Minting with constant per-block reward

General
1

During the chat with the community, Sunny King said that he is looking for a minting model with stronger incentive for minting while keeping the 1% inflation model. He mentioned minting with constant per-block reward. I consider this to be a major issue and in this discussion, I would like to provide as much feedback as possible and discuss about the consequences if such change should happen. Also, if there are alternatives that differ in details, we should try to find out which one is better.

Statistics of current minting
At first, I would present few statistics about the minting process as it works now. If we take 2000 POS blocks from the block 100,000, we find several numbers about the mint rewards:

Mean: 1.706
Median: 0.440
“Effective securing mint” calculated from mean difficulty: 0.225

If we sort all mints and take the 5% quantiles, we get:

 0%      5%     10%     15%     20%     25%     30%     35%     40%     45%     50%

0.000 0.099 0.130 0.160 0.190 0.220 0.250 0.280 0.330 0.380 0.440

 55%     60%     65%     70%     75%     80%     85%     90%     95%    100%

0.510 0.600 0.710 0.820 0.990 1.184 1.673 2.262 3.775 137.120

If we do the same for last 50,000 blocks (much larger sample, but also with bigger fluctuations), we get:

Mean: 1.751
Median: 0.310
Effective securing mint: 0.167

5% quantiles:

 0%      5%     10%     15%     20%     25%     30%     35%     40%     45%     50%

-0.07 0.06 0.09 0.12 0.15 0.17 0.20 0.22 0.25 0.28 0.31

 55%     60%     65%     70%     75%     80%     85%     90%     95%    100% 
0.35     0.39     0.45     0.53     0.67      0.91     1.37      2.32     5.08    1285.40

If all wallets would be minting and they would mint evenly, one would expect one mint to be:

‘all wallets minting evenly’ mint = money supply / 52595 / 100 = 4.03 PPC

Here, 52595 is number of blocks in a year. What is very interesting fact from these statistics is, that more than 60 % of the rewards come from 5% of rare mints with large rewards. The network difficulty corresponds to reward below median, it is much closer to 1st quartile (25% from the table). I do not know why is it so. As a result, I would say: yes, we should do something about this - although POS security is better than POW, we could do (in principle) 4.03/0.167 = 24x better.

mints2.png3300×2550

Constant reward, 90-days capped coinage
Here, I will try to specify the technical details - one possibility how could the constant per-block reward model look like. There are basically two things that need to be satisfied: time between blocks should be 10 minutes and inflation should be 1%. Let us assume that we do not change the way how probability of finding a blocks works, we only change the reward. This means that the coinage is still capped on 90 days for security reasons. How big should the reward be? It can be calculated from the 1% inflation:

block reward = money supply / 52595 / 100

Now, the block reward would be 4.03 PPC, with increasing money supply, it will slowly increase.

How does it help the incentive to mint? You need to stay online to take your share of the mints. This is a strong incentive, whenever you are offline, you are (statistically) losing profit at range 1% of your coins. Actually, you are losing even slightly higher profit, since whole 1% of money supply is divided among the actively minting wallets over the year, so the actively minting wallet typically gets “1% * money supply / actively minting PPC”, which could be around 3% or more. This is much better motivation compared to the state when you are just losing compound interest.

What is the position of small stake-holders? If you are holding 100 PPC now, you can expect to get a mint block approximately in 90 days. (60% at POS difficulty 13). With constant block reward, the difficulty will very probably rise as many people will want to keep their wallets online permanently to get the 1% profit. Since you should get reward 4 PPC in 4 years, you can expect to mint a block once in 1460 days. In such case, the POS difficulty would be 210. But since you are not competing with everybody, just with people who really do mint, you can probably expect successful mint once a year and a smaller POS difficulty. It is not that bad.

