, 80% of NXT spread between accounts with, say, 10K NXT each, then individually each of those accounts doesn't amount to much, but all of them together make 51% attacks impossible.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: devphp on July 27, 2014, 11:41:51 am
Does anybody know about how many forgers there are that are big enough to not forge in a pool? That would be interesting.

300-350 accounts have successfuly forged blocks over the past 30 days.

This code can be used to select public keys of successful forgers from the blockchain.

Code: [Select]
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.ResultSet;
import java.sql.SQLException;
import java.sql.Statement;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.Iterator;

public class inspectDB {

public static void runQuery(int startHeight, int endHeight) {

Connection conn = null;

String dbUrl = "jdbc:h2:nxt_db/nxt;DB_CLOSE_ON_EXIT=FALSE";
dbUrl += ";DB_CLOSE_ON_EXIT=FALSE";

try {

Class.forName("org.h2.Driver");

conn = DriverManager.getConnection(dbUrl, "sa", "sa");

Statement stmt = conn.createStatement();
// generator_public_key
ArrayList<String> pubgenkeys = new ArrayList<String>();
String stmtStr = "SELECT generator_public_key, height, timestamp, total_amount FROM block where total_amount >= 0 and height > "
+ startHeight + " and height < " + endHeight;
ResultSet selectRS = stmt.executeQuery(stmtStr);// total_amount
while (selectRS.next()) {
String a = selectRS.getString(1);
pubgenkeys.add(a);

}

Iterator<String> it = pubgenkeys.iterator();
HashMap<String, Integer> occ = new HashMap<String, Integer>();
while (it.hasNext()) {
String cur = it.next();
// System.out.println(cur);
if (occ.containsKey(cur)) {
int last = occ.get(cur).intValue();
occ.remove(cur);
occ.put(cur, new Integer(last + 1));
} else
occ.put(cur, 1);
}

int numBlocks = endHeight - startHeight;
int numAddr = occ.values().size();

System.out.println("startBlock  " + startHeight
+ " generated by # addresses : " + numAddr
+ " (numBlocks: " + numBlocks + ")");

conn.close();

} catch (SQLException e) {
e.printStackTrace();
} catch (ClassNotFoundException e) {

e.printStackTrace();
} finally {
if (conn != null)
try {
conn.close();
} catch (SQLException e) {
e.printStackTrace();
}
}

}

public static void main(String[] args) {
int maxh = 200000;
int step = 10000;
for (int bh = 0; bh < maxh; bh += step) {
runQuery(bh, bh + step);
}
}
}

https://nxtblocks.info/#section/blockexplorer_charts
this graph has a 'Others' section that according to the blockchain represents 300+ successfully forging accounts.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 27, 2014, 09:16:51 pm
If you would want to process say >100 tx per second and have 10 second block times you would not only need high bandwidth, reliability / up-time (determining that someone missed his block with TF costs time and makes the network less predictable which might not be acceptable under high market competition) as well as a lot of CPU and memory although I can not assess at which point (how many tx per sec etc.) the CPU/memory begins to play a role.

1000 transactions are not an issue for today's processors. Even less for those to come. Within several years, that will be peanuts. I suspect that even smartphones will be able to handle that without effort.

CPU speed is not really the limiting constraint. Memory capacity (need more RAM), memory bandwidth (need more CPUs), and network bandwidth (need access to bigger cables) are the real killers.
It seems difficult to assess how much of a role hardware and bandwidth and reliability costs play. It is an open question. Maybe one of the bitshares delegates can make a precise calculation.

@ChuckOne, what did you mean by this?
Quote
However, considering the growth rate, we will not outpace the real-world technology and infrastructure.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 27, 2014, 09:29:28 pm
You said it right, "It all depends on the amount of stake and amount of luck". All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.

A possible way to solve that would be to create a portal site. Those with high capacity nodes point their server at the portal, and those with Nxt to lease point their Nxt at the portal. The portal ensures that the leased Nxt is distributed throughout the network of high capacity nodes, creating a robust decentralized network.

Don't know if that is technically possible but the result would be pretty close to my scenario for NXT POS + possibility for delegation/pools which was: Big stakholders that can froge without pools + one or two big pools with the rest of the leased stake.

I have a question though: You said:

One other thing to throw into the mix, the current plan is that Nxt is going to not allow leasors to lease to anyone with more than 0.5% of the forging power.  Meaning that it should be harder for anyone person to get a large percentage of the forging power.  Basically the same thing, right?

If I get that right, it is only possible, for a big stakeholder, to either be able to lease all his stake to an account hi controls as well so he can put his NXT in cold storage OR to limit the size of pools one can delegate to to xy%.  By cold storage I mean what ever is possible with NXT (atm), for example what Brangdon suggested "the password is more buried and rarely used."
This would also effect the scenario above. 
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 27, 2014, 09:45:28 pm
Does anybody know about how many forgers there are that are big enough to not forge in a pool? That would be interesting.
Kind-of everyone is? Forging in a pool actually makes less money, because there are transaction fees first to lease and second to withdraw funds. If your income from forging is likely to be small compared to the fees, but you happen to have a computer online 24/7, it's probably better not to lease. So it isn't really to do with the size of the account, but how cheap bandwidth is for you.
Agreed! Whether this stays this way would depend on the outcome of the forging costs and rewards under conditions of market competition (competing for customers with low tx fees) and increased traffic and increased need for reliability. As said above, this is an open question.
 
All nodes receive transactions and can check whether they are unspent outputs but they don't have to forge (my definition of forging/mining: Selecting which transactions are included in the next block and selecting which previous block to reference. Agreed?). Everyone can be a node but a node can only be a significant forger if he has a significant stake.
Everyone can be a node, and everyone can be a forger, but not every forger is individually significant.
Well said!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 27, 2014, 09:46:48 pm
@ChuckOne, what did you mean by this?
Quote
However, considering the growth rate, we will not outpace the real-world technology and infrastructure.

Well, even though Nxt is PoS and thus requires a minimum, when scaling up the number of transactions, more will be demanded from the existing infrastructure and technology (resources).

However, Nxt is not growing at high speed but steadily, so does the demand.

On the other hand side, technology and existing infrastructure and technology is supposed to grow with or without Nxt. In any case, Nxt's demand will not outpace the available resources.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 27, 2014, 09:50:00 pm
Agreed! Whether this stays this way would depend on the outcome of the forging costs and rewards under conditions of market competition (competing for customers with low tx fees) and increased traffic and increased need for reliability. As said above, this is an open question.

It is an open question, indeed and time will tell. However, future hardware and infrastructure will provide cheap means to do the stuff on the side.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on July 27, 2014, 09:51:30 pm
Agreed! Whether this stays this way would depend on the outcome of the forging costs and rewards under conditions of market competition (competing for customers with low tx fees) and increased traffic and increased need for reliability. As said above, this is an open question.

It is an open question, indeed and time will tell. However, future hardware and infrastructure will provide cheap means to do the stuff on the side.
I linked your argument here https://bitsharestalk.org/index.php?topic=5564.new#new
Maybe someone with more of an insight into hardware requirements than me can come up with a well grounded estimate.

I also updated my assessment in the OP. I appreciate your feedback!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on July 27, 2014, 11:18:53 pm
@ChuckOne, what did you mean by this?
Quote
However, considering the growth rate, we will not outpace the real-world technology and infrastructure.

