A Better Coin

[SgtSpike]

- Hmmm... now you've got me all confused about the same thing.  Would putting the lost coin rate into inflation increase prices of goods over time?  I believe the argument for demurrage is that it would help keep prices stable, even though a person might lose coins over time.  So instead of a $1 burger being $5 30 years down the road, it would still be $1, but you would only have $0.20 left after 30 years.  And that would be a very extreme example with a much higher demurrage rate than what we are talking about here, but you get the idea.

Even more confused now, how can demurrage achieve this if it has no control over the demand for money? As I see it, demurrage just encourages an increase in money velocity, which has the same effect on inflation as an increase in money supply.

- Not much we can do about change in growth rate.  It's completely unpredictable.  You argue that the rate will grow due to technological advances, and I might argue that it will shrink due to limited resources.  This will introduce some amount of mild inflation or deflation, depending on whether growth rate increases or decreases.  Without including some sort of manipulatable index, there's no way to account for this potential change, so it must be accepted as a caveat of the currency.

But is there any evidence to suggest that we are reaching a peak in resources? Even if we were to reach a peak in some physical resources during the next decades or centuries, I don't think it would hit GDP growth too hard as information keeps grabbing an increasingly bigger share of GDP, and information can be almost infinite.

How about a compromise? Let's say I grab all the historical data I can find and extrapolate it to the year 2200 or whatever. I tell you then that we should expect an average annual world GDP growth of 9% by that time, so we start with your 4% supply growth and very slowly increase it over 10 million blocks or so until we reach the 9% which would be a hard limit. How does that sound?

- The initial 10 years (or whatever period of time is chosen) would be the "adoption period".  I would guess that this would involve lots of mining and very little actual merchant transactions.  The purpose of this period would be to spread coins to anyone who wanted them.  There should be no "early adopter" problem, where a single individual ends up with more than 1% of the currency.  We could have that time period only be 1 year, or increase it to 20, depending on whatever is deemed appropriate before the project is started.  Even after the block reward reaches maturity and only inflates by 4%/year, the actual inflation rate of the currency will be very high for several more years, as the new block reward/year will be much higher than the number of coins in circulation.

The problem with starting out with a minimum block reward, as you have suggested, is the early adopter problem.  It could easily end up with a single person holding 1% or more of the currency if enough people don't jump the gun and get onboard with mining with a constant block reward.  That's why there needs to be a "ramp up" to the ending block reward.  Now maybe 10 years is a bit extreme for that ramp up, so the actual amount of time to "ramp up" should be considered.

Ok I can go with that, but it should be like a year at most or we risk irrelevancy.

By "this", do you mean keep prices stable?  I'm not sure, maybe someone who knows demurrage more than I do can jump in on that question.

I don't like it.  I don't think we can assume the GDP growth rate will continue increasing.  9% economic growth sounds absolutely insane... haha.

Perhaps you're right about the ramp up stage.  1 year into bitcoin would have put it around, what, September 2010?  Maybe 2 years?  *shrug*

[1713580850]

1713580850

[1713580850]

1713580850

[JohnDoe]

If you reread your post #66, you'll see your only objection to steady deflation was volatility.  And then you even agreed that steady deflation, not volatility, was the problem later on:

By steady deflation I meant "always in a deflationary state", not something like "steady 3% annual deflation, forever" as that's not something that could happen. My argument was that it is virtually impossible to asses if an investment will pay off in an "always deflationary state" and that in most cases it will be superior to just keep the money and do nothing than make the investment and risk ending up worse off.

By "this", do you mean keep prices stable?  I'm not sure, maybe someone who knows demurrage more than I do can jump in on that question.

I don't like it.  I don't think we can assume the GDP growth rate will continue increasing.  9% economic growth sounds absolutely insane... haha.

- Yeah, by "this" I meant keep price stability.

- You were the one who suggested to look into historical data in order to extrapolate and get a more accurate number. I did just that but it seems you didn't like the numbers. There's no denying that the world keeps moving faster and faster, specially since the industrial revolution, and you haven't provided any evidence of the supposed wall we are about to hit, so I don't see how 9% is insane at all. Maybe some further reading on the subject could convince you: http://en.wikipedia.org/wiki/Accelerating_change

Remember that GDP is just the market value of all goods and services produced in a given amount of time. It would be possible for physical goods to account for 1% of GDP while information and services account for 99%.

- Block rewards start at 1, and increase linearly to a rate of 500,000/block over the next 10 years
- Block reward then inflates at a rate of 4% / year.  So the next year, it would be 520,000, etc.

