[PPC] Coin-age conservation out of proof-of-stake kernel transactions?

[freequant]

Sunny King,

I read your whitepaper, but a few things are not clear.

• If there are no transaction fees, how do you prevent transaction "dust" spam?
  
  
• You define coin age as "currency amount times holding period" but you do not seem to make a distinction between the two ways a coin can be spent (as opposed to holding it), namely:
  1) by paying it to yourself in a kernel transaction
  2) by paying it to someone in a regular transaction
  
  But you do mention that you introduced one more timestamp.
  Is this timestamp used to track the age accrued between kernel transactions?
  
  If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :
  - In order not to waste the stake when doing a regular transaction, you could attempt to use your stake first before attempting any regular transaction with it, but if you haven't got that much stake on this input, you could end up waiting a fairly long time before managing to mint a proof-of-stake block, which could reveal unpractical if the other party is awaiting payment.
  - A solution to the above would be to proactively spend your stake so that your coins don't carry much and can be spent immediately at no opportunity-cost. But the downside of that is that if everyone does that, there will never be any reserve of stake in the network as the stake is spent in real time as it is accrued. The first undersirable side effect of that is that difficulty will be always max for everyone (no stake, not difficulty modifier) and the network will confirm transactions too slowly. The second much more worrisome effect is that attackers are the only ones to have an incentive to accrue stake, and can fork the blockchain with ease while other stake holders are struggling trying to mint blocks with small amounts of stake.
  - A third solution to transfer a coin without loosing stake would be to give away the private key of the address holding the coins to the buyer, and let him spend the stake before he transfers the coins to another of his addresses, but this is not secure as it requires to communicate the private key over non-secure communication channels, and the recipient won't be able to confirm the transaction before he first minted a block and then transfered the balance to one of his own addresses.
  - Last solution, the most likely: people seat on their coins and let them generate stake, and never use them for doing transactions but when they want to close their position of PPC. If this is what happens, the coin is dead, because its only purpose is to sustain its own existence, with no added value for anyone.
  
  On the other hand, if you do make a distinction between coin-age between kernel transactions and coin-age between regular transactions, this creates a different kind of problems: coins stop being fungible. Let me explain.
  If coins can be used in regular transactions in such a way that their stake isn't affected, then you will have coins with all sorts of stake levels being exchanged. But since stake can be used to mint proof-of-stake blocks that yield 1% commission, stake is like a coupon, and the more you have on a coin, the more valueable it is.
  if follows that all the coins on the network will start having different values, and their value can't be agregated anymore without making some approximation that custom implementations will refuse to do. It will also be a headache to exchange them with other currencies.
  
  
• Last question regarding the minting rate, and Proof-of-Work. In the paper you say "Over longer term the proof-of-work mint curve would not be too dissimilar to that of
  Bitcoin in terms of the inflationary behavior, given the continuation of Moore’s Law". Wnat do you mean?
  I I guess you make the assumption that difficulty will keep increasing because the hardware will keep improving. But this is not true. Difficulty would keep increasing following Moore's low given a constant or growing number of miners. In reality, some miners will stop mining the coin as the minting rate crosses the point where they would be more profitable with another coin. Accordingly the difficulty won't convege to infinite, and the minting rate won't converge to 0. Given enough time, the difficulty will likely converge to some asymptote, and the minting rate with then become constant and will define the coin inflation that could be significant.
  If you really care making proof-of-work come to an end for the sake of long-term energy efficiency, why not hook the minting rate to something more mathematically guaranteed, like a function of the block number that converges slowly to 0.

edit: added the url of the PPCoin whitepaper

[Sunny King]

Hi freequant,
  Thanks for the comments. See below answers:

• There is a transaction fee, only that fee is not paid to miners/minters, but is destroyed. We simplified transaction fee calculation and set a minimum 1 cent per KB fee. To further reduce dust spam we also enforce that minimum payment amount is 1 cent as well.
• Stake is withheld to maturity (520 blocks) so you cannot spend them until they mature. Once matured it returns to balance and then you can spend them. One item on my todo list is to automatically select coins with less coin age on them when spending, to reduce coin age loss of the user.
• Yes the proof-of-work mint curve is also heavily affected by market participation. If market participation stays constant then Moore's Law will ensure mint reduction. Even if Moore's Law discontinues and difficulty stops increasing (after many years), mint quantity would stay constant but inflation rate may not be that high. For example if you reach 100M money supply and 2M constant annual production, annual inflation is only 2%. Some alt-coins are already running with constant generation because over really-long term inflation is still approaching zero. But ppcoin should approach low inflation much faster than those, comparable to bitcoin's reduction schedule I think.

