Author Topic: How Well Does DPoS Really Work?  (Read 4381 times)

0 Members and 1 Guest are viewing this topic.

Offline Thom

Quote
Three of the 100% pay rate delegates ranked in the top half run software at least 2 revisions out of date
If that was the case, they wouldn't sign blocks at all. I guess they just forgot to publish their new version.
I for myself updated the version of one of my delegates just yesterday ..

This observation is an indication of voter apathy, as is the fact that some of the devs don't publish feeds. IMO nobody has much of a basis for claiming devs don't pull their own weight in supporting this ecosystem. What kind of a message does it send if the devs that created the system in the first place not use it as they intend, my failing to keep their published delegate version up to date or publish feeds? This is a matter of integrity as well.

And shame on you too xeroc!  ;) We all KNOW you're a major contributor to BitShares.

This thread has surfaced some really great comments & discussion, which was my primary goal. As usual BM chimes in with short, concise inputs to realign and coalesce snippets of thought from the community into a positive cohesive energy.

Improvements in the DPoS strategy are just around the corner. Troglodactyl was quick to offer a reason why some of the apathy may be explained. As I posited in the OP, why so much more discussion about BitAssets ideas and so little about delegate reform? Clearly there is plenty of material, this thread has surfaced several areas of thought that could use attention:

  • Accuracy of info shareholders rely on to gauge witness / worker / delegate performance
  • What are minimum standards for being a witness / worker / delegate? (this has been discussed in the past too)
  • Should better mechanisms be built besides delegate slates to help people express their voting preferences? +5% to Tuch
  • Should price feed publishers be a separate role from block producers?

Bytemaster, you're oh so right about incentives for price feeds, and where they should originate. We should hold you accountable to develop that idea further, into the pay structure of those witnesses that decide to publish price feeds. It may not come tomorrow, but you just gave us a very good foundation of how price feeds should be paid for. Not only is that function quite distinct from block production, there is no need for it to require the same level of hardware support to publish feeds as a block producer does. There are less technical requirements but more mathematical and economic requirements. Having a separate role for feed publishers also provide more direct accountability to the market and shareholders. A separate role for elected feed publishers has real merit.

« Last Edit: June 07, 2015, 03:54:37 pm by Thom »
Injustice anywhere is a threat to justice everywhere - MLK |  Verbaltech2 Witness Reports: https://bitsharestalk.org/index.php/topic,23902.0.html

Offline Akado

  • Hero Member
  • *****
  • Posts: 2752
    • View Profile
  • BitShares: akado
Can an automated way to vote be created (or has it been?), where the user can select certain criteria to vote yes and no? If running version less than x, if no feed, if payrate=x, etc.
Delegates can publish slates to achieve this.
Perhaps one in particular could specialize in providing a slate that meets the criteria that you suggested.

It's not the same thing though, you have to rely on a someone else and although we can trust some delegates, it doesn't mean any one of the 101 will have a slate which I agree 100% with. A custom voting option would be awesome
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline BunkerChainLabs-DataSecurityNode

Can an automated way to vote be created (or has it been?), where the user can select certain criteria to vote yes and no? If running version less than x, if no feed, if payrate=x, etc.

That would be awesome.  +5% +5%

I still don't get why people don't publish feeds, they have no financial incentive but they are still contributing to the network and it's needed. I thought it would be one of the basic "to-dos" on the delegates list. I honestly see no excuse for a delegate not to publish feeds. The fact that a new system is coming out is no excuse as we are still not there yet and the fact that his situation has been going on for a long time. But I guess that's just my opinion on the requisites to become a delegate.. other people probably have it different, I respect that. I just don't understand why, atm. For me it's like they're skipping part of their job, which they are already being paid for, just because they apparently don't want to. I haven't seen a good excuse atm for a delegate not to publish a feed, that's what annoys me.

zero excuse for any 100% delegate.. for 3% though the transaction costs of publishing feeds out of what little they get to cover their server expense becomes significant. Depending on how often you publish, it can be as much as 300-600BTS a month.