Well, even though Nxt is PoS and thus requires a minimum, when scaling up the number of transactions, more will be demanded from the existing infrastructure and technology (resources).

However, Nxt is not growing at high speed but steadily, so does the demand.

On the other hand side, technology and existing infrastructure and technology is supposed to grow with or without Nxt. In any case, Nxt's demand will not outpace the available resources.
Well, if I dont implement some sort of aggregation and InstantDEX tradebots go crazy, it could start pushing the current tx limits...
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on July 28, 2014, 06:43:23 am
Well, if I dont implement some sort of aggregation and InstantDEX tradebots go crazy, it could start pushing the current tx limits...

Do it. ;)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 01, 2014, 07:00:41 pm
One Question: who gets the fees in Nxt:  a) the forger who included the block or b) all forgers share all fees equally (based on their stake).
Title: Re: NXT POS vs. Bitshares DPOS
Post by: devphp on August 01, 2014, 07:06:48 pm
One Question: who gets the fees in Nxt:  a) the forger who included the block or b) all forgers share all fees equally (based on their stake).

a)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 01, 2014, 07:10:57 pm
Both are roughly the same.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 01, 2014, 07:13:14 pm
One Question: who gets the fees in Nxt:  a) the forger who included the block or b) all forgers share all fees equally (based on their stake).

a)
Ok. Thanks. That is what I guessed...
Title: Re: NXT POS vs. Bitshares DPOS
Post by: allwelder on August 02, 2014, 01:10:58 am
BTS---In practice BitShares isn't a distributed system, it is more like coordinated synchronized time shared centralized system. ---BM.
https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829 (https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829)

NXT---fully decentralized.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 04, 2014, 10:44:12 am
BTS---In practice BitShares isn't a distributed system, it is more like coordinated synchronized time shared centralized system. ---BM.
https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829 (https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829)

NXT---fully decentralized.
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: devphp on August 04, 2014, 10:48:16 am
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.

Can one person run more than one delegate in Bitshares? How long will it take before there are 101 different people each running one delegate? If only 20 people run 101 delegates (each averaging to 5), that's not very decentralized. What are the economic incentives to run a delegate in Bitshares? I am just trying to understand it better, need answers.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 04, 2014, 11:07:50 am
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.

Can one person run more than one delegate in Bitshares? How long will it take before there are 101 different people each running one delegate? If only 20 people run 101 delegates (each averaging to 5), that's not very decentralized. What are the economic incentives to run a delegate in Bitshares? I am just trying to understand it better, need answers.
One person can run more than 1 delegate. But he would make that public then. Delegates are mostly people from the forum or publicly known individuals like Adam B Levine. So it would be unlikely to get votes from shareholders if you set up a sock puppet account on the forum and campaign with this to be a delegate of which people don't know that is is run by the same individual that also runs another delegates under another forum name. So it is possible for 20 people to run the 101 delegates if the shareholders vote this way.
The question is whether this is more centralized than the few major stakeholders that find most of the blocks in NXT. It was discussed here before whether economies of scale will lead to centralization of NXT forgers, see https://nxtforum.org/general-discussion/nxt-pos-vs-bitshares-dpos/msg71586/#msg71586 We agreed that it is an open question. You might also want to look at this https://bitsharestalk.org/index.php?topic=5564.0;all (the OP + some new arguments have been added recently).

Economic incentives are the tx fees the delegates - the same with NXT.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: devphp on August 04, 2014, 11:17:40 am
One person can run more than 1 delegate. But he would make that public then. Delegates are mostly people from the forum or publicly known individuals like Adam B Levine. So it would be unlikely to get votes from shareholders if you set up a sock puppet account on the forum and campaign with this to be a delegate of which people don't know that is is run by the same individual that also runs another delegates under another forum name. So it is possible for 20 people to run the 101 delegates if the shareholders vote this way.
The question is whether this is more centralized than the few major stakeholders that find most of the blocks in NXT. It was discussed here before whether economies of scale will lead to centralization of NXT forgers, see https://nxtforum.org/general-discussion/nxt-pos-vs-bitshares-dpos/msg71586/#msg71586 We agreed that it is an open question. You might also want to look at this https://bitsharestalk.org/index.php?topic=5564.0;all (the OP + some new arguments have been added recently).

Economic incentives are the tx fees the delegates - the same with NXT.

Well, the difference is, DPOS doesn't make a conscious effort to make it more decentralized, while NXT does. And as distribution of NXTs among users gets better with time (less whales, more dolphins), the number of stakeholders who forge blocks should increase too. There is also a correlation to the price of NXT, higher price incentivizes more NXTers to forge.

In Bitshares DPOS, which I find interesting and worthy of attention too, developers sort of admitted their defeat as far as decentralization aspect goes, and set themselves on a predefined number of delegates, which will probably work out to an even smaller number of real persons behind those delegates, since there is an issue of trust as you say.

Tx fees in Bitshares are very low too, from that comparative table I've seen, there is no way tx fees alone can be an incentive at this point, until there are significantly more txs.

Of course, it's still too early to make final judgements and conclusions, let's wait a year, it should get more clear then.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: allwelder on August 04, 2014, 01:18:23 pm
BTS---In practice BitShares isn't a distributed system, it is more like coordinated synchronized time shared centralized system. ---BM.
https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829 (https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829)

NXT---fully decentralized.
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.
So,BM,the inventor of BTS is just kidding according to you.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 04, 2014, 02:57:52 pm
BTS---In practice BitShares isn't a distributed system, it is more like coordinated synchronized time shared centralized system. ---BM.
https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829 (https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829)

NXT---fully decentralized.
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.

As far as I remember, the delegates can be anonymous, too. People just say that it will be unlikely because other will not vote for them.

So, it is basically the perception only. We in Nxt could do the same as well and educate people not to lease out to anonymous people. But I think that is not an issue currently.

Am I right about this assumption?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 04, 2014, 07:25:24 pm
Of course, it's still too early to make final judgements and conclusions, let's wait a year, it should get more clear then.

Agree 100%

I am not here to promote anything. I wanted to spark of a quality discussion an analyze the two approaches as rational as possible.

I am a bitshares as well as a NXT holder.... I am mostly on the bitshares forum because you can find a lot of critical and constructive discussions. I appreciate the difference of perspectives of different communities!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 04, 2014, 07:29:53 pm
BTS---In practice BitShares isn't a distributed system, it is more like coordinated synchronized time shared centralized system. ---BM.
https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829 (https://bitsharestalk.org/index.php?topic=5402.msg72829#msg72829)

NXT---fully decentralized.
It is not that simple and black and white. In the end there is probably not that much difference in terms of decentralization.
As far as I remember, the delegates can be anonymous, too. People just say that it will be unlikely because other will not vote for them.
So, it is basically the perception only. We in Nxt could do the same as well and educate people not to lease out to anonymous people. But I think that is not an issue currently.
Am I right about this assumption?
Your assumption that delegates can be anonymous is correct! I assume though that shareholders will value and trust those delegates more that are out there with their real identity. It is a market for votes...
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 04, 2014, 07:37:21 pm
Thank you. That makes even more sense now.

As far as the spirit of Nxt is concerned (being a multi-layer system), having real-world identities published could be another level on top of the completely anonymous system now.