[freequant]

• Stake is withheld to maturity (520 blocks) so you cannot spend them until they mature. Once matured it returns to balance and then you can spend them. One item on my todo list is to automatically select coins with less coin age on them when spending, to reduce coin age loss of the user.

The problem with that is that it forces the participants to choose between either protecting the network by commiting their coins in what amounts to some sort of 512-block time deposit with 1% interest rate, OR keep their coins handy, so that they can spend them anytime, hereby supporting the economy.
If your economy is doing too well and people spend more than they hoard, you'll end up with a weaker network.
Conversely, if people hoard and put in time deposit, the economy will be anemic. Contraction of the money supply due to high stake usage could even spark a deflationary spiral.

Yes the proof-of-work mint curve is also heavily affected by market participation. If market participation stays constant then Moore's Law will ensure mint reduction. Even if Moore's Law discontinues and difficulty stops increasing (after many years), mint quantity would stay constant but inflation rate may not be that high. For example if you reach 100M money supply and 2M constant annual production, annual inflation is only 2%. Some alt-coins are already running with constant generation because over really-long term inflation is still approaching zero. But ppcoin should approach low inflation much faster than those, comparable to bitcoin's reduction schedule I think.

So you concur that, out a a specifc set of favorable circomstances, there is no guarantee that PPCoin will be prove energy-efficient in the long run...

A last question that you probably already answered, but no time to dig up earlier endless threads.
What does PPC mean?

[Sunny King]

The problem with that is that it forces the participants to choose between either protecting the network by commiting their coins in what amounts to some sort of 512-block time deposit with 1% interest rate, OR keep their coins handy, so that they can spend them anytime, hereby supporting the economy.
If your economy is doing too well and people spend more than they hoard, you'll end up with a weaker network.
Conversely, if people hoard and put in time deposit, the economy will be anemic. Contraction of the money supply due to high stake usage could even spark a deflationary spiral.

I don't think 1% annual interest is a valid trigger of deflationary spiral. Deflationary spiral is a debt problem as far as I understand.

Although it is possible that most users stop generating stakes and the network loses security. It's a possibility but of course the same can be said for bitcoin as well. If it does materialize that the incentive structure is not adequate to maintain reasonable security then improvements would be considered. For example, there is already a discussion on-going regarding the possibility of allowing minters to offline spend keys and reduce risk.

So you concur that, out a a specifc set of favorable circomstances, there is no guarantee that PPCoin will be prove energy-efficient in the long run...

A last question that you probably already answered, but no time to dig up earlier endless threads.
What does PPC mean?

Since we did not try to address the initial minting problem so proof-of-work is still needed, and we feel it's kind of arbitrary to fix a schedule to stop proof-of-work minting, so ppcoin remains a hybrid design and no preplanned phaseout of proof-of-work is built in. Also as a real cryptocurrency you have to consider from market's point of view if miners are all excluded from the ppcoin market after a short time period then adoption is likely very limited. That's why a relatively slow declining mint curve is chosen to match bitcoin's.

Even if ppcoin itself doesn't become fully energy efficient, we feel that the design already demonstrates how it can be achieved for future cryptocurrencies.

ppc = peer-to-peer coin

[freequant]

If it does materialize that the incentive structure is not adequate to maintain reasonable security then improvements would be considered.

You seem to rely on the assumption that current shortcomings in the design can be fixed in the future by modifying the protocol.
This is not a safe assumption to make. Have you considered that miners and stake holders may decide not to upgrade? Modifying the protocol comports significant risks of introducing regressions or vulnerabilities. People can become risk avert when they have money at stake and more to loose than to earn.
All it takes is one person stepping up, cloning the repo, and rallying people. This already happened with Solidcoin.

Wait-and-see / Ship-and-patch approach also makes the network weaker and more centralized, because it becomes reliant on maintenance and proper future decision making by developers.

[cunicula]

If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :

You are ignoring the tiny incentive involved here. 1% interest per year is a tiny motivator. In over 99% of cases, loss of 1% interest is not going to effect my consumer behavior at all...ever. Consider the last time you bought something... the current price is higher than the sale price. You think that in 1 year there will be a sale and the item will be discounted by 1%. Can you think of a single case where you wait to buy until next year because of the exciting 1% off sale?

[AndyRossy]

If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :

You are ignoring the tiny incentive involved here. 1% interest per year is a tiny motivator. In over 99% of cases, 1% interest vs 0% interest is not going to effect my consumer behavior at all...ever.

I'd like to see a study on this.

Also, I like doublec's idea of a proof-of-stake pool. 