Now that I think about it, with this scheme you would be increasing the money supply by way more than 4% for many years. Assuming there are 52560 blocks in a year and you want 4% annual inflation, then the block reward has to be 0.000075% of the money supply. 500k/block means you would first need a supply of 667,000,000,000 before you start increasing the reward, and you wont be anywhere near that amount when you reach the 500k/block rate.

[SgtSpike]

If you reread your post #66, you'll see your only objection to steady deflation was volatility.  And then you even agreed that steady deflation, not volatility, was the problem later on:

By steady deflation I meant "always in a deflationary state", not something like "steady 3% annual deflation, forever" as that's not something that could happen. My argument was that it is virtually impossible to asses if an investment will pay off in an "always deflationary state" and that in most cases it will be superior to just keep the money and do nothing than make the investment and risk ending up worse off.

By "this", do you mean keep prices stable?  I'm not sure, maybe someone who knows demurrage more than I do can jump in on that question.

I don't like it.  I don't think we can assume the GDP growth rate will continue increasing.  9% economic growth sounds absolutely insane... haha.

- Yeah, by "this" I meant keep price stability.

- You were the one who suggested to look into historical data in order to extrapolate and get a more accurate number. I did just that but it seems you didn't like the numbers. There's no denying that the world keeps moving faster and faster, specially since the industrial revolution, and you haven't provided any evidence of the supposed wall we are about to hit, so I don't see how 9% is insane at all. Maybe some further reading on the subject could convince you: http://en.wikipedia.org/wiki/Accelerating_change

Remember that GDP is just the market value of all goods and services produced in a given amount of time. It would be possible for physical goods to account for 1% of GDP while information and services account for 99%.

- Block rewards start at 1, and increase linearly to a rate of 500,000/block over the next 10 years
- Block reward then inflates at a rate of 4% / year.  So the next year, it would be 520,000, etc.

Now that I think about it, with this scheme you would be increasing the money supply by way more than 4% for many years. Assuming there are 52560 blocks in a year and you want 4% annual inflation, then the block reward has to be 0.000075% of the money supply. 500k/block means you would first need a supply of 667,000,000,000 before you start increasing the reward, and you wont be anywhere near that amount when you reach the 500k/block rate.

*shrug*  I just don't think it is reasonable to expect continued growth in the GDP growth rate.  I mean, are we someday going to hit 100% growth in GDP/year?  Not likely... so where do we cap it?  It just becomes an uneducated guessing game.  That's why I like an unchanging rate better - it just seems more reasonable.

I will do further reading on the subject though, as you suggested.  Certainly wouldn't hurt. 

Yes, inflation would be quite high for quite a number of years even after the 500k block reward is hit.  Kind of like how Bitcoin will continue to be inflationary for a good half-century or so, even though it is touted as a deflationary currency.  By my calculations, the 4% inflation rate would be reached around year 25.  I don't think there is any way around this without giving early adopters a huge advantage (or having a bell-curve-like distribution schema, which is an interesting thought), but hopefully, adoption would be able to match the huge inflation rate and keep it somewhat stable.  Or, at the very least, semi-predictable.  There is no doubt that the early stages of the currency would be filled with volatility though, just like Bitcoin.  That is downright inevitable for a decentralized currency.

[d'aniel]

If you reread your post #66, you'll see your only objection to steady deflation was volatility.  And then you even agreed that steady deflation, not volatility, was the problem later on:

By steady deflation I meant "always in a deflationary state", not something like "steady 3% annual deflation, forever" as that's not something that could happen.

Of course not.  My purpose for using that ideal scenario was to decouple the expected steady downward drift in prices seen over a large enough time scale from the obviously inevitable shorter term fluctuations, and addressing just the first problem of steady downward price drift.  It's these shorter term fluctuations that I was referring to as volatility, but what I see now is important for the discussion, and what I didn't specify, is the relevant time scale.

Going to leave the deflation discussion alone for a bit until I'm convinced it's currently relevant.  And don't worry, I'm not asking you to convince me this time  

[jtimon]

We have different concerns. Let's recapitulate a bit.

1) Early adopters reward

I''m not concerned about giving them too much nor about giving them too few.
With merged mining most miners will join even if the reward is low.

2) Initial Volatility

It's impossible to address this without price information input and we have agreed we don't want it.
Demurrage would still mitigate this partially by increasing V and making it more constant.

For long term price stability based on expected GDP growth...
Many experts claim we're currently at a peak in energy production and it will have consequences in the GDP growth. See this guy, for example.
I don't like the GDP statistics because they're usually manipulated by governments and monetary manipulations. See this site.

**I still have to read d'aniel's post #119 to undesrstand his proposal to solve this.

3) Miners long term reward

I think this is cunnicula's main concern.
There's only two possibilities: exponential growth of the monetary base or coin destruction equivalent to creation. Destructive fees will affect V the opposite way demurrage does.
So the solution with more stable M and V is demurrage.

4) Steady deflation

If we have a constant M and a pretty stable V, we're going to face steady deflation with growth and steady inflation with economic contraction.
But the financial effects of steady deflation are diminished with demurrage too.

5) Economic cycles

Some economists argue that even with sound money, the so called business cycles are caused by the debt exponential growth that interest favors.
During the first phase, debt growths and price inflation (note that credit competes with money as a medium of exchange) appears. When the growth cannot be sustained anymore, the (bad) debts start to be liquidated (forescloses, bankrupcies, etc) in some sort of chain reaction.
At that moment deflation appears (even with economic contraction and fixed M0, because credit is shrinking and it is part of M in some sense).
Printing to fight deflation only allows further debt growth, making it sustainable for a little longer and making the later chain reaction worse. Also "injecting liquidity" always represents an unexpected transfer of wealth, that springs mal-investments and ruins business plans.     
Since demurrage (unlike monetary inflation) lowers interest rates, it attacks directly the root cause of these cycles instead of only its final destructive (but clearing) effect, which is deflation. In fact deflation helps to reduce faster the general debt level.
If you study "free money" interest theory, you can see other problems with interest. Another effect of interest that Gesell didn't saw is the short-term thinking that interest impregnates the financial system with. 

I know most of you probably won't agree with me in that last part, it was just in case someone becomes interested in these economic ideas.
Of course, you can try to refute my reasonings.

It would be nice if we try to discuss the different problems separately.

[cunicula]

2) Initial Volatility

It's impossible to address this without price information input and we have agreed we don't want it.
Demurrage would still mitigate this partially by increasing V and making it more constant.

For long term price stability based on expected GDP growth...
Many experts claim we're currently at a peak in energy production and it will have consequences in the GDP growth. See this guy, for example.
I don't like the GDP statistics because they're usually manipulated by governments and monetary manipulations. See this site.

**I still have to read d'aniel's post #119 to undesrstand his proposal to solve this.

I clearly disagree with everyone else on this. Volatility is cause by unanticipated changes in demand. Theoretically, volatility is completely unrelated to any predetermined supply rule you choose (10% destruction, 10% inflation, constant supply, etc.). If you choose a non-predetermined rule, then you can affect volatility. If the money supply growth rate increases in response to an increase in the demand growth rate, volatility will go down. If the money supply growth rate decreases in response to an increase in the demand growth rate, volatility will go up.  Having supply linked to difficulty seems like a no-brainer to me. No one else has proposed a decentralized solution which links supply to demand, so it seems like the only reasonable option. If you don't go for this, then you might as well stick with bitcoin because you can't improve on its volatility properties.

[jtimon]

2) Initial Volatility

It's impossible to address this without price information input and we have agreed we don't want it.
Demurrage would still mitigate this partially by increasing V and making it more constant.

For long term price stability based on expected GDP growth...
Many experts claim we're currently at a peak in energy production and it will have consequences in the GDP growth. See this guy, for example.
I don't like the GDP statistics because they're usually manipulated by governments and monetary manipulations. See this site.

**I still have to read d'aniel's post #119 to undesrstand his proposal to solve this.

I clearly disagree with everyone else on this. Volatility is cause by unanticipated changes in demand. Theoretically, volatility is completely unrelated to any predetermined supply rule you choose (10% destruction, 10% inflation, constant supply, etc.). If you make the supply rule a function of demand information, you can either increase or decrease volatility. Having supply linked to difficulty seems like a no-brainer to me, but no one seems to agree.

I agree only partially with you. I think the system could decrease volatility only with demand information, but I'm afraid that's not a feasible possibility. The difficulty is related to the price (and price to demand), but not as directly as you claim. As JohnDoe pointed out, technology improvements cannot be predicted as easily as you propose. With a disruptive computation improvement the price could drop at the same time the difficulty rises.
Another proposal that I've heard for using internal information is to watch the transactions volume to measure V, but the problem there is that the system cannot tell if a transaction is a real trade/payment or a user paying to himself to move and reorganize funds.

[jtimon]

post #119

Mmmm. Your idea seems pretty centralized to me.
I suggest you read beertoken, stablecoin and reservecoin proposals in the list of my sign.

[d'aniel]

2) Initial Volatility

It's impossible to address this without price information input and we have agreed we don't want it.
Demurrage would still mitigate this partially by increasing V and making it more constant.

For long term price stability based on expected GDP growth...
Many experts claim we're currently at a peak in energy production and it will have consequences in the GDP growth. See this guy, for example.
I don't like the GDP statistics because they're usually manipulated by governments and monetary manipulations. See this site.

**I still have to read d'aniel's post #119 to undesrstand his proposal to solve this.

I clearly disagree with everyone else on this. Volatility is cause by unanticipated changes in demand. Theoretically, volatility is completely unrelated to any predetermined supply rule you choose (10% destruction, 10% inflation, constant supply, etc.). If you make the supply rule a function of demand information, you can either increase or decrease volatility. Having supply linked to difficulty seems like a no-brainer to me, but no one seems to agree.

I do agree that it's probably the best way to do this in an decentralized fashion.  I actually emailed a friend a while back when you first proposed it telling him how cool I thought it was.   

I just think it would be much better done with external inputs like exchange rates and human control, since difficulty has a significant lag on exchange rate changes, and the temporal resolution is quite large.  Not to mention potential anomalies such as ASICS coming onboard, and that it introduces a new avenue of attack.  You need humans anyway to distribute merchant subsidies and developer rewards, anyway, so why not let them help here, too?  This is especially the obvious way to go if you want to use the distributed central bank idea I proposed in order to avoid splitting the user base, and causing economic disruption.

[jtimon]

@d'aniel

If you have the external input of exchange rates in the chain, why do you need human management?
But as you said, it introduces a new avenue of attack.

[d'aniel]

post #119

Mmmm. Your idea seems pretty centralized to me.
I suggest you read beertoken, stablecoin and reservecoin proposals in the list of my sign.

Thanks jtimon, I will read them.

I think there are distinct short term and long term problems, and that only the short term ones are relevant now since there exists an obvious way down the road to address the long term ones when they become relevant.  (The particular solution details are not obvious - they're what you guys are coming up with here - but the way to roll them out is.)  So because the added centralization from this proposal only exists temporarily to give Bitcoin a leg up over the adoption hurdle, I don't see it as a problem.  Especially considering the organizational structure I proposed.

I also see it as necessary since these short-term problems clearly require, or would at least be much better solved using, external inputs and human management.

And it's especially beneficial since it avoids splitting the user base and causing economic disruption, unlike rolling out a whole new block chain or forking the existing one.

I'd also like to state that I remain quite hopeful that Bitcoin will make it over the adoption hurdle without this kind of help, and that this should only be seen as a kind of contingency plan.

Don't think I can make the case any better than that, so there you have it.

[d'aniel]

@d'aniel

If you have the external input of exchange rates in the chain, why do you need human management?

You do?  Apart from the rough proxy of difficulty, as cunicula suggested?

[cunicula]

2) Initial Volatility

It's impossible to address this without price information input and we have agreed we don't want it.
Demurrage would still mitigate this partially by increasing V and making it more constant.

For long term price stability based on expected GDP growth...
Many experts claim we're currently at a peak in energy production and it will have consequences in the GDP growth. See this guy, for example.
I don't like the GDP statistics because they're usually manipulated by governments and monetary manipulations. See this site.

**I still have to read d'aniel's post #119 to undesrstand his proposal to solve this.

I clearly disagree with everyone else on this. Volatility is cause by unanticipated changes in demand. Theoretically, volatility is completely unrelated to any predetermined supply rule you choose (10% destruction, 10% inflation, constant supply, etc.). If you make the supply rule a function of demand information, you can either increase or decrease volatility. Having supply linked to difficulty seems like a no-brainer to me, but no one seems to agree.

I agree only partially with you. I think the system could decrease volatility only with demand information, but I'm afraid that's not a feasible possibility. The difficulty is related to the price (and price to demand), but not as directly as you claim. As JohnDoe pointed out, technology improvements cannot be predicted as easily as you propose. With a disruptive computation improvement the price could drop at the same time the difficulty rises.
Another proposal that I've heard for using internal information is to watch the transactions volume to measure V, but the problem there is that the system cannot tell if a transaction is a real trade/payment or a user paying to himself to move and reorganize funds.

you can get around the txn volume measurement problem with a small % tax. However, there is a lingering problem. Txn volume tends to shoot up in periods of high volatility. Look at the stock market for evidence. In some cases, a price collapse could cause an increase in txn volume as people run for the exit. If the algorithm floods the market with coins in response, it would just make the problem worse. One possibility is to use only very long run trends in txn volume to adjust coin output. These long run trends are less likely to be related to volatility.

[cunicula]

Okay, I accept human control of currency generation. However, we need to tie human hands to prevent them from seriously fucking up. What if human input could produce a 5% rate of currency destruction annualy or a 5% rate of currency production annualy or anything in between, but that any changes outside these bounds are blocked in the code.

[jtimon]

@d'aniel

If you have the external input of exchange rates in the chain, why do you need human management?

You do?  Apart from the rough proxy of difficulty, as cunicula suggested?

Dacoinminster and morpheo (among others) propose different chains in which miners must report certain outside exchange values and the validity of their block depends on the network accepting those reports as truth or not. They can average various exchange and an error must be allowed.
If you know the prices for btc/usd, usd/oil, usd/gas, usd/rice, usd/corn, etc. You can make a target stable value to estimate dP.

[jtimon]

you can get around the txn volume measurement problem with a small % tax. However, there is a lingering problem. Txn volume tends to shoot up in periods of high volatility. Look at the stock market for evidence. In some cases, a price collapse could cause an increase in txn volume as people run for the exit. If the algorithm floods the market with coins in response, it would just make the problem worse. One possibility is to use only very long run trends in txn volume to adjust coin output. These long run trends are less likely to be related to volatility.

Yes, a small mandatory proportional fee should help.
I think high volume  should be interpreted as a symptom of inflation, not deflation, so the network would act the opposite way you describe.
With high volume, the demurrage would decrease but the reward would decrease even more (making dM negative).
Just exploring ideas...

[d'aniel]

Dacoinminster and morpheo (among others) propose different chains in which miners must report certain outside exchange values and the validity of their block depends on the network accepting those reports as truth or not. They can average various exchange and an error must be allowed.
If you know the prices for btc/usd, usd/oil, usd/gas, usd/rice, usd/corn, etc. You can make a target stable value to estimate dP.

Neat idea.  Would be interesting to see it in action.

Although with the current centralization of mining due to pooling, I'm not sure how this would be any less centralized than my proposal.

[cunicula]

you can get around the txn volume measurement problem with a small % tax. However, there is a lingering problem. Txn volume tends to shoot up in periods of high volatility. Look at the stock market for evidence. In some cases, a price collapse could cause an increase in txn volume as people run for the exit. If the algorithm floods the market with coins in response, it would just make the problem worse. One possibility is to use only very long run trends in txn volume to adjust coin output. These long run trends are less likely to be related to volatility.

Yes, a small mandatory proportional fee should help.
I think high volume  should be interpreted as a symptom of inflation, not deflation, so the network would act the opposite way you describe.
With high volume, the demurrage would decrease but the reward would decrease even more (making dM negative).
Just exploring ideas...

You are right, I shouldn't have been so careless about the relationship between velocity and price.

Our equation for change in price:

%ΔP = %ΔM + %ΔV – %ΔQ

Our goal in math: find a function f(%ΔV), such that if we set %ΔM = f(%ΔV), then E[(%ΔP)] = 0
Our goal in English: find a money supply rule that minimizes expected price changes given what we observe about velocity

One obvious f(%ΔV) to choose is %ΔM = -%ΔV, this takes velocity out of the price change equation. However, if %ΔV and %ΔQ are correlated, then this is not a good rule.
The best rule to pick is f(%ΔV) = E[ -%ΔV + %ΔQ |  %ΔV ]. However, we have no idea what E[ -%ΔV + %ΔQ |  %ΔV ] is and there is no obvious way of estimating it.

In short, we are up shit creek.
 

[d'aniel]

Okay, I accept human control of currency generation. However, we need to tie human hands to prevent them from seriously fucking up. What if human input could produce a 5% rate of currency destruction annualy or a 5% rate of currency production annualy or anything in between, but that any changes outside these bounds are blocked in the code.

All for tying human hands.

I was thinking another way the managers might be kept accountable and competent might be if several of these banks were operating for profit and in competition with one another.  There's nothing to stop viable for profit competitors from springing up if the stability etc. they offer is compelling enough.

[cunicula]

@d'aniel

If you have the external input of exchange rates in the chain, why do you need human management?

You do?  Apart from the rough proxy of difficulty, as cunicula suggested?

Dacoinminster and morpheo (among others) propose different chains in which miners must report certain outside exchange values and