I see many 3%es also having a 100% delegate who don't publish feeds.. So it's really just a small handful of players that don't care to do it.. surprisingly enough a good chunk are dev delegates.
+-+-+-+-+-+-+-+-+-+-+
www.Peerplays.com | Decentralized Gaming Built with Graphene - Now with BookiePro and Sweeps!
+-+-+-+-+-+-+-+-+-+-+

Offline Permie

  • Hero Member
  • *****
  • Posts: 606
  • BitShares is the mycelium of the financial-earth
    • View Profile
  • BitShares: krimduss
Can an automated way to vote be created (or has it been?), where the user can select certain criteria to vote yes and no? If running version less than x, if no feed, if payrate=x, etc.
Delegates can publish slates to achieve this.
Perhaps one in particular could specialize in providing a slate that meets the criteria that you suggested.
JonnyBitcoin votes for liquidity and simplicity. Make him your proxy?
BTSDEX.COM

Offline Akado

  • Hero Member
  • *****
  • Posts: 2752
    • View Profile
  • BitShares: akado
Can an automated way to vote be created (or has it been?), where the user can select certain criteria to vote yes and no? If running version less than x, if no feed, if payrate=x, etc.

That would be awesome.  +5% +5%

I still don't get why people don't publish feeds, they have no financial incentive but they are still contributing to the network and it's needed. I thought it would be one of the basic "to-dos" on the delegates list. I honestly see no excuse for a delegate not to publish feeds. The fact that a new system is coming out is no excuse as we are still not there yet and the fact that his situation has been going on for a long time. But I guess that's just my opinion on the requisites to become a delegate.. other people probably have it different, I respect that. I just don't understand why, atm. For me it's like they're skipping part of their job, which they are already being paid for, just because they apparently don't want to. I haven't seen a good excuse atm for a delegate not to publish a feed, that's what annoys me.
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Tuck Fheman

  • Guest
Can an automated way to vote be created (or has it been?), where the user can select certain criteria to vote yes and no? If running version less than x, if no feed, if payrate=x, etc.

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc


The direction the innovation is headed makes me feel like the competition might as well be standing still  +5%
haha .. so true ..
hint: bitcoin block size :)


Offline Ben Mason

  • Hero Member
  • *****
  • Posts: 1070
  • Integrity & Innovation, powered by Bitshares
    • View Profile
  • BitShares: benjojo
Great thoughts Thom. We have to keep analysing and pushing for the best Bitshares can be.

BM your succinct clarity is inspiring as always.  What has been accomplished with DPOS already simply blows my mind. The direction the innovation is headed makes me feel like the competition might as well be standing still  +5%

Offline bytemaster

How well does DPOS really work compared to.... ????

It seems like it is being compared to some theoretical ideal DPOS...

It seems like our long-term average delegate participation is above 95% which means the DPOS part of the network is working perfectly. 

DPOS still functions even with 55% average delegate participation.   

So why does DPOS function... because it pays for blocks to be produced and delegates don't get paid if blocks are not produced.   Very simple cause/effect reward system.   

Why don't feeds get published... simple: no direct financial incentive to do so and punishment is "slow".    But this is an issue with BitAssets not delegates.   BitAssets need to pay feed producers.   Thus why we have privatized bitassets.

If you boil DPOS down to only focus on block production then I think it is working VERY well.    Everything else is just a matter of scale. 

For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline Samupaha

  • Sr. Member
  • ****
  • Posts: 479
    • View Profile
  • BitShares: samupaha
This leads me to believe the majority of our community is focused on trading and profit rather than the underlying facilities that allow that to occur. It makes me wonder if the voter apathy and overall lack of concern about the fundamentals of DPoS indicate an inherent weakness in its viability as a mechanism to sustain an effective and efficient crypto ecosystem.

Should we focus on the quality of the shareholders?

Cryptoworld is full of people who just want to get rich fast and do not care how the projects will work in the long run. I think we do not need these kind of people as shareholders, because they will not vote, or if they do, they will mostly vote for short term profits.

