elective-stereophonic
elective-stereophonic
NXT POS and Bitshares DPOS  
Please login or register.

Login with username, password and session length
Advanced search  

News:

Latest Stable Nxt Client: Nxt 1.12.2

Pages: [1] 2 3 ... 7  All

Author Topic: NXT POS and Bitshares DPOS  (Read 26530 times)

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
NXT POS and Bitshares DPOS
« 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.
« Last Edit: August 24, 2014, 10:12:22 am by delulo »
Logged

Damelon

  • Hero Member
  • *****
  • Karma: +792/-54
  • Offline Offline
  • Posts: 2314
    • View Profile
    • Nxt Inside
Re: NXT POS vs. Bitshares DPOS
« Reply #1 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!
Logged
Member of the Nxt Foundation | Donations: NXT-D6K7-MLY6-98FM-FLL5T
Join Nxt Slack! https://nxtchat.herokuapp.com/
Founder of Blockchain Workspace | Personal Site & Blog

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #2 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!
Logged

Damelon

  • Hero Member
  • *****
  • Karma: +792/-54
  • Offline Offline
  • Posts: 2314
    • View Profile
    • Nxt Inside
Re: NXT POS vs. Bitshares DPOS
« Reply #3 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
Logged
Member of the Nxt Foundation | Donations: NXT-D6K7-MLY6-98FM-FLL5T
Join Nxt Slack! https://nxtchat.herokuapp.com/
Founder of Blockchain Workspace | Personal Site & Blog

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #4 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...]
Logged

mczarnek

  • Hero Member
  • *****
  • Karma: +68/-4
  • Offline Offline
  • Posts: 898
    • View Profile
    • Nxt Place - Craigslist for Nxt
Re: NXT POS vs. Bitshares DPOS
« Reply #5 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?
« Last Edit: July 25, 2014, 05:08:57 pm by mczarnek »
Logged
NXT Organization: Tech
Donations greatly appreciated: NXT-DWVJ-G89C-RHNL-6QW6Q

Agent86

  • Newbie
  • *
  • Karma: +0/-0
  • Offline Offline
  • Posts: 6
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #6 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.
Logged

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #7 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?
« Last Edit: July 24, 2014, 11:42:44 pm by delulo »
Logged

Damelon

  • Hero Member
  • *****
  • Karma: +792/-54
  • Offline Offline
  • Posts: 2314
    • View Profile
    • Nxt Inside
Re: NXT POS vs. Bitshares DPOS
« Reply #8 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.
Logged
Member of the Nxt Foundation | Donations: NXT-D6K7-MLY6-98FM-FLL5T
Join Nxt Slack! https://nxtchat.herokuapp.com/
Founder of Blockchain Workspace | Personal Site & Blog

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #9 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?
Logged

Damelon

  • Hero Member
  • *****
  • Karma: +792/-54
  • Offline Offline
  • Posts: 2314
    • View Profile
    • Nxt Inside
Re: NXT POS vs. Bitshares DPOS
« Reply #10 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 :)
Logged
Member of the Nxt Foundation | Donations: NXT-D6K7-MLY6-98FM-FLL5T
Join Nxt Slack! https://nxtchat.herokuapp.com/
Founder of Blockchain Workspace | Personal Site & Blog

Agent86

  • Newbie
  • *
  • Karma: +0/-0
  • Offline Offline
  • Posts: 6
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #11 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!
Logged

Damelon

  • Hero Member
  • *****
  • Karma: +792/-54
  • Offline Offline
  • Posts: 2314
    • View Profile
    • Nxt Inside
Re: NXT POS vs. Bitshares DPOS
« Reply #12 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
Logged
Member of the Nxt Foundation | Donations: NXT-D6K7-MLY6-98FM-FLL5T
Join Nxt Slack! https://nxtchat.herokuapp.com/
Founder of Blockchain Workspace | Personal Site & Blog

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #13 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.
Logged

Agent86

  • Newbie
  • *
  • Karma: +0/-0
  • Offline Offline
  • Posts: 6
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #14 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.
Logged

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #15 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?
Logged

Agent86

  • Newbie
  • *
  • Karma: +0/-0
  • Offline Offline
  • Posts: 6
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #16 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)
« Last Edit: July 25, 2014, 01:07:16 am by Agent86 »
Logged

ThomasVeil

  • Hero Member
  • *****
  • Karma: +183/-11
  • Offline Offline
  • Posts: 1400
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #17 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?
Logged
ARDOR-BPV3-837M-QZTQ-9DQ69  oxpal.com

Agent86

  • Newbie
  • *
  • Karma: +0/-0
  • Offline Offline
  • Posts: 6
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #18 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.
Logged

delulo

  • Jr. Member
  • **
  • Karma: +4/-0
  • Offline Offline
  • Posts: 43
    • View Profile
Re: NXT POS vs. Bitshares DPOS
« Reply #19 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.
Logged
Pages: [1] 2 3 ... 7  All
 

elective-stereophonic
elective-stereophonic
assembly
assembly