[cunicula]

If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :

You are ignoring the tiny incentive involved here. 1% interest per year is a tiny motivator. In over 99% of cases, 1% interest vs 0% interest is not going to effect my consumer behavior at all...ever.

I'd like to see a study on this.

Also, I like doublec's idea of a proof-of-stake pool.  

There are plenty of studies discussing the effects of monetary policy shocks on the economy.  A readable treatment is here:
http://faculty.wwu.edu/kriegj/Econ407/Reading%20List/Bernanke%20and%20Gertler-Inside%20the%20Black%20box%20The%20Credit%20Channel%20of%20Monetary%20Policy.pdf
Take away message: a one-standard deviation increase in the federal funds rate (increase of about 0.8%, close enough to 1) decreases short-run GDP by about 0.1%.
This is a short-run effect, there should be no long-run effect at all. The effect occurs because this is a short-run fluctuation rather than fully anticipated, but let's ignore all that.
I'm happy to overestimate things here. After all, after grossly exaggerating all we are left with is a 0.1% effect!

Note that this is what we are talking about. FED raises interest rate by 1%, your bank interest rate goes up by 1%, price level tends to remain fixed for about 12 months.
You can now purchase either today, or 12 months from now at a 1% discount. The discount occurs because your bank pays higher interest on your savings while the price level remains fixed.

Outcome: in the aggregate, the economy purchases 0.1% less.

Now is freequent's discussion fear-mongering or is there some other word for it?

[cunicula]

• If there are no transaction fees, how do you prevent transaction "dust" spam?
  
  

Regarding txn fees. I agree that some portion of txn fees should be remitted to miners. Currently, txn fees are completely shared, in the sense that generation will = fee destruction in long-run equilibrium.

However, there is no incentive for inclusion of txns. Some small incentive should be introduced. We were discussing how this could be done before without introducing incentives not to broadcast txns (as in bitcoin).

I also agree that it is better to fix as much as possible as soon as possible.

Sonny, whack job idiots (aka users) will complain about everything you do; good or bad. Just look at bitcoin.

Better to fix things sooner, before you have a larger population of whack job idiots to contend with. They will try to prevent you from doing anything at all.

[molecular]

freequant, could you reference the paper you're talking about?

[freequant]

If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :

You are ignoring the tiny incentive involved here. 1% interest per year is a tiny motivator. In over 99% of cases, loss of 1% interest is not going to effect my consumer behavior at all...ever. Consider the last time you bought something... the current price is higher than the sale price. You think that in 1 year there will be a sale and the item will be discounted by 1%. Can you think of a single case where you wait to buy until next year because of the exciting 1% off sale?

You have a very naive view of economy.

Not all agents in an economy are consumers (or even people), and not all consumers are irrational.
Saving accounts and time deposit interest rates are lower than 1% p.a. in many countries.
And most countries have <1% yield on short term bonds.
Some countries like Hong Kong and Japan even manage to have a sub-percent yield on 10-year bonds.
Yet, they manage to raise trillions with these rates.

[freequant]

freequant, could you reference the paper you're talking about?

I added the url to the initial post.

[cunicula]

If you do not make a distrinction between coin-age between kernel transactions and coin-age between regular transactions, then spending your coins in a regular transaction would make you loose the accrued stake, but you wouldn't receive the 1% new mint. In the face of this, the most rational behaviors (in terms of outcome) are all undesirable for the network :

You are ignoring the tiny incentive involved here. 1% interest per year is a tiny motivator. In over 99% of cases, loss of 1% interest is not going to effect my consumer behavior at all...ever. Consider the last time you bought something... the current price is higher than the sale price. You think that in 1 year there will be a sale and the item will be discounted by 1%. Can you think of a single case where you wait to buy until next year because of the exciting 1% off sale?

You have a very naive view of economy.

Not all agents in an economy are consumers (or even people), and not all consumers are irrational.
Saving accounts and time deposit interest rates are lower than 1% p.a. in many countries.
And most countries have <1% yield on short term bonds.
Some countries like Hong Kong and Japan even manage to have a sub-percent yield on 10-year bonds.
Yet, they manage to raise trillions with these rates.

Okay, I cited some evidence documenting the negligible effect on demand (0.1%) of a 1% increase in interest rates (e.g. 0% -> 1%). Perhaps you can find some evidence for a larger effect. Go searching, maybe you can unearth 0.2%. I'll stand corrected.  Ohh, but wait 0.2% is still negligible.

Good luck.

Note: I did not try to clear up your misunderstanding. Here goes. The question is not whether you can raise money at 0% or 0.01% or 0.0001%. You can, even at 0 and in huge amounts. The question is
how much extra money you raise when you move from 0% to 1%. The answer: There is very little change at all.