We need more shareholders who are actively interested in developing BitShares to be the best blockchain in the long run.

Offline puppies

  • Hero Member
  • *****
  • Posts: 1659
    • View Profile
  • BitShares: puppies
When BM announced DPOS, I was not a fan.  In the interim I believe that it has secured our blockchain very well.  I am not sold on the utility of all of the 100% delegates.  You bring up some very good points in regards to improving delegate performance.  In regards to how well DPOS works, I would ask you.  Compared to what?  Even with its flaws I think it is better than bitcoin and all other POW schemes.  I think it's better than NXT.  Has something new come along that I haven't seen?  This of course does not mean that we should not work on getting better.  I am eagerly awaiting details on the new structure.  I think it is a very good thing to split protecting the network away from growing the network.

To touch upon Xeroc's point.  I have definitely not made a cent being a delegate.  I do it because its fun, kind of a hobby.  I also believe in the mission, and believe that we are making history.  Sadly I didn't make your top ten, I've only got 14.32% of the vote, I have moved up a single slot in the last week, and I only publish feeds 16 times per day. 
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline BunkerChainLabs-DataSecurityNode

25% of delegates don't even publish price feeds
29% of the 100% pay rate delegates don't publish feeds
That really sucks!

Quote
Three of the 100% pay rate delegates ranked in the top half run software at least 2 revisions out of date
If that was the case, they wouldn't sign blocks at all. I guess they just forgot to publish their new version.
I for myself updated the version of one of my delegates just yesterday ..

Quote
Ten of the 100% delegates have questionable reliability (less than 98% reliability rating)
That measure might not be fair at all considering we had some outages that were
not the delegates' fault (e.g. last december). Hence, delegates that are longer
part of the top 101 necessarily have a lower reliability.

I think a reasonable "reliability" measure should consider a moving average
over the last x weeks. (reasonable value for x has to be determined)

Quote
80% of the top 10 delegates (see below) earn 100% pay rate
Isn't that obvious? People working hard for the ecosystem get the most votes?

Quote
I have to ask, do BitShares delegates take their role seriously? Do the
shareholders? On the positive side, we have seen action taken rather quickly to
remove poorly performing delegates, once they are brought to the attention of
the community.
Let's do the math: 3% of 50 BTS per block signed is about 130 BTS a day and
gives rougly $40 a month.
That hardly pays for
a) the machine
b) the time you need to for updates
c) the stress involved with running a delegate 24/7
d) the time you need to keep yourself up to date
e) the time and resources you need to maintain your feeds, payouts, reliability
... not considering the fees you need to pay for feed publishing.

So, now I have to ask you: Are you not satisfied about how awesome this
community is running their delegates despite the fact that it is by far not
profitable for them?

However, the math above holds true for 3% delegates but I am too no so
impressed by the performance of most 100% delegates.

Quote
I am rather concerned however of the very limited discussion surrounding the
heart & soul of the BitShares ecosystem, DPoS. Although full disclosure of DPoS
2.0 has yet been revealed, neither have the details about BitAssets and yet
that topic is being discussed at length. Why?

This leads me to believe the majority of our community is focused on trading
and profit rather than the underlying facilities that allow that to occur.
Trading is important .. very important .. it provides liquitdity!

Quote
It makes me wonder if the voter apathy and overall lack of concern about the
fundamentals of DPoS indicate an inherent weakness in its viability as a
mechanism to sustain an effective and efficient crypto ecosystem.
... remains to be seen. BitShares is still an experiment ... though I would say
it is on it's way to succeed.

Quote
I am however grateful to BM & the dev staff for their dedication to thinking
about the underlying infrastructure and the innovation that will hopefully
balance out the tradeoffs between a reliable and high performance blockchain
network and the elements of it that will attract people to use it. So I remain
cautiously optimistic.
And that's your right to do.
Thanks anyway for your analysis.

 +5% +5%

Wow good answers.

