CPOS (cooperative proof of stake) discussion thread

[bytemaster]

To be clear we will not be changing code to accommodate these ideas in the first release.


Sent from my iPhone using Tapatalk

[luckybit]

I'm very conservative about the initial Bitshares X chain. I think for future chains we can test some of these ideas out but I don't see why it's necessary to do this right now when Bitshares X isn't even released yet to fulfill the social consensus to the current set of owners.

I see nothing wrong with having different chains with different BitAssets. Some of these different chains can try slightly different business models.

But the way I see it working is if there is a inflationary chain and a deflationary chain and we have shares in both then it's a lot easier to talk about experimenting. If we just have one chain and it's not even launched yet then it sounds like there could be another major change at the last minute and this kind of change is as major as switching to Proof of Stake.

Let's release Bitshares X first. After it is released then Bitshares X 2.0 can be based on a snapshot of Bitshares X 1.0 but with these newer features.

For marketing there are a lot of ways to do that so I don't think you really need to dilute shares for that. I do understand that somehow funding can run out, especially when you look at the funding of Mastercoin and potentially Ethereum. Why not just use Angelshares to fund marketing campaigns in the future? Why mess with the DAC?

I'm not against attempting it because Bitshares itself is an experiment in next generation capitalism. But I don't think it's wise to take the corporate model of capitalism complete with all it's corruption and flaws and import it all into the decentralized capitalism 2.0. If we are going to find a way to do it we have to up our standards to a level of efficiency beyond that of corporate crony capitalism.

Our model must be designed to be fault tolerant, error tolerant, collusion resistant, the incentives should be moved around until the perfect incentive model is found to optimize for maximum value increase of the shares. Bytemaster is right in that if you can increase the value of the shares enough to the point where that value increase goes beyond the dilution then you can get away with dilution.

But there is still a problem. When you dilute or inflate the shares you create a sort of black hole where corruption and collusion can form. Cartels and other groups might form around the creation process to corrupt it in such a way that once you start diluting the shares the social and political forces prevent you from ever cutting the flow of shares off.

These shares would have to be tracked, it would cause all sorts of potential problems. If we look at it as corporate bonuses then you have to actually measure the real world effect. Unfortunately the real world bankers give bonuses even when the effects as bad as a reward for what? So is it possible that this new bonus structure could be abused in the same sort of way if the delegates or the social engineering become too strong?

[santaclause102]

Also stop to consider the issue mathematically. For example, in a group of 100 people, you contribute 1% to the total if everyone works equally hard, which of course they won’t. If you do less than the others, you contribute even less than 1%, so your efforts are statistically meaningless; if you do more, your efforts are subsidizing the slackers – but you’ll still have to share the reward with them.

The quote above describes the common pool resource problem / tragedy of the commons. It is a similar problem among shareholders but there are some differences. If you differentiate shareholders functionally completely from the active operational functions they might have as well there is nothing one shareholder does more than another. Every shareholder gives his capital to the company exactly to the degree that he will profit from it later (through share price appreciation or dividends). There are two problems here: (1) The "group trap" here is between all shareholders and those the shareholders entrust the operational work with (delegates, dac dev team) - this is described as the "principle agent problem". (2) The problematic (but unavoidable (?) and before hand known) fact/problem that there is a majority vote over operational and delegate decisions (share dilution is an operational decision, see next paragraph).

I basically agree with troc that it can be a moral problem to dilute shareholders by majority vote. But this is not the same dilution like with fiat money. It has a different purpose. The goal of company share dilution is capital infusion and therefore growth for all. There might be a moral problem when the minority thinks that the share dilution will not have the desired effect (effect = value appreciation per share: Because the effect of the investment in marketing etc. is stronger than the dilution effect).

establishing an objective measure of the value of the capital infusion and who is responsible for that infusion

The only solution is the critical and foreseeing (and therefore never perfect) judgement of shareholders about the delegates policies and / or about the effectiveness of a share dilution.