Suspected problems

[ul][li]Increased incentive to mint increases incentive to pooled minting. For small stake-holders, there is quite strong incentive for pooled minting. Even without the cold-lock minting, the small stakeholders might be motivated to give their private key to some trustworthy-looking pool. With the cold-locked minting, this might be severe problem to decentralization, which makes the cold-locked minting a sensitive issue we need to discuss properly. [/li]
[li]Coinage-based mint probability no longer makes sense and leads to wrong incentives. The increase in probability of finding block from 30 days of coinage to 90 days of coinage no longer makes sense. In current model, you are directly trading coinage for reward. In the proposed model, you are not. It is possible to divide your 10,000 PPC into 10,000 stakes by 1 PPC or into 1000 separate wallets by 10 PPC that you run separately. (There shouldn’t really be incentive to do that, but with coinage-based probability, there is.) In such way, you always lose only small amount of coinage, while the rest of your money keep it. Since this cannot be prevented (you do not know whose are which money), all stakes should have probability of finding block proportional to their coins, not to their coinage. [/li]
[li]With bigger incentive to mint comes bigger incentive to mint on many forks of blockchain and allow propagation of such mints throught the network. From what I understand, it is unsolved and important issue, that could lead to serious security thread, unless there is some solution for it. Also look here: http://www.peercointalk.org/index.php?topic=1956.0[/li][/ul]

2

[quote=“irigi, post:1, topic:2424”]block reward = money supply / 52595 / 100

Now, the block reward would be 4.03 PPC, with increasing money supply, it will slowly increase.

How does it help the incentive to mint?[/quote]

We don’t need 100% of the network to mint. It would be great to have let’s say around 5% minting participation. That way network stays secure and it would mean that incentive is not to strong for minters.
Why would strong incentive for minter be a bad thing? Because if incentive is stronger then we need it to be, it means we are overpaying for security.

If 5% of people are actually minting that means they will mint for whole year 1%*20=20% of their stack.
If reward per block is 4 ppc then it means somebody with 1000 ppc will solve 50 blocks per year etc.

If people think 20% return is big incentive then more of the network will join minting and they will drive reward down and up to some value. Most important thing is that in any moment there is enough minting participation to keep network secure.

Also for the stage we are in 1% per year might be low, it could be designed as bitcoin inflation to reduce over time. I think bitcoin design is really great from all angels except it doesn’t rely on PoS.

3

I also thought about this problem and my take is that if a LOT of people are trusting somebody will ALL their money then it means that that entity is trust-worthy.

4

I thought about another issue with the fixed reward:
I believe the “nothing at stake” problem is not really a problem because the incentive to mint all forks is too light (you only gain a few compounds interests). If the reward is constant then you have a strong incentive to mint all the chains. You don’t want to miss a single block. So the supposed problem may becomes a real problem.

There may be other situations where the fact the annual 1% reward was guaranteed led to conclusions that would become wrong with this change.

5

[quote=“sigmike, post:4, topic:2424”]I thought about another issue with the fixed reward:
I believe the “nothing at stake” problem is not really a problem because the incentive to mint all forks is too light (you only gain a few compounds interests). If the reward is constant then you have a strong incentive to mint all the chains. You don’t want to miss a single block. So the supposed problem may becomes a real problem.

There may be other situations where the fact the annual 1% reward was guaranteed led to conclusions that would become wrong with this change.[/quote]

I think problem with “minting all the forks while nothing at stake” is when you are able to mint consequential blocks with many stakes and that way you are able to double spend if you find long enough chain. That problem remains the same I believe.

6

we had some ideas about this also, see http://www.peercointalk.org/index.php?topic=2802,

thought about it a little longer, and say for some reason this (giving tx fee to minters) isn’t such a good idea maybe, because it could make the coin capped (if you replace it with the PoS reward), so , there I thought about it some more, and was thinking, hmmm hmm isnt there some coins that use demurrage, see freicoin?

add demurrage, for not doing anything with your coins for one year, they vanish (or just some percentage :slight_smile: )

to me it doesnt seem obvious right away why cap or no-cap is an advantage, I guess you can still keep it capped if wanted while using demurrage, and still use demurrage to incentivice minters

can someone explain myself to me? ???

7

I always thought that the more minting we have the more secure is the network.

If only 5% of the network is minting, one needs only 5% of the coins to make a 51% attack, no?

Or am I missing something?