I agree that there is some drift wood among the delegates... but as xeroc pointed out.. this is despite the lack of funds provided to do all that. It's all in how you choose to view it. I am operating these 100% delegates that I don't even see a penny from because it's all going into what they run. We all know what's coming down with the TBD so I don't know why we want to wage a finger at DPOS 1.0. I agree with Xeroc that overall it has been extremely successful under the harsh conditions.

Aaaand if we are all about being judgey judgey here about delegate performance.. it should be noted that minebts1.bunkermining-com has missed the least amount of blocks along with solly. :)

Now if you just vote in the other 5.. we can put some of that drift wood you mentioned aside and have some prime nuke proof delegates make their way in. :) http://vote.bunkermining.com !
+-+-+-+-+-+-+-+-+-+-+
www.Peerplays.com | Decentralized Gaming Built with Graphene - Now with BookiePro and Sweeps!
+-+-+-+-+-+-+-+-+-+-+

Offline testz

...
Feed update frequency at least 27 (once per hour > 24?)
...

IMHO it's can't delegate performance counter, almost all delegates update price feeds by percentage of price swing if price stable it's can be 1 per 24 hours if not it's can be 100+ updates per 24 hours. So you think that delegates which spam network with price feeds perform better?
All rest looks good, thanks for analysis.

Offline karnal

  • Hero Member
  • *****
  • Posts: 1068
    • View Profile
Guess I'll have to setup a couple of delegates one of these days..

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
25% of delegates don't even publish price feeds
29% of the 100% pay rate delegates don't publish feeds
That really sucks!

Quote
Three of the 100% pay rate delegates ranked in the top half run software at least 2 revisions out of date
If that was the case, they wouldn't sign blocks at all. I guess they just forgot to publish their new version.
I for myself updated the version of one of my delegates just yesterday ..

Quote
Ten of the 100% delegates have questionable reliability (less than 98% reliability rating)
That measure might not be fair at all considering we had some outages that were
not the delegates' fault (e.g. last december). Hence, delegates that are longer
part of the top 101 necessarily have a lower reliability.

I think a reasonable "reliability" measure should consider a moving average
over the last x weeks. (reasonable value for x has to be determined)

Quote
80% of the top 10 delegates (see below) earn 100% pay rate
Isn't that obvious? People working hard for the ecosystem get the most votes?

Quote
I have to ask, do BitShares delegates take their role seriously? Do the
shareholders? On the positive side, we have seen action taken rather quickly to
remove poorly performing delegates, once they are brought to the attention of
the community.
Let's do the math: 3% of 50 BTS per block signed is about 130 BTS a day and
gives rougly $40 a month.
That hardly pays for
a) the machine
b) the time you need to for updates
c) the stress involved with running a delegate 24/7
d) the time you need to keep yourself up to date
e) the time and resources you need to maintain your feeds, payouts, reliability
... not considering the fees you need to pay for feed publishing.

So, now I have to ask you: Are you not satisfied about how awesome this
community is running their delegates despite the fact that it is by far not
profitable for them?

However, the math above holds true for 3% delegates but I am too no so
impressed by the performance of most 100% delegates.

Quote
I am rather concerned however of the very limited discussion surrounding the
heart & soul of the BitShares ecosystem, DPoS. Although full disclosure of DPoS
2.0 has yet been revealed, neither have the details about BitAssets and yet
that topic is being discussed at length. Why?

This leads me to believe the majority of our community is focused on trading
and profit rather than the underlying facilities that allow that to occur.
Trading is important .. very important .. it provides liquitdity!

Quote
It makes me wonder if the voter apathy and overall lack of concern about the
fundamentals of DPoS indicate an inherent weakness in its viability as a
mechanism to sustain an effective and efficient crypto ecosystem.
... remains to be seen. BitShares is still an experiment ... though I would say
it is on it's way to succeed.

Quote
I am however grateful to BM & the dev staff for their dedication to thinking
about the underlying infrastructure and the innovation that will hopefully
balance out the tradeoffs between a reliable and high performance blockchain
network and the elements of it that will attract people to use it. So I remain
cautiously optimistic.
And that's your right to do.
Thanks anyway for your analysis.