The overall question that dominates is: What is the original motivation for capital owners to join capital efforts to form a company at all and therefore buy in to some unavoidable group traps / principle agent problems? There are reasons obviously and some degree (the least possible) of accepting group traps / principle agent problems might be necessary and justified because those problems are known before hand and are inherent to every company.
We could also say that by joining other shareholders you enter a group and by doing this you accept the majority vote of the group. Isn't it like this de facto with any company you invest in? 


I will define the words I used and are used here often with respect to our specific case (DACs). We might be at cross purposes here if we understand the words differently.
Share / shareholder dilution and capital infusion: Offering new shares to the market that can be bought by everyone. The revenue from this public share sale is the capital infusion.
Value of the capital infusion, as used by bm: I would rather say "efficiency of the share dilution". Value of the capital infusion could be understood as the monetary value of what you get from the sale of shares = monetary value of the capital infusion.

[bytemaster]

In the short term, it may seem that redistributing property in order to stimulate growth is catalytic and beneficial, but in the long run I think efficiency will win out, and forced redistribution is inherently inefficient and destabilizing.

You say that you know the rules going in, but the rules to which you're referring are the rules for changing the rules.  People should read the fine print and see this, but if they do, do you think they would still invest?  I would hate to see the "decentralized solutions to centralized problems" goal die here and the project become just another tool some people use to dominate others.

There are lots of philosophical and moral issues here, and to deny them I think would be shortsighted.  They're part of the product, and sacrificing them for apparent expediency could alienate potential customers and investors.

I am going to start off with the fundamental philosophical perspective: All shared ownership systems fall into the Group Trap...

Next is the Group Trap, which is the belief that you can accomplish more by acting in groups than you can by acting on your own. Harry didn’t believe that there’s anything inherently wrong with participating in groups; you may enjoy the social aspect or something else about it. But you should be consciously aware that, if you just want to accomplish something, you not only don’t have to go through a group, but it’s actually easier to act on your own.

.....


Also stop to consider the issue mathematically. For example, in a group of 100 people, you contribute 1% to the total if everyone works equally hard, which of course they won’t. If you do less than the others, you contribute even less than 1%, so your efforts are statistically meaningless; if you do more, your efforts are subsidizing the slackers – but you’ll still have to share the reward with them.

So attempting to accomplish something together via 'shares' is already very close to falling into this trap.   Those who work hard to market the DAC subsidize those who do not.   

The best you can do to mitigate the damage from the group trap is to avoid falling into the government trap.  In this case some people are put in authority over assets belonging to others.   This kind of situation introduces the moral hazard we hope to avoid with DACs.   

All of that said, the concept of diluting shareholders makes sense only when then the value-per-share increases (or stays the same) as a result of a capital infusion.  In this sense you can view the system like salt water where the salt is the value and the water is the shares.   If you add more salt than water then you don't actually dilute the shareholders. 

From this perspective it would be immoral to accept a capital infusion without issuing new shares because the person giving the infusion would be getting nothing (no shares) for something (capital infusion). 

And this is where the group trap and economic reality come to the foreground, there is no objective measure of the value of the capital infusion and thus each individual shareholder perceives the 'dilution' differently.   Some see it as a net gain, others a net loss and they are both right.

Now we have identified the crux of the matter:  establishing an objective measure of the value of the capital infusion and who is responsible for that infusion.   If it were possible to compensate the delegates proportional to the real rise in the value of the shares, then it would work like corporate bonuses for stock gains.  This works if the value increase can be directly tied to the actions of a set of delegates.   The problem is that the delegates themselves are in a group trap.  A lazy delegate can do nothing and reap huge rewards from the delegates that are actually doing something useful.   

With BitUSD DACs have an objective internal measure of their own value.  This means we have the potential to tie dilution rates to the rise in value of the DAC.    If you combine this dilution with a vesting period (say 3 years) then the delegates will have to work for long-term growth rather than short-term. 

As you can see this is a complex issue with many pitfalls.   

[Troglodactyl]

That's right, the majority doesn't have to open their wallets.  They can take no action.  They don't have any more right to force the minority to open their wallets than the minority has to force the majority to open their wallets.  The default state is that each is free to use his own resources as he sees fit, but none are free to use the resources of others without their consent.  If I choose to do something that I claim will also benefit you, that does not entitle me to use your money to pay for it unless you agree to contribute.

Snapshotting to split a network by strategy is an interesting idea, but it sacrifices network effect for purity.  It might be necessary if the main network is headed for collapse and can't be turned, but it's a sacrifice technique.

Talking about "they don't have a right" as if it's a moral question is missing the point.  By not allowing dilution on some kind of philosophical ground you are just limiting the options and tying the hands of the company.  I believe that a company that refuses under all circumstances to allow dilution would be out competed in the market by a company that can do this.

Virtually no large successful companies could ever have made it without being able to raise capital and/or bring in new stakeholders by issuing new shares.

These DACs can be quickly hiring developers/marketers/lawyers/executives making deals companies like Coinbase...  While you sit around waiting for enough transaction fees to come in so you can finally do something, and/or appealing to the charitable nature of some of the stakeholders while others profit more from inaction.

It's not a moral issue, you know the rules going in, if you don't like the way the rules are written you don't have to buy in to the DAC or you can sell your shares.  If you don't like to go along with what a majority want than don't buy shares of a company, because that's what you have to do.

Tying the hands of "the company"?  Who exactly do you mean by this?  The only hands we're tying are the ones otherwise headed for their neighbor's pockets.  Designing the company such that the majority stakeholders can confiscate the stake of the minority undermines trust in the company.  The more power you give the delegates to take from the shareholders, the more corruption you invite.

Dilution is just redistribution.  The only difference is opacity.  If delegates can collect seignorage in addition to transaction fees, the likelihood of long term social engineering attack vectors increases significantly.  There are certainly differences because it's a company with voluntary participation, but some lessons from central banking are still applicable.

In the short term, it may seem that redistributing property in order to stimulate growth is catalytic and beneficial, but in the long run I think efficiency will win out, and forced redistribution is inherently inefficient and destabilizing.

You say that you know the rules going in, but the rules to which you're referring are the rules for changing the rules.  People should read the fine print and see this, but if they do, do you think they would still invest?  I would hate to see the "decentralized solutions to centralized problems" goal die here and the project become just another tool some people use to dominate others.

There are lots of philosophical and moral issues here, and to deny them I think would be shortsighted.  They're part of the product, and sacrificing them for apparent expediency could alienate potential customers and investors.

[Agent86]

That's right, the majority doesn't have to open their wallets.  They can take no action.  They don't have any more right to force the minority to open their wallets than the minority has to force the majority to open their wallets.  The default state is that each is free to use his own resources as he sees fit, but none are free to use the resources of others without their consent.  If I choose to do something that I claim will also benefit you, that does not entitle me to use your money to pay for it unless you agree to contribute.

Snapshotting to split a network by strategy is an interesting idea, but it sacrifices network effect for purity.  It might be necessary if the main network is headed for collapse and can't be turned, but it's a sacrifice technique.

Talking about "they don't have a right" as if it's a moral question is missing the point.  By not allowing dilution on some kind of philosophical ground you are just limiting the options and tying the hands of the company.  I believe that a company that refuses under all circumstances to allow dilution would be out competed in the market by a company that can do this.

Virtually no large successful companies could ever have made it without being able to raise capital and/or bring in new stakeholders by issuing new shares.

These DACs can be quickly hiring developers/marketers/lawyers/executives making deals companies like Coinbase...  While you sit around waiting for enough transaction fees to come in so you can finally do something, and/or appealing to the charitable nature of some of the stakeholders while others profit more from inaction.

It's not a moral issue, you know the rules going in, if you don't like the way the rules are written you don't have to buy in to the DAC or you can sell your shares.  If you don't like to go along with what a majority want than don't buy shares of a company, because that's what you have to do.

[luckybit]

I don't think it would be a good idea to hard code a share dilution for a fixes amount of time. Would there be a possibility to do a temporary share dilution without hard forking? Would be helpful, especially at the beginning, in case tx fees are not enough and the gains from adoption would be the highest.   

Why not just let all delegates keep as much of the fee from their blocks as they want, and destroy the rest?  If they abuse that, they'll be voted out quickly, and the market will naturally regulate the dividend rate and strategies for reinvestment.

This is a good idea and would be necessary if delegates also make reinvestment decisions. My concern was that an entity with an interest to harm the network that has a lot of financial backup (like a state) can survive this low compensation better. But there is probably no way to prevent it and a harmful actor could obfuscate its identity anyway. Could it?

I thought about whether it is a problem that the max. amount of votes a delegate can get is 2% of all votes (pro - contra votes). But is should be no problem if it is not possible to vote for a delegate that would have more than 2% if you voted for him. Then the personal or team behind the delegate has to set up a new delegate...

If transaction fees are insufficient I'd suggest convincing shareholders to reinvest directly and voluntarily.  Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

The concern at surviving low compensation isn't solved by hard coding compensation, since any delegate can reduce his own compensation by destroying some of it manually, and can advertise these dividends to the shareholders.  If a state wants to run a delegate at low compensation, they're welcome to, but if they try to use it to exert power negatively, they should be voted out quickly just like anyone else would be.

How about offering something equivalent to a tax credit for those who go along with it voluntarily? It doesn't even have to come from the same DAC as long as another DAC can send value through cross chain transactions. I'm not exactly sure how atomic cross chain transactions will work but I don't see any reason why different blockchains couldn't give credits of some sort.

The problem is it would be overly complex. Maybe it also might not be something which is immediately necessary.

Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

I disagree with this.  Just because some small group decides they want to be "freeloaders" doesn't mean that the majority has to open their wallets and pay for them while they get all the growth benefits at no charge.  This would be like making taxes optional, no one will pay them.

It's not like majority can vote that the minority has to give them all the money or something, the minority still has representation.

My previous idea is everyone votes for an inflation rate and take the median, or some other way of giving the DAC flexibility toward growth or dividends.  0% inflation would be all fees are given to delegates with no dividend. 2% inflation would be creating new shares to fund growth.  -1% inflation would be destroying some percent of transaction fees for dividends (might be more appropriate for a long established DAC).

The way we take a snapshot now to honor a DAC, in the future, people might honor all shares of a DAC that vote for a particular group of delegates.  That way people who want to take the company in a different direction can fork it and bring the people who share that vision with them and then compete in the market.

Inflation is dangerous. First explain why we need inflation at all and what problem does it solve which wouldn't be solved without inflation?

1) Diluting shareholders as Bitcoin does now and spending this money / those coins on marketing ("paying Amazon to accept Bitcoin") instead of paying it to miners or shareholders - would this be an advantage to DPOS/bitshares?

I think it's debatable. No I don't think you should dilute shareholders because why get rid of mining saying its to prevent dilution of shares only to put the dilution back without any of the benefits of mining?

I don't think new shares need to be created to fund growth. Reward people who already have shares and who donate them to projects which fund growth.

Value isn't in the shares it's in the products and services. Any problem can be solved by crowd funding. Use Angelshares for this purpose because isn't that the whole point of Angelshares?

[luckybit]

Is it not obvious that inflation is a brilliant way to force everyone to continually donate a tiny amount? Does it not effectively solve the tragedy of the commons problem? Are we sure our idealist attitudes are not interfering with our judgement? What does it matter that voting etc. on how to spend this influx of money is flawed, when we know that it can solve at least one problem that individuals as individuals can not, namely the situation where each donating a tiny amount would be rational if they all did it, but they do not do it since they can not be sure everyone or even anyone else does. Pending a perfect solution, do we not already know that we require both types of spending for the whole system to be supra-rational?

I think of it more like shareholders agreeing to a dilution in exchange for a capital infusion.    I think it is indeed a good idea in the short run and works well for DPOS.   Early on in a networks life transaction fees are unlikely to be sufficient to motivate growth.  They will merely cover costs.   

So I have thought about it some and recognized that no company pays a dividend while in its growth phase.   Only after it has matured does it start paying dividends.   

So paying 100% of transaction fees to the delegates is a middle ground.   Have the percent paid to delegates decrees down to 10% over 10 years as the network matures.   

On the other hand, paying dividends is marketing in its own right and is effectively paying every shareholder to grow the network.   

I think that transaction fees will more than cover infrastructure growth as the network grows.

Why set it by years and not by something more measurable like activity of some sort?

[Troglodactyl]

...
The best solution might be to make the tx fees grandiose from the beginning and let the market / shareholders decide then.

I think each delegate should be free to set his own transaction fee, and if they set them too high they can be voted out.

Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

I disagree with this.  Just because some small group decides they want to be "freeloaders" doesn't mean that the majority has to open their wallets and pay for them while they get all the growth benefits at no charge.  This would be like making taxes optional, no one will pay them.

It's not like majority can vote that the minority has to give them all the money or something, the minority still has representation.

My previous idea is everyone votes for an inflation rate and take the median, or some other way of giving the DAC flexibility toward growth or dividends.  0% inflation would be all fees are given to delegates with no dividend. 2% inflation would be creating new shares to fund growth.  -1% inflation would be destroying some percent of transaction fees for dividends (might be more appropriate for a long established DAC).

The way we take a snapshot now to honor a DAC, in the future, people might honor all shares of a DAC that vote for a particular group of delegates.  That way people who want to take the company in a different direction can fork it and bring the people who share that vision with them and then compete in the market.

That's right, the majority doesn't have to open their wallets.  They can take no action.  They don't have any more right to force the minority to open their wallets than the minority has to force the majority to open their wallets.  The default state is that each is free to use his own resources as he sees fit, but none are free to use the resources of others without their consent.  If I choose to do something that I claim will also benefit you, that does not entitle me to use your money to pay for it unless you agree to contribute.

Snapshotting to split a network by strategy is an interesting idea, but it sacrifices network effect for purity.  It might be necessary if the main network is headed for collapse and can't be turned, but it's a sacrifice technique.

[Agent86]

Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

I disagree with this.  Just because some small group decides they want to be "freeloaders" doesn't mean that the majority has to open their wallets and pay for them while they get all the growth benefits at no charge.  This would be like making taxes optional, no one will pay them.

It's not like majority can vote that the minority has to give them all the money or something, the minority still has representation.

My previous idea is everyone votes for an inflation rate and take the median, or some other way of giving the DAC flexibility toward growth or dividends.  0% inflation would be all fees are given to delegates with no dividend. 2% inflation would be creating new shares to fund growth.  -1% inflation would be destroying some percent of transaction fees for dividends (might be more appropriate for a long established DAC).

The way we take a snapshot now to honor a DAC, in the future, people might honor all shares of a DAC that vote for a particular group of delegates.  That way people who want to take the company in a different direction can fork it and bring the people who share that vision with them and then compete in the market.

[santaclause102]

Sorry, I meant if transaction fees were insufficient, not share dilution.  What I typed originally didn't make much sense...

read it right anyway...

[Troglodactyl]

Sorry, I meant if transaction fees were insufficient, not share dilution.  What I typed originally didn't make much sense...

[santaclause102]

The concern at surviving low compensation isn't solved by hard coding compensation, since any delegate can reduce his own compensation by destroying some of it manually, and can advertise these dividends to the shareholders.  If a state wants to run a delegate at low compensation, they're welcome to, but if they try to use it to exert power negatively, they should be voted out quickly just like anyone else would be.

Agree. That is the best solution.

If share dilution is insufficient I'd suggest convincing shareholders to reinvest directly and voluntarily.  Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

It is problematic but it is a workable solution a vast majority could perceive as just. Voluntary donation without getting something back is the best ethical solution but the outcome if no one donates because everyone is a small fish by himself might be worse than if everyone was diluted and everyone benefits. It is likely that everyone sees this as justified if it is widely accepted that investment makes sense for all.
The best solution might be to make the tx fees grandiose from the beginning and let the market / shareholders decide then.
edit: After having thought about it a bit more making tx fees grandiose wouldn't work of course. It would be only one stream of capital that can be reinvested and it might not be enough, especially at the beginning.

[toast]

I don't think it would be a good idea to hard code a share dilution for a fixes amount of time. Would there be a possibility to do a temporary share dilution without hard forking? Would be helpful, especially at the beginning, in case tx fees are not enough and the gains from adoption would be the highest.   

Why not just let all delegates keep as much of the fee from their blocks as they want, and destroy the rest?  If they abuse that, they'll be voted out quickly, and the market will naturally regulate the dividend rate and strategies for reinvestment.

This is a good idea and would be necessary if delegates also make reinvestment decisions. My concern was that an entity with an interest to harm the network that has a lot of financial backup (like a state) can survive this low compensation better. But there is probably no way to prevent it and a harmful actor could obfuscate its identity anyway. Could it?

I thought about whether it is a problem that the max. amount of votes a delegate can get is 2% of all votes (pro - contra votes). But is should be no problem if it is not possible to vote for a delegate that would have more than 2% if you voted for him. Then the personal or team behind the delegate has to set up a new delegate...

If share dilution is insufficient I'd suggest convincing shareholders to reinvest directly and voluntarily.  Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

The concern at surviving low compensation isn't solved by hard coding compensation, since any delegate can reduce his own compensation by destroying some of it manually, and can advertise these dividends to the shareholders.  If a state wants to run a delegate at low compensation, they're welcome to, but if they try to use it to exert power negatively, they should be voted out quickly just like anyone else would be.

Bingo. Right at launch we would have "dev delegates" and "dividend delegates". The shareholders pick whatever they like.

[Troglodactyl]

I don't think it would be a good idea to hard code a share dilution for a fixes amount of time. Would there be a possibility to do a temporary share dilution without hard forking? Would be helpful, especially at the beginning, in case tx fees are not enough and the gains from adoption would be the highest.   

Why not just let all delegates keep as much of the fee from their blocks as they want, and destroy the rest?  If they abuse that, they'll be voted out quickly, and the market will naturally regulate the dividend rate and strategies for reinvestment.

This is a good idea and would be necessary if delegates also make reinvestment decisions. My concern was that an entity with an interest to harm the network that has a lot of financial backup (like a state) can survive this low compensation better. But there is probably no way to prevent it and a harmful actor could obfuscate its identity anyway. Could it?

I thought about whether it is a problem that the max. amount of votes a delegate can get is 2% of all votes (pro - contra votes). But is should be no problem if it is not possible to vote for a delegate that would have more than 2% if you voted for him. Then the personal or team behind the delegate has to set up a new delegate...

If transaction fees are insufficient I'd suggest convincing shareholders to reinvest directly and voluntarily.  Allowing even a majority of stakeholders the discretion to dilute the stake of the minority without their consent is problematic.

The concern at surviving low compensation isn't solved by hard coding compensation, since any delegate can reduce his own compensation by destroying some of it manually, and can advertise these dividends to the shareholders.  If a state wants to run a delegate at low compensation, they're welcome to, but if they try to use it to exert power negatively, they should be voted out quickly just like anyone else would be.