This post is interesting to read in this regard: http://www.peercointalk.org/index.php?topic=2515.0

8

Actually, I noticed one problem with the constant reward per block model. It is not so much a problem as a feature: The increase in probability of finding block from 30 days of coinage to 90 days of coinage no longer makes sense. In current model, you are directly trading coinage for reward. In the proposed model, you are not. It is possible to divide your 10,000 PPC into 10,000 stakes by 1 PPC or into 1000 separate wallets by 10 PPC that you run separately. (There shouldn’t really be incentive to do that, but with coinage-based probability, there is.) In such way, you always lose only small amount of coinage, while the rest of your money keep it. Since this cannot be prevented (you do not know whose are which money), all stakes should have probability of finding block proportional to their coins, not to their coinage. I will add this into the problems section in the initial post.

[quote=“petar87, post:5, topic:2424”][quote=“sigmike, post:4, topic:2424”]I thought about another issue with the fixed reward:
I believe the “nothing at stake” problem is not really a problem because the incentive to mint all forks is too light (you only gain a few compounds interests). If the reward is constant then you have a strong incentive to mint all the chains. You don’t want to miss a single block. So the supposed problem may becomes a real problem.

There may be other situations where the fact the annual 1% reward was guaranteed led to conclusions that would become wrong with this change.[/quote]

I think problem with “minting all the forks while nothing at stake” is when you are able to mint consequential blocks with many stakes and that way you are able to double spend if you find long enough chain. That problem remains the same I believe.[/quote]
Could you please explain the “minting all the forks while nothing at stake”-problem to me, or send some reference to where it is explained? Thanks!

I think that the 1% inflation is great design and that it shouldn’t be reduced. Currently, the Bitcoin inflation is still above 10% and the drop in inflation presents a grave danger to its security model. When people will have to pay for security by transaction fees, they will be very unwilling to do that, IMHO. Yes, the POS is much cheaper, but since the minting always presents some (even if small) danger compared to paper wallet, motivation to do it is needed. Moreover, almost all economists agree that deflation is a bad thing for a currency. We may disagree with them, but I would still not risk deflationary model - 1% inflation is not that much! (And you can get the value of your coins back if you are minting.) If nothing, deflation may discourage people with money who trust what the economists say about deflation from buying PPC.

I believe that the 1% inflation model is much better than the ‘deflation model + transaction fees to minters’ (NXT). When there will be hoarding time (price going up), people will stop transact and the network will remain unsecured. When there will be (for whatever reason) drop in number of transactions, it is also drop in security, etc. (+ the potential problems with deflation).

[quote=“petar87, post:2, topic:2424”]We don’t need 100% of the network to mint. It would be great to have let’s say around 5% minting participation. That way network stays secure and it would mean that incentive is not to strong for minters.
Why would strong incentive for minter be a bad thing? Because if incentive is stronger then we need it to be, it means we are overpaying for security.[/quote]

As I see it, we are not overpaying them. We are introducing 1% inflation which is better than deflation, even if it wasn’t intended as a reward for securing the network. We are paying 1% not only to minters to improve security, but also for the stability of the currency, to avoid possible negative impacts of deflation. Too strong incentive for minters could be a bad thing for people, who prefer the best security they can get. Such people will never leave the paper wallets and we shouldn’t force them too strongly to mint, otherwise they will simply not invest into Peercoin. 1% Inflation seems to be a reasonable price for not participating in the network security, no?

[quote=“irritant, post:6, topic:2424”]add demurrage, for not doing anything with your coins for one year, they vanish (or just some percentage :slight_smile: )

to me it doesnt seem obvious right away why cap or no-cap is an advantage, I guess you can still keep it capped if wanted while using demurrage, and still use demurrage to incentivice minters[/quote]
We all understand that 1% inflation (when you don’t mint) and 1% demurrage is basically the same thing - either way you are losing 1% value of your coins. But psychologically, the demurrage looks much worse. Any the time people open their wallets, they will be wondering: “Why am I being punished? Why do my coins disappear?” It seems to me that the uncapped (but controlled) money supply will be much better accepted by people than the idea that they are losing money while holding them. But this is only my opinion, of course.

9

Yes, I can’t agree more. When/if we move to fixed reward we shouldn’t keep coinage in calculation. We will still need some locked period like few days or so, that is needed so that nobody could guess winning public key.