[freequant]

Okay, I cited some evidence documenting the negligible effect on demand (0.1%) of a 1% increase in interest rates (e.g. 0% -> 1%). Perhaps you can find some evidence for a larger effect. Go searching, maybe you can unearth 0.2%. I'll stand corrected.  Ohh, but wait 0.2% is still negligible.

You cited one case where a change in key interest rates didn't have a measurable impact on some hardly measurable output of a chaotic system that involves more parameters than you can even imagine, within a time frame chosen abitrarily. That hardly qualifies as evidence of anything. But let's put aside the question of how unscientific are economic whitepapers in general, and whether economics deserve to be called a science.
More interestingly, you are making an analogy that doesn't fit because in your example the currency that is affected is legal tender, and there is no question of choice. The demand and offer are captive because denominated in dollars. Whether a small change of interest rates in these circumstances will or will not have an effect on the demand in dollar is complex but also irrelevant. In the case of cryptocurrencies, offer and demand aren't captive, and trades could happen outside of the system in "classic" fiat currencies. In these circumstances, the question isn't whether the interest rate will affect the demand and savings in general, but rather whether it will affect the proportion of the existing non-captive demand and savings that will seek fulfilment using this currency. In that case, it is all about incentives and counter incentives in using the currency for dealing or saving.
There is no simple authoritative answer on what would exactly happen. But if you think at the margin, 1% interest rate does create a positive incentive to use the PPCoin for time-deposit, which would arguably lead to a shrinking monetary base.

And even if you put aside the interest rate, there is still a major dilemma with PPCoin in the fact that savings secure the network, which creates a lose-lose situation because whatever is saved isn't used in the economy, and whatever is used in the economy isn't in time deposit securing the network.

Note: I did not try to clear up your misunderstanding. Here goes. The question is not whether you can raise money at 0% or 0.01% or 0.0001%. You can, even at 0 and in huge amounts.

If this doesn't involve a nigerian prince, I am all ears... 

The question is how much extra money you raise when you move from 0% to 1%. The answer: There is very little change at all.

Off the top of your head.
In the real world though, central banks do achieve macroscopic impact on the money supply by performing key interest rate changes of hardly more than 0.5% at a time.

[cunicula]

. In these circumstances, the question isn't whether the interest rate will affect the demand and savings in general, but rather whether it will affect the proportion of the existing non-captive demand and savings that will seek fulfilment using this currency.

Let's ignore Ben Bernanke paper then. "Helicopter Ben's" review article tell us nothing about monetary policy. Clearly, the guy is not qualified to write on this subject. Like you say, those ivory tower folks / Federal reserve guys don't know shit about monetary policy or economics in general. Real knowledge is only found on the bitcoin forum.

You are claiming that 1% nominal interest would kill the demand for purchases using PPC and that cause everyone to spend using fiat. Let's assume that this is true and generate a contradiction. The interest rate that is relevant for purchasing behavior is the real interest rate = nominal interest rate - inflation rate. Predictable increases in the money supply lead directly to predictable inflation. Let's assume that 100% of PPC users refrain from spending and diligently mine their stake blocks. This then produces 1% growth in the money supply. So what is the real interest rate? 0%. You get nothing whatsoever from delaying a purchase. You get some extra PPC (1% more), but the price has also risen by 1%. To reiterate, you mine 1% extra PPC and this compensates for the anticipated price rise. Therefore, there is no benefit at all from using fiat over PPC for expenditures. Accordingly, we know this corner solution can't hold.

In reality, there might be say 10% of money that is circulating and 90% which is sitting around mining. In this case, the nominal interest rate will be 1%, whereas the money supply growth rate will be 0.9 and the predictable component of inflation will be 0.9%. So there will be 0.1% real interest. So you do gain something by waiting. Let's consider how much you gain... If you want to buy something, and you delay that purchase by 700 years, you can now purchase two of that same thing. Incredible!

Maybe you are worried that inflation is procyclical here. It's not.

negative shock -> increased hoarding -> higher inflation -> positive short-run effect on economy

No downward spiral there. Instead you have a negative feedback loop.
Monetary policy operates just like a Keynesian would like in PPCoin (at least for proof-of-stake blocks, no comment on the proof-of-work portion). This is a bit of an overstatement. The Keynesian would probably want a lot more umph than just fluctuation between 0 and 1%.

[molecular]

freequant, could you reference the paper you're talking about?

after researching ppcoin for a couple of minutes it became pretty clear what you were talking about...

thanks for putting the think anyhow.