Do you think Nxt could benefit from such increased transparency about who are the hub owners?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: eimon on August 04, 2014, 09:07:06 pm
I wouldn't bother to vote..
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 04, 2014, 09:11:22 pm
I wouldn't bother to vote..

How come?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: eimon on August 04, 2014, 09:24:31 pm
I wouldn't bother to vote..

How come?

Who has the time to vote on who forges the blocks for their cryptos? I'd rather just leave it up to an automated system and avoid the chore of researching a bunch of campaigning forgers to figure out which ones are the most trustworthy.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 04, 2014, 09:26:23 pm
How would such automated system work?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: eimon on August 04, 2014, 09:36:09 pm
How would such automated system work?

Isn't that how POS works? How is the NXT forging process not automated?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 04, 2014, 10:20:05 pm
Here is how I see it:
NXT started out as a non-delegation Proof of Stake (see my OP, it is a comparison of delegated vs. non delegates POS), meaning it was not possible to lease your forging power to a pool. Then NXT has added the leasing feature, I figure because it is more secure to not vote with the account where your whole stake is + because some forgers have a too small stake to take the hassle having a machine online 24/7.
Now with NXT which pool you lease to depends on how much the pool you lease to gives you as a kickback. With DPOS such pools (in DPOS called "delegates) are the default option but like with NXT you can also vote for your own delegate and get all the tx fees (equals a stakeholder in NXT that leases to himself). So basically it is just that the default options are twisted for NXT and Bitshares. Therefore saying one system is by default completely centralized and the other one is not is not that realistic. Some degree of centralization is unavoidable with Bitcoin, NXT or Bitshares (economies of scale is the reason: We discussed this before and came to the conclusion that it is difficult to assess how big the effect is). Both systems are quite similar actually. As the delegation/leasing is the default option there is a lot you can do with it apart from determining who is forging: You can vote for delegates that not only are trustworthy to forge blocks, you can also vote for delegates that promise to develope the ecosystem with the tx fees those delegates get.

If delegates /pool operators are publicly known with their real world identity they have more of an incentive to stay honest - they have more to loose.
 
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 23, 2014, 11:07:57 pm
IDK if anyone posted this excerpt yet but I feel like it is critical:

wiki (dot) bitshares (dot) org/index.php/DPOS#Scalability

tl;dr  ALL consensus protocols centralize at scale. How well do you manage it? Does yours squeeze out small players (or do they have to do it at a loss like with BTC or soon PPC... and eventually NXT?)

Having even 1 delegate would still be "decentralized" if stakeholders had instant reaction time. 101 is just an arbitrary number where we guess the tradeoff where 1% marginal decentralization is not worth the extra incurred cost at the network.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: 2Kool4Skewl on August 23, 2014, 11:46:50 pm
IDK if anyone posted this excerpt yet but I feel like it is critical:

wiki (dot) bitshares (dot) org/index.php/DPOS#Scalability

tl;dr  ALL consensus protocols centralize at scale. How well do you manage it? Does yours squeeze out small players (or do they have to do it at a loss like with BTC or soon PPC... and eventually NXT?)

Having even 1 delegate would still be "decentralized" if stakeholders had instant reaction time. 101 is just an arbitrary number where we guess the tradeoff where 1% marginal decentralization is not worth the extra incurred cost at the network.

This "theory" is based on the fact that all validators need to make a profit off forging to continue forging.  This is flawed.  People will run nodes not because they will make a profit from forging but because they have a business that requires NXT to operate and that business makes a profit.

Why are you posting this in multiple threads?  Are you a Bitshill?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 12:07:18 am
IDK if anyone posted this excerpt yet but I feel like it is critical:

wiki (dot) bitshares (dot) org/index.php/DPOS#Scalability

tl;dr  ALL consensus protocols centralize at scale. How well do you manage it? Does yours squeeze out small players (or do they have to do it at a loss like with BTC or soon PPC... and eventually NXT?)

Having even 1 delegate would still be "decentralized" if stakeholders had instant reaction time. 101 is just an arbitrary number where we guess the tradeoff where 1% marginal decentralization is not worth the extra incurred cost at the network.

This "theory" is based on the fact that all validators need to make a profit off forging to continue forging.  This is flawed.  People will run nodes not because they will make a profit from forging but because they have a business that requires NXT to operate and that business makes a profit.

Why are you posting this in multiple threads?  Are you a Bitshill?

Yes, I am a bitshill.

Fine, suppose they don't need to make a profit, then neither does any other system. Point is, somebody has to incur the costs, and eventually the number of people that are willing to incur the costs will decrease. All such systems centralize.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ThomasVeil on August 24, 2014, 12:22:41 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

Quote
then neither does any other system.

It depends on the costs. Bitcoin has massive costs and rising. So they do need high fees. They've put themselves in a corner.
Normal POS forging costs are negligible. Why would two dollar per month loss matter?

Quote
Point is, somebody has to incur the costs, and eventually the number of people that are willing to incur the costs will decrease.

Why? Is there any evidence or even logic behind this?
Evidence to the contrary: The number of forgers for NXT has risen so far.
Logic to the contrary: If I can forge on any machine, and I have to run the software anyways for normal usage - then everyone can forge. Should that alt be used in every store, then basically every cash register can forge.

Payoff: Even without fees, a working safe system has huge benefits.

Quote
All such systems centralize.

Why? What is "such systems"? It always depends on lots of factors.
Nature for example is full of wonderfully decentralized systems.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: 2Kool4Skewl on August 24, 2014, 12:52:25 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

(https://encrypted-tbn3.gstatic.com/images?q=tbn:ANd9GcQODv_yLmyrRItGPekz7eh4W82ddRrY1sRf9RVvuQf-FP8CrnkdOQ)
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 12:58:06 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

Suppose they don't need to make an "explicit" profit but it needs to be worth their while to be running the node. The number of people for whom it is worth their while will decrease.

Quote
Quote
Point is, somebody has to incur the costs, and eventually the number of people that are willing to incur the costs will decrease.

Why? Is there any evidence or even logic behind this?
Evidence to the contrary: The number of forgers for NXT has risen so far.
Logic to the contrary: If I can forge on any machine, and I have to run the software anyways for normal usage - then everyone can forge. Should that alt be used in every store, then basically every cash register can forge.

Payoff: Even without fees, a working safe system has huge benefits.
Cost to the network scales linearly with the number of validators. Your evidence doesn't mean much because none of these systems are at any scale worth talking about. Even bitcoin is only at around 5 tps, I'm talking about on the order of 2,000 tps.

Quote
Quote
All such systems centralize.

Why? What is "such systems"? It always depends on lots of factors.
Nature for example is full of wonderfully decentralized systems.
All consensus protocols where you must validate each transaction. Nature has plenty of decentralized systems which do not need to reach consensus about a shared state.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: 2Kool4Skewl on August 24, 2014, 01:11:59 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

Suppose they don't need to make an "explicit" profit but it needs to be worth their while to be running the node. The number of people for whom it is worth their while will decrease.

It's worth their while to forge because they have to run a node to process business transactions.

Bitshares is intentionally capping the network at 101 forgers.  Why would you impose a self-limitation?  It seems like flawed thinking.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 01:53:29 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

Suppose they don't need to make an "explicit" profit but it needs to be worth their while to be running the node. The number of people for whom it is worth their while will decrease.

It's worth their while to forge because they have to run a node to process business transactions.

Bitshares is intentionally capping the network at 101 forgers.  Why would you impose a self-limitation?  It seems like flawed thinking.

For every increase in cost there is a decrease in number of willing forgers. At some scale there it is more expensive to run a node than the risk premium for outsourcing the job of full validation. We just have not reached that scale yet so it is not visible - heck you can even run a bitcoin full node for $40/month.

At some bandwidth/RAM requirement level it switches from "intentionally capping" to "intentionally decentralizing past what you need". Look at the graphs I posted above, you can only argue "it will never reach a scale/cost ratio to where 101 is the sweet spot" and not "that trend doesn't happen".

That said, if it turns out 101 is not the sweet spot, it could be changed. We have no reason to think it is greater than 101 and in fact I suspect it is less. Again, having a single forger would still be decentralized if you assumed very active stakeholders.

Right now BTC has effectively 2 forgers that control the network, but people still call it decentralized. Why? Because it's a social consensus and not a technological one. All you need is for a single block producer to be able to suddenly stop producing blocks and for the network to recover from it.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 01:59:50 am
Fine, suppose they don't need to make a profit,

That is the stated core reason for making this whole thing up though.

Suppose they don't need to make an "explicit" profit but it needs to be worth their while to be running the node. The number of people for whom it is worth their while will decrease.

It's worth their while to forge because they have to run a node to process business transactions.

Bitshares is intentionally capping the network at 101 forgers.  Why would you impose a self-limitation?  It seems like flawed thinking.

For every increase in cost there is a decrease in number of willing forgers. At some scale there it is more expensive to run a node than the risk premium for outsourcing the job of full validation. We just have not reached that scale yet so it is not visible - heck you can even run a bitcoin full node for $40/month.

At some bandwidth/RAM requirement level it switches from "intentionally capping" to "intentionally decentralizing past what you need". Look at the graphs I posted above, you can only argue "it will never reach a scale/cost ratio to where 101 is the sweet spot" and not "that trend doesn't happen".

That said, if it turns out 101 is not the sweet spot, it could be changed. We have no reason to think it is greater than 101 and in fact I suspect it is less. Again, having a single forger would still be decentralized if you assumed very active stakeholders.

Right now BTC has effectively 2 forgers that control the network, but people still call it decentralized. Why? Because it's a social consensus and not a technological one. All you need is for a single block producer to be able to suddenly stop producing blocks and for the network to recover from it.
I think it would be more productive to just stick to the facts, instead of injecting subjective assessments into the evaluations. Then everyone can make up their minds on whether it is decentralized enough for them. Some people actually like things that are centralized.

I bolded the parts I felt were commentary and not factual. Also the BTC PoW flaw are not really relevant, are they? Clearly almost anything is more decentralized than Ghash.iocoin (aka BTC)

so just describe how many delegates (101), how many different people are they (?), etc.

Also I dont understand how the activity of the stakeholders would allow a system with one forger to be decentralized. I am thinking maybe we have different definitions for such words?

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 02:32:51 am
Cost to the network scales linearly with the number of validators. Your evidence doesn't mean much because none of these systems are at any scale worth talking about. Even bitcoin is only at around 5 tps, I'm talking about on the order of 2,000 tps.

Weird guy indeed... What is he thinking... ;)

TheOnion
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 02:42:37 am
By my definition "decentralization" is a continuous metric which describes how directly the 51% majority stakeholders can preserve their property rights against individuals who can be coerced.

The "one delegate is still decentralized" depends heavily on the assumption that active stakeholders are in fact very active to the point where it is unrealistic. I do not actually think 1 delegate would be decentralized because stakeholders do not have reaction time less than 10 seconds, it is just to emphasize that at the end of the day it is about how directly do the 51% stakeholders control who the block producers are. What is the "churn rate" for bad delegates, how much redundancy do you need?  (actually you would need 2 delegates at least if the bad one starts blocking transactions... but I think you see the point I'm making).

The previous point I was making is that *at scale*, there will not be many distinct full validating nodes. In bitcoin there are fewer than 5k full nodes, and bitcoin processes like 2tps and has only 15gb blockchain.

With DPOS it is economically feasible for 51% stake which is distributed among 1,000 individuals to overpower a single 10% stakeholder. This is also economically feasible in NXT with leased forging! It is only NOT economically feasible in PPC style POS where there is no stake delegation. So actually my argument is not that NXT preserves decentralization worse at scale, I am first just trying to get people to admit that all consensus systems lead to centralization and that accounting for centralized entities is required. With NXT it is an ad-hoc process where stakeholders have to monitor forging pools.

Assuming I can get people to agree this far then the only difference between NXT and DPOS is the voting mechanism, direct vs approval voting. This is basically, "how effectively can you keep bad stake from producing blocks".

tl;dr I am not arguing that DPOS is significantly better than NXT as much as I am trying to get you guys to admit that 101 distinct block producers is not more centralized than NXT at scale.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 03:03:35 am
By my definition "decentralization" is a continuous metric which describes how directly the 51% majority stakeholders can preserve their property rights against individuals who can be coerced.

The "one delegate is still decentralized" depends heavily on the assumption that active stakeholders are in fact very active to the point where it is unrealistic. I do not actually think 1 delegate would be decentralized because stakeholders do not have reaction time less than 10 seconds, it is just to emphasize that at the end of the day it is about how directly do the 51% stakeholders control who the block producers are. What is the "churn rate" for bad delegates, how much redundancy do you need?  (actually you would need 2 delegates at least if the bad one starts blocking transactions... but I think you see the point I'm making).

The previous point I was making is that *at scale*, there will not be many distinct full validating nodes. In bitcoin there are fewer than 5k full nodes, and bitcoin processes like 2tps and has only 15gb blockchain.

With DPOS it is economically feasible for 51% stake which is distributed among 1,000 individuals to overpower a single 10% stakeholder. This is also economically feasible in NXT with leased forging! It is only NOT economically feasible in PPC style POS where there is no stake delegation. So actually my argument is not that NXT preserves decentralization worse at scale, I am first just trying to get people to admit that all consensus systems lead to centralization and that accounting for centralized entities is required. With NXT it is an ad-hoc process where stakeholders have to monitor forging pools.

Assuming I can get people to agree this far then the only difference between NXT and DPOS is the voting mechanism, direct vs approval voting. This is basically, "how effectively can you keep bad stake from producing blocks".

tl;dr I am not arguing that DPOS is significantly better than NXT as much as I am trying to get you guys to admit that 101 distinct block producers is not more centralized than NXT at scale.
I just realized the title of this thread is quite directly comparing the two systems. Have you checked the size of the largest forging pool? Even though it is theoretically possible that some forging pool could get to 51% of NXT, this wont happen. i think currently the largest pool has around 8% , maybe a bit more, but I think this is because one of the 5% stakeholders is using it. if any pool starts getting above 10%, I will start making noises. If it gets to 20%, I will probably sell all my NXT and encourage everybody else to. So forging wont ever get to 50%, or if it does, it wont be my concern anymore. Due to marketcap and liquidity, just buying up 50% NXT is not feasible, so I am comfortable with NXT forging process.

relying on people to vote properly seems quite dangerous. I point to any election as my proof :)

So, DPOS using the wisdom of people voting to ensure good results within a max of 101 forging nodes, as opposed to NXT that has no need for this and also no limit to the number of forging nodes, though practically speaking at the current distribution clumpage it is around 300 nodes that are able to forge and probably the top 100 forge most of the blocks.

Do the 100 DPOS nodes all have equal chance to make the next block? If so, at current levels (ignoring the wisdom of popular voting results), the two systems appear to be roughly the same, at least from the view of how many nodes are eligible to forge the next block.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 03:52:32 am
I think you have it about right. When the number of pools is high ("high" to me is over 20) then it doesn't really matter. When the number of pools is low, approval voting and the very explicit "one human runs more than one delegate? vote em out!" gives DPOS a slight advantage in my eyes. The tradeoff is the perception of centralization and more visible / "exposed" block producers.

edit:
Quote
Do the 100 DPOS nodes all have equal chance to make the next block? If so, at current levels (ignoring the wisdom of popular voting results), the two systems appear to be roughly the same, at least from the view of how many nodes are eligible to forge the next block.

yes, they are shuffled every round
Title: Re: NXT POS vs. Bitshares DPOS
Post by: toast on August 24, 2014, 03:56:38 am
BTW  do any of you want to join a dev hangout to discuss this with Dan?  He has a better understanding of the nuances of different consensus models than I do. Some guy on our forum organizes them:  http://beyondbitcoinx.net/

Contact "fuznuts" on our forum if you're down
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 04:04:50 am
I think you have it about right. When the number of pools is high ("high" to me is over 20) then it doesn't really matter. When the number of pools is low, approval voting and the very explicit "one human runs more than one delegate? vote em out!" gives DPOS a slight advantage in my eyes. The tradeoff is the perception of centralization and more visible / "exposed" block producers.

edit:
Quote
Do the 100 DPOS nodes all have equal chance to make the next block? If so, at current levels (ignoring the wisdom of popular voting results), the two systems appear to be roughly the same, at least from the view of how many nodes are eligible to forge the next block.

yes, they are shuffled every round
each of the large stakeholders sort of acts like a pool, but really no practical way to go above 3% to 5% chance, toss in a few leased pools of 5% or 8% and then a hundred+ with small chances, but on aggregate maybe 15% to 25%, and without any need for human monitoring/voting. I prefer the NXT approach (especially at larger scales) for the blockchain, but to each their own. Some people even like soy milk!

By being more centralized, btsx probably gets more speed, at least until I get InstantDEX done. We can debate which is better, etc. but the numbers tell me that NXT is already a bit more decentralized than bitshares and will continue to be more so as time passes. Add in the human factor needed and bitshares is not directly comparable to NXT, but still bitshares is much closer to NXT than most other things, so let us consider each other the crazy cousin up in the attic :)

I think that we can agree that regardless of which is "better", either is miles ahead of Ghash.iocoin
I also think the human factor which is not desirable for blockchain matters is quite suitable and even preferable for assets. I assume bitshares will have assets? Would there be a way to easily make an asset gateway between the two systems so that the same asset can be on either?

I think that would benefit both

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 04:07:10 am
BTW  do any of you want to join a dev hangout to discuss this with Dan?  He has a better understanding of the nuances of different consensus models than I do. Some guy on our forum organizes them:  http://beyondbitcoinx.net/

Contact "fuznuts" on our forum if you're down
I dont do realtime voice. my speech synthesizer is hard to operate at full speed, so it is better to use messages.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: lucky331 on August 24, 2014, 04:36:21 am
Quote
Would there be a way to easily make an asset gateway between the two systems so that the same asset can be on either?

yes please. that would be awesooome! 
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 04:48:25 am

By being more centralized, btsx probably gets more speed, at least until I get InstantDEX done. We can debate which is better, etc. but the numbers tell me that NXT is already a bit more decentralized than bitshares and will continue to be more so as time passes. Add in the human factor needed and bitshares is not directly comparable to NXT, but still bitshares is much closer to NXT than most other things, so let us consider each other the crazy cousin up in the attic :)


I like the ever softening tone ever since I stared posting (in that regard here). I knew, it has to happen... and I like it! +5%  +1440

Happy Onion!  :)  here!
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 05:15:30 am
Wow, this is one of those best threads. Thank you delulo for starting this thread. Your comments are neutral and objective. The discussion save me a lot of energy in collecting information, reading, learning and thinking.


As of the 90% attack resistant feature, I think it is buried in different threads. Combining the following threads together should give the whole picture (AFAIK).

description about Transparent Forging
https://bitcointalk.org/index.php?topic=364218.0

description about Economic Clustering
https://nxtforum.org/index.php?topic=1635.0
https://nxtforum.org/index.php?topic=1691.0

and some discussions about the so-called Nothing at Stake and some materials mentioned there.
https://nxtforum.org/index.php?topic=3343.0
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 05:24:33 am
About the pools/delegates, I think the biggest difference lies in that in Nxt, the choice is in the hand of the stakers, they are not forced to lease (or vote for delegates), they have the choice of forging on their own or leasing to a pool, and in Nxt, the number of pools are determined automatically by the market instead of being some arbitrarily predefined number.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 05:25:43 am

By being more centralized, btsx probably gets more speed, at least until I get InstantDEX done. We can debate which is better, etc. but the numbers tell me that NXT is already a bit more decentralized than bitshares and will continue to be more so as time passes. Add in the human factor needed and bitshares is not directly comparable to NXT, but still bitshares is much closer to NXT than most other things, so let us consider each other the crazy cousin up in the attic :)


I like the ever softening tone ever since I stared posting (in that regard here). I knew, it has to happen... and I like it! +5%  +1440

Happy Onion!  :)  here!
we are just not liking aggressive marketing stuff that plays loose with facts. As long as it is a fact based discussion, then there is always common ground.

I keep seeing that +5%? What does that mean?

Also, do you have a URL to where I can get API for btsx?

James

P.S. it also seems that btsx is stabilizing right at $50 mil marketcap. Infusion? Coincidence? full moon?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 05:33:17 am
About the pools/delegates, I think the biggest difference lies in that in Nxt, the choice is in the hand of the stakers, they are not forced to lease (or vote for delegates), they have the choice of forging on their own or leasing to a pool, and in Nxt, the number of pools are determined automatically by the market instead of being some arbitrarily predefined number.
yes, no human factor, other than people deciding to use a pool. Since there is no hardware needed to run a pool, it is hard for any company to get some significant competitive advantage over the others, lessening the chance of a ghash.io Though their dominance comes from their cex.io trading market that they operate without commissions. So, it gets lots of trading, bots, etc. It also lets them just sell new shares whenever they want. This allows them nearly unlimited capital to buy more hash power. It seems none of the other pools is doing anything like this, at least not the last time I checked.

the few small hashing assets in NXT AE is one vector that can impact ghash.io's dominance.

Back to the thread's topic...

In a few years, when most of the big stakeholders have reduced their holdings to the ~10 million NXT level, there really wont be any 5% accounts anymore. I foresee a lot of different businesses in the 1 to 10 million NXT level and we could see 500+ different accounts actively forging. A few more years and as the average NXT total reduces more (due to commerce, varying fortunes, etc.) if the average business acct ends up in the 1 million NXT range, then we can see 1000 active forging accts, and so on.

So as the crypto economy grows, NXT is designed to get more and more active forging accts. So btsx uses the magic hand of the market to create "holographic" pegs, NXT uses the magic hand of the market to create the forging system.

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 05:51:52 am

By being more centralized, btsx probably gets more speed, at least until I get InstantDEX done. We can debate which is better, etc. but the numbers tell me that NXT is already a bit more decentralized than bitshares and will continue to be more so as time passes. Add in the human factor needed and bitshares is not directly comparable to NXT, but still bitshares is much closer to NXT than most other things, so let us consider each other the crazy cousin up in the attic :)


I like the ever softening tone ever since I stared posting (in that regard here). I knew, it has to happen... and I like it! +5%  +1440

Happy Onion!  :)  here!
we are just not liking aggressive marketing stuff that plays loose with facts. As long as it is a fact based discussion, then there is always common ground.


James

P.S. it also seems that btsx is stabilizing right at $50 mil marketcap. Infusion? Coincidence? full moon?

>>As long as it is a fact based discussion, then there is always common ground.

For us too.

>>I keep seeing that +5%? What does that mean?

+5% is the same as +1440 in (NXT terms); generally +1/ me too/etc.

>>Also, do you have a URL to where I can get API for btsx?

Ask 'toast' for the API, but generally the code is here (I know you are playing dumb, but anyway):
https://github.com/BitShares/bitshares_toolkit


Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 06:09:17 am
Quote
2.1) Delegates do not have the same interest in the success of the system as shareholders that are forgers when delegation is not possible.
2.2) An attacker could put up delegates and deceive shareholders. If the attacker manages to put up 50+% of the delegates by letting them gain the trust of shareholders over a long time and then he can perform one big attack after which he would be voted out. This would be cheaper than buying 51 % of the stake but maybe deceiving shareholders on such a scale is not possible at all.

I really have very bad feelings about these.

Nxt bases its security on its algorithm and leasing is only introduced later. While BTS uses these things as the base to protect the security of the system. It introduces trust between people in its first step. The different interests of the delegates and the stakeholders in 2.1 and the trust of the stakeholders in delegates needed in 2.2 make me think whether it is Proof of Stake any more. It sounds more like a Proof of Trust system. In this way, why are we making cryptocurrencies in the first place? We can just vote bankers as our delegates, they are faster, and if one banker fail, we vote another.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: blackyblack1 on August 24, 2014, 06:16:25 am
In a few years, when most of the big stakeholders have reduced their holdings to the ~10 million NXT level, there really wont be any 5% accounts anymore. I foresee a lot of different businesses in the 1 to 10 million NXT level and we could see 500+ different accounts actively forging. A few more years and as the average NXT total reduces more (due to commerce, varying fortunes, etc.) if the average business acct ends up in the 1 million NXT range, then we can see 1000 active forging accts, and so on.
No big businesses won't do forging. I requires dedicated host, cold storage and has less than 1% yearly interest.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 06:22:16 am
In a few years, when most of the big stakeholders have reduced their holdings to the ~10 million NXT level, there really wont be any 5% accounts anymore. I foresee a lot of different businesses in the 1 to 10 million NXT level and we could see 500+ different accounts actively forging. A few more years and as the average NXT total reduces more (due to commerce, varying fortunes, etc.) if the average business acct ends up in the 1 million NXT range, then we can see 1000 active forging accts, and so on.
No big businesses won't do forging. I requires dedicated host, cold storage and has less than 1% yearly interest.
really?
I think that most would, as all NXT based businesses have to run a NXT node, otherwise they arent a NXT business. So what is the incremental cost of clicking the button to start forging? THAT is the cost of forging for a business that is already running a NXT node. It doesnt make sense to count the entire cost of running the node as a forging cost when they need to do that anyway. Do you know of any business that will not click the start forging button to get incremental revenues, no matter how small? If so, that is a poorly run business

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 06:26:17 am

By being more centralized, btsx probably gets more speed, at least until I get InstantDEX done. We can debate which is better, etc. but the numbers tell me that NXT is already a bit more decentralized than bitshares and will continue to be more so as time passes. Add in the human factor needed and bitshares is not directly comparable to NXT, but still bitshares is much closer to NXT than most other things, so let us consider each other the crazy cousin up in the attic :)


I like the ever softening tone ever since I stared posting (in that regard here). I knew, it has to happen... and I like it! +5%  +1440

Happy Onion!  :)  here!
we are just not liking aggressive marketing stuff that plays loose with facts. As long as it is a fact based discussion, then there is always common ground.


James

P.S. it also seems that btsx is stabilizing right at $50 mil marketcap. Infusion? Coincidence? full moon?

>>As long as it is a fact based discussion, then there is always common ground.

For us too.

>>I keep seeing that +5%? What does that mean?

+5% is the same as +1440 in (NXT terms); generally +1/ me too/etc.

>>Also, do you have a URL to where I can get API for btsx?

Ask 'toast' for the API, but generally the code is here (I know you are playing dumb, but anyway):
https://github.com/BitShares/bitshares_toolkit
thanks! I have been too lazy to figure out where it was :)
Maybe you can have someone come and post how best to do the AE <-> AE dual gateway. I can of course do the NXT side, maybe someone from bitshares can do the bitshares side? I just dont have much time to wade through another codebase to find the right way to do the dual gateway

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: blackyblack1 on August 24, 2014, 06:30:16 am
Good business will be having effective balance close to zero because of constantly moving funds. And they'd prefer having client without anything, just transferring money from one point to another.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 06:38:05 am
Good business will be having effective balance close to zero because of constantly moving funds. And they'd prefer having client without anything, just transferring money from one point to another.
This is an interesting definition. I guess that Apple is not a good business.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: blackyblack1 on August 24, 2014, 06:39:55 am
That is effective balance in NXT terms. Balance having 1440+ confirmations.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 06:44:15 am
That is effective balance in NXT terms. Balance having 1440+ confirmations.
What makes you believe that Nxt businesses will not have some effective balance of Nxt as Apple does to fiat?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 06:46:24 am
thanks! I have been too lazy to figure out where it was :)
Maybe you can have someone come and post how best to do the AE <-> AE dual gateway. I can of course do the NXT side, maybe someone from bitshares can do the bitshares side? I just dont have much time to wade through another codebase to find the right way to do the dual gateway

James

Sure, the same way BTC sent their people for their side, right?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: blackyblack1 on August 24, 2014, 06:49:28 am
That is effective balance in NXT terms. Balance having 1440+ confirmations.
What makes you believe that Nxt businesses will not have some effective balance of Nxt as Apple does to fiat?
I actually do not know what is the bank balance of Apple. Fiat does not loose it's "effectiveness" due to rotation and NXT does.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 06:52:26 am
That is effective balance in NXT terms. Balance having 1440+ confirmations.
What makes you believe that Nxt businesses will not have some effective balance of Nxt as Apple does to fiat?
I actually do not know what is the bank balance of Apple. Fiat does not loose it's "effectiveness" due to rotation and NXT does.
You didn't answer my question
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 07:07:43 am
thanks! I have been too lazy to figure out where it was :)
Maybe you can have someone come and post how best to do the AE <-> AE dual gateway. I can of course do the NXT side, maybe someone from bitshares can do the bitshares side? I just dont have much time to wade through another codebase to find the right way to do the dual gateway

James

Sure, the same way BTC sent their people for their side, right?
yes, I worked with satoshi, he was very helpful
are you saying bitshares wont do the same?
much disappointment

James
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 24, 2014, 07:14:32 am
Wow, this is one of those best threads. Thank you delulo for starting this thread. Your comments are neutral and objective. The discussion save me a lot of energy in collecting information, reading, learning and thinking.


As of the 90% attack resistant feature, I think it is buried in different threads. Combining the following threads together should give the whole picture (AFAIK).

description about Transparent Forging
https://bitcointalk.org/index.php?topic=364218.0

description about Economic Clustering
https://nxtforum.org/index.php?topic=1635.0
https://nxtforum.org/index.php?topic=1691.0

and some discussions about the so-called Nothing at Stake and some materials mentioned there.
https://nxtforum.org/index.php?topic=3343.0

What an honor. I am greatful for the discussion here.

With regard to the 90 % attack I posted here: https://bitcointalk.org/index.php?topic=364218.msg8508849#msg8508849 Maybe someone has an answer there.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 07:26:59 am
thanks! I have been too lazy to figure out where it was :)
Maybe you can have someone come and post how best to do the AE <-> AE dual gateway. I can of course do the NXT side, maybe someone from bitshares can do the bitshares side? I just dont have much time to wade through another codebase to find the right way to do the dual gateway

James

Sure, the same way BTC sent their people for their side, right?
yes, I worked with satoshi, he was very helpful
are you saying bitshares wont do the same?
much disappointment

James

No, to the contrary... he will  work with you again... :)

Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 07:27:27 am
With regard to the 90 % attack I posted here: https://bitcointalk.org/index.php?topic=364218.msg8508849#msg8508849 Maybe someone has an answer there.

About that question, I think the detection part has already been implemented
Added first part of Economic Clustering - fork detection. After DGS block, each
transaction will include a reference to a recent blockId. For now this is used
only to detect forks, and transactions refering to a different fork are still
not rejected.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 24, 2014, 07:36:11 am
Quote
2.1) Delegates do not have the same interest in the success of the system as shareholders that are forgers when delegation is not possible.
2.2) An attacker could put up delegates and deceive shareholders. If the attacker manages to put up 50+% of the delegates by letting them gain the trust of shareholders over a long time and then he can perform one big attack after which he would be voted out. This would be cheaper than buying 51 % of the stake but maybe deceiving shareholders on such a scale is not possible at all.

I really have very bad feelings about these.

Nxt bases its security on its algorithm and leasing is only introduced later. While BTS uses these things as the base to protect the security of the system. It introduces trust between people in its first step. The different interests of the delegates and the stakeholders in 2.1 and the trust of the stakeholders in delegates needed in 2.2 make me think whether it is Proof of Stake any more. It sounds more like a Proof of Trust system. In this way, why are we making cryptocurrencies in the first place? We can just vote bankers as our delegates, they are faster, and if one banker fail, we vote another.
Note, that my comparison is about forging vs. not-forging.
If NXT would have no leasing it would suffer from 1.2) and 1.4). If there is both, the possibility to forge and to lease as it is atm, I guess the network will converge to one method for block production (forging or forging via leasing) at scale. No empirical evidence yet though for this.
Remember that leasing = delegation.
As for trust:
- Bitcoiners have to trust the pool operators
- NXTers have to trust the forging pools
- Bitshares holders have to trust delegates
The one power the coin holders have is to vote the pools/delegates out as soon as they misbehave. The goal is to make this as fast and outomated as possible.
On the other side there is a sense of trustlessness with all these system: You don't have to entrust your banker with your money/private keys.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: delulo on August 24, 2014, 07:40:21 am
With regard to the 90 % attack I posted here: https://bitcointalk.org/index.php?topic=364218.msg8508849#msg8508849 Maybe someone has an answer there.

About that question, I think the detection part has already been implemented
Added first part of Economic Clustering - fork detection. After DGS block, each
transaction will include a reference to a recent blockId. For now this is used
only to detect forks, and transactions refering to a different fork are still
not rejected.
Didn't get how this should work :( Maybe we can exchange arguments on BTT as it is another topic?
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 07:53:10 am
That is effective balance in NXT terms. Balance having 1440+ confirmations.
What makes you believe that Nxt businesses will not have some effective balance of Nxt as Apple does to fiat?
I actually do not know what is the bank balance of Apple. Fiat does not loose it's "effectiveness" due to rotation and NXT does.
You didn't answer my question

Can you put your questions in laymen's terms Evan?
So I can attempt answering at the 'high level' you deserve.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 07:58:30 am
Quote
2.1) Delegates do not have the same interest in the success of the system as shareholders that are forgers when delegation is not possible.
2.2) An attacker could put up delegates and deceive shareholders. If the attacker manages to put up 50+% of the delegates by letting them gain the trust of shareholders over a long time and then he can perform one big attack after which he would be voted out. This would be cheaper than buying 51 % of the stake but maybe deceiving shareholders on such a scale is not possible at all.

I really have very bad feelings about these.

Nxt bases its security on its algorithm and leasing is only introduced later. While BTS uses these things as the base to protect the security of the system. It introduces trust between people in its first step. The different interests of the delegates and the stakeholders in 2.1 and the trust of the stakeholders in delegates needed in 2.2 make me think whether it is Proof of Stake any more. It sounds more like a Proof of Trust system. In this way, why are we making cryptocurrencies in the first place? We can just vote bankers as our delegates, they are faster, and if one banker fail, we vote another.
Note, that my comparison is about forging vs. not-forging.
If NXT would have no leasing it would suffer from 1.2) and 1.4). If there is both, the possibility to forge and to lease as it is atm, I guess the network will converge to one method for block production (forging or forging via leasing) at scale. No empirical evidence yet though for this.
Remember that leasing = delegation.
As for trust:
- Bitcoiners have to trust the pool operators
- NXTers have to trust the forging pools
- Bitshares holders have to trust delegates
The one power the coin holders have is to vote the pools/delegates out as soon as they misbehave. The goal is to make this as fast and outomated as possible.
On the other side there is a sense of trustlessness with all these system: You don't have to entrust your banker with your money/private keys.

I think the point is that leasing is just a choice in Nxt while delegation is a must in BTS. So Nxt leasing is not exactly equal to BTS delegation. The freedom to choose makes the trust in pools just an option for Nxters. The security of the network is protected by the algorithm not by leasing. In BTS, the trust is a must since you have to vote one delegate or the other. The delegation is a part of the network security for BTS.

Nxters can lease to themselves to forge, thus avoid 1.2 and forge on their own at the same time.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 08:02:11 am
Didn't get how this should work :( Maybe we can exchange arguments on BTT as it is another topic?
Sorry I cannot discuss that with you. I don't know enough.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: tonyk on August 24, 2014, 08:04:37 am
@ Evan
read it, how you want to read it.

Cost to the network scales linearly with the number of validators. Your evidence doesn't mean much because none of these systems are at any scale worth talking about. Even bitcoin is only at around 5 tps, I'm talking about on the order of 2,000 tps.

Weird guy indeed... What is he thinking... ;)

TheOnion


Title: Re: NXT POS vs. Bitshares DPOS
Post by: Evan on August 24, 2014, 08:19:23 am
Can you put your questions in laymen's terms Evan?
So I can attempt answering at the 'high level' you deserve.
Sorry, I need to get back to work soon, don't have much time.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ThomasVeil on August 24, 2014, 10:46:50 am
DPOS is a solution in search of a problem. I have yet to see any factual argument why normal POS is not even better. All assumptions (i.e. "the network can't scale") are so far baseless - they already fall apart on face value:
- NXT has more nodes, and the number is growing
- running a node has lots of benefits beyond fees
- voting has a bad track record - Bitcoin was in 51% danger for weeks (was over several times), nothing happened. Still now 3 pools are enough to could collude.
- even the core philosophy "All such systems centralize." has zero merits. Please provide facts or research if you have any.
Again: Nature is full of decentralized systems ... which do in fact reach consensus about a shared state.


To talk about 2000 tps then as argument is a bit silly. Should we ever get there, then that would mean: Much more distribution. Much more potential forgers, much more interest in a stable system. Much higher computation speeds. And: Likely vastly evolved cryptos. Just look at what happened in happened in the last 6 months. Who knows what's going on in 10 years - if we're so lucky reach 2k tps by then.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Brangdon on August 24, 2014, 11:51:51 am
The previous point I was making is that *at scale*, there will not be many distinct full validating nodes. In bitcoin there are fewer than 5k full nodes, and bitcoin processes like 2tps and has only 15gb blockchain.
That's partly because in Bitcoin, mining is separate to running a full node. Running a full node helps with redundancy, but doesn't make the network more secure cryptographically in the same way that mining does, so there's less reason to do it. At the same time, mining is expensive so not many people get involved with that, either.

In Nxt, running a forging node has a greater benefit; it helps to secure the network so there is more reason to do it. Currently, running a forging node does not cost significantly more than running a full node. I don't know how much scale will change that. I would expect that even at 7,000 transactions a second, there will be far more than 101 companies that can handle that volume.

In a few years, when most of the big stakeholders have reduced their holdings to the ~10 million NXT level, there really wont be any 5% accounts anymore.
Bter had control of 5%. That probably won't change, so long as we have exchanges. I agree there won't be many accounts like that.

Good business will be having effective balance close to zero because of constantly moving funds.
It doesn't work like that. If you have a balance of 1000 NXT, and someone sends you a payment of 100 NXT, and you then pay 50 NXT to a supplier, your forging power is 1000, not 950 (or 1050). In other words, although you can't forge with the 100 NXT you received until 1440 blocks have passed, you can spend it, and the spending comes out of the newly arrived, non-forging coins rather than the old, forging ones. So as long as you keep your balance above 1000 NXT, you'll have 1000 NXT of forging power, even if transaction are coming in and out.

I think most companies try to keep a minimum balance, as a reserve so they don't risk going over-drawn if transactions happen in the wrong order.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Brangdon on August 24, 2014, 12:56:35 pm
IDK if anyone posted this excerpt yet but I feel like it is critical:

wiki (dot) bitshares (dot) org/index.php/DPOS#Scalability

tl;dr  ALL consensus protocols centralize at scale. How well do you manage it? Does yours squeeze out small players (or do they have to do it at a loss like with BTC or soon PPC... and eventually NXT?)
So the argument is that as the scale grows, the costs of forging rise to the point where few can afford to be altruistic about it. When that happens, the number of forgers will be proportional to the income from forging. If Nxt has more than 101 forgers, it will be because the Nxt transaction fees are higher than Bitshares.

One question I have about this: won't the costs be bursty? I'm assuming transparent forging: that is, that most transactions are sent to the few nodes most likely to forge the next block. In this model, there will be a base cost from running a full node: you need to maintain a full copy of the block-chain and relay transactions. If you are also a forging node, there will be the additional cost of occasionally forging a block. Every so often, you'll receive a deluge of 7,000 transactions per second for a minute or so, and your node needs to be able to handle that load.

According to Bitshares, if we double the number of forging nodes, we need to double the transaction fees. I agree that twice as many people will be paying the base cost. However, the number of people paying the burst cost will stay the same, because the number of blocks per day won't change. A given node will have to handle the deluge of transactions half as often. Won't that be cheaper for the individual nodes? The internet is used to bursty loads and you can often arrange to only pay for the capacity you are using, while you are using it.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: jl777 on August 24, 2014, 07:47:09 pm
IDK if anyone posted this excerpt yet but I feel like it is critical:

wiki (dot) bitshares (dot) org/index.php/DPOS#Scalability

tl;dr  ALL consensus protocols centralize at scale. How well do you manage it? Does yours squeeze out small players (or do they have to do it at a loss like with BTC or soon PPC... and eventually NXT?)
So the argument is that as the scale grows, the costs of forging rise to the point where few can afford to be altruistic about it. When that happens, the number of forgers will be proportional to the income from forging. If Nxt has more than 101 forgers, it will be because the Nxt transaction fees are higher than Bitshares.

One question I have about this: won't the costs be bursty? I'm assuming transparent forging: that is, that most transactions are sent to the few nodes most likely to forge the next block. In this model, there will be a base cost from running a full node: you need to maintain a full copy of the block-chain and relay transactions. If you are also a forging node, there will be the additional cost of occasionally forging a block. Every so often, you'll receive a deluge of 7,000 transactions per second for a minute or so, and your node needs to be able to handle that load.

According to Bitshares, if we double the number of forging nodes, we need to double the transaction fees. I agree that twice as many people will be paying the base cost. However, the number of people paying the burst cost will stay the same, because the number of blocks per day won't change. A given node will have to handle the deluge of transactions half as often. Won't that be cheaper for the individual nodes? The internet is used to bursty loads and you can often arrange to only pay for the capacity you are using, while you are using it.
dont forget NXT value will grow as usage grows, so if double the traffic, very likely NXT is worth close to double
Title: Re: NXT POS vs. Bitshares DPOS
Post by: ChuckOne on August 24, 2014, 09:14:36 pm
Yes, I am a bitshill.

Fine, suppose they don't need to make a profit, then neither does any other system. Point is, somebody has to incur the costs, and eventually the number of people that are willing to incur the costs will decrease. All such systems centralize.

I would not say so. It depends on how big those costs are.
Title: Re: NXT POS vs. Bitshares DPOS
Post by: Brangdon on August 25, 2014, 10:41:54 am
dont forget NXT value will grow as usage grows, so if double the traffic, very likely NXT is worth close to double
Yes; that's independent of the number of forging nodes, though. Bitshares are saying that if the number of forging nodes doubles, the income for each node will halve, because the same number of NXT will be divided between them. That's unrelated to the traffic level, or the value of 1 NXT.

It's also independent of the currency; Bitshares are arguing that if we have more forging nodes for a given number of transactions, it will be because our fees are higher. Even if NXT is worth more, presumably the same applies to Bitshares so it cancels out.

7,000 tps for 60 seconds is about 420,000 transactions per block. Assuming no altruism, the real-world value of an average NXT transaction fee will be 1/420,000 times the real-world cost of processing a block, whatever that comes out to in NXT. Maybe it will be low enough that lowering it further won't matter; I don't know.
elective-stereophonic
elective-stereophonic
assembly
assembly