If you have let’s say 10% of minting coins in ppc or 10% of mining power in btc. You theoretically could still perform 51% attack if you are lucky enough. With btc for every failed attempt you would lose a lot of electricity so it doesn’t pay off, but with ppc you would not lose anything so you can try it with every block and even if chances are low to chain 6 blocks eventually you might succeed.

So if you have something like 10% of minting coins over 100 addresses you are trying all minting combinations (forks) to chain 6 blocks in small amount of time (faster then rest of the network) and use that to double spend.

[quote=“irigi, post:8, topic:2424”]I think that the 1% inflation is great design and that it shouldn’t be reduced. Currently, the Bitcoin inflation is still above 10% and the drop in inflation presents a grave danger to its security model. When people will have to pay for security by transaction fees, they will be very unwilling to do that, IMHO. Yes, the POS is much cheaper, but since the minting always presents some (even if small) danger compared to paper wallet, motivation to do it is needed. Moreover, almost all economists agree that deflation is a bad thing for a currency. We may disagree with them, but I would still not risk deflationary model - 1% inflation is not that much! (And you can get the value of your coins back if you are minting.) If nothing, deflation may discourage people with money who trust what the economists say about deflation from buying PPC.

I believe that the 1% inflation model is much better than the ‘deflation model + transaction fees to minters’ (NXT). When there will be hoarding time (price going up), people will stop transact and the network will remain unsecured. When there will be (for whatever reason) drop in number of transactions, it is also drop in security, etc. (+ the potential problems with deflation).[/quote]

Here I disagree when it comes to inflation/deflation issue. I know SK has said that he thinks small inflation is good but Satoshi on the other hand favors deflation. Also except Satoshi most austrian economists favors deflation, but I agree that general consensus in schools and colleges is keynesian. 1% is not that much and even higher wouldn’t hurt us much. But I will always be for design that will reach 0% inflation over time.

People can also be punished even when there is 0% inflation and their balance is intact. They are punished by missed profits.

10

I always thought that the more minting we have the more secure is the network.

If only 5% of the network is minting, one needs only 5% of the coins to make a 51% attack, no?

Or am I missing something?

This post is interesting to read in this regard: http://www.peercointalk.org/index.php?topic=2515.0[/quote]
Yes more minting is more security. But like with everything else you don’t go into store and say give me your most quality product regardless of price. If there is a lot of participation somebody (ppc holders) must pay for it.

Well actually if there is 5% minting participation, for double spend you can be sure that it can be achieved with 1%. But imagine that somebody holds 1% of btc (120k), do you think that somebody that invested would do anything to harm btc?

11

[quote=“petar87, post:9, topic:2424”]If you have let’s say 10% of minting coins in ppc or 10% of mining power in btc. You theoretically could still perform 51% attack if you are lucky enough. With btc for every failed attempt you would lose a lot of electricity so it doesn’t pay off, but with ppc you would not lose anything so you can try it with every block and even if chances are low to chain 6 blocks eventually you might succeed.

So if you have something like 10% of minting coins over 100 addresses you are trying all minting combinations (forks) to chain 6 blocks in small amount of time (faster then rest of the network) and use that to double spend.[/quote]
Great, now I understand. Thank you!

At the first glance it seems to me that this danger is the same both in the current coinage-based reward system and in the proposed constant per-block reward system. Is this correct? So the only difference is, that the constant per-block reward provides higher incentive to mint, which makes the “all forks minting”-attack proportionally more difficult.

12

Don’t know if it should be taken into account but I suppose that quite a few coins are lost every year by people loosing their private key or wallet password.

What percentage of the global coin volume could it be? If we consider the recent MtGox affair, it could be a lot :wink:

13

[quote=“mably, post:12, topic:2424”]Don’t know if it should be taken into account but I suppose that quite a few coins are lost every year by people loosing their private key or wallet password.

What percentage of the global coin volume could it be? If we consider the recent MtGox affair, it could be a lot ;)[/quote]

with demurrage they would not be lost anymore

14

This is my take on lost coins. When somebody lose any form of money coins, gold, dollars etc he is effectively just gave it to everybody else proportionally to their stake. So everybody else is better off in same amount as the guy that lost coins is worst off. And inflation is exactly opposite, guy that receive new money is better off in amount that everybody else is worst off.

So lost coins and in general “total amount of coins” is not that important imo. Only downside of lost coins that I see is from technical perspective, if somebody lose 100ppc we will always have to track that forever. When block chain grows to big we will be able to write client that starts from e.g 1 000 000th block but we will always have to track and carry all unspent balances including that 100 ppc balance.

Maybe some demurrage like if amount is not moved for 10-20 years then consider it lost and just drop it. It would reduce amount of memory needed for running ppc node.

15

From the long-term design perspective, this makes perfect sense to me. (Peercoin, after all, is a lot about long-term sustainability.) From the bussiness perspective, I do not dare to guess how many people would be offended by the fact that they must do something with their money once in 10-20 years. Hopefully not that many - this should be an understandable rule.

The fact that there will be an increasing amount of lost coins will also cause that coin market capitalization will not mean that much anymore for most coins and it will be less predictable how many active coins there is. But this is not such a big problem, in my opinion.

16

From the long-term design perspective, this makes perfect sense to me. (Peercoin, after all, is a lot about long-term sustainability.) From the bussiness perspective, I do not dare to guess how many people would be offended by the fact that they must do something with their money once in 10-20 years. Hopefully not that many - this should be an understandable rule.

The fact that there will be an increasing amount of lost coins will also cause that coin market capitalization will not mean that much anymore for most coins and it will be less predictable how many active coins there is. But this is not such a big problem, in my opinion.[/quote]

If we do this we kill the currency as a space currency. When we go into space and communication takes more then 10-20 years, the economies will separate and when one gets back to earth one will find that all those years cleaning toilets and putting food on peoples plates, just… evaporated. Even worse, those coins never even existed on earth. :pearcoin:

17

[quote=“irritant, post:6, topic:2424”]hmmm hmm isnt there some coins that use demurrage, see freicoin?

add demurrage, for not doing anything with your coins for one year, they vanish (or just some percentage :slight_smile: )[/quote]

Demurrage in Freicoin seems too radical for most people. I proposed a scheme to implement demurrage in POS reward, i.e. if you don’t mint often your reward diminishes, here http://www.peercointalk.org/index.php?topic=2810.msg25189#msg25189

18

FYI, Reddcoin will implement constant block reward PoS:

A constant PoSV block reward is what we are going to implement. In the FAQs document I even mentioned the PoSV block reward to be 3125. In Peercoin, the block reward is linearly proportional to the coin-age consumed in mining that single block. This means someone with a large holding only needs to come online once a year, solve a block and immediately gets paid all the interest. This is bad design. In PoSV, coin-age consumed only affects the chance of solving a block. As a result 5% inflation rate will be a probabilistic expectation.

http://www.reddit.com/r/reddCoin/comments/26m3z2/peercoin_considers_pos_modification_to_increase/cht061k

19

The “nothing at stake” problem appears when there’s a fork of the network. Since you don’t consume anything while you mint, when you find a new block you can propagate it on all forks. If a lot of people do that then reaching a consensus on the main chain takes more time.

It becomes much easier to double spend because everyone minting on all forks basically helps you doing that. So the more people doing that, the more confirmations we need to be certain to reach a consensus. And if everybody mints on all chains the coin is dead because there’s no consensus at all.

20

[quote=“sigmike, post:19, topic:2424”]The “nothing at stake” problem appears when there’s a fork of the network. Since you don’t consume anything while you mint, when you find a new block you can propagate it on all forks. If a lot of people do that then reaching a consensus on the main chain takes more time.

It becomes much easier to double spend because everyone minting on all forks basically helps you doing that. So the more people doing that, the more confirmations we need to be certain to reach a consensus. And if everybody mints on all chains the coin is dead because there’s no consensus at all.[/quote]
Thanks. What does the current protocol do to prevent that? Is this the reason to hold the coins of a stake that just successfully minted for some time? Is it any different for the constant per-block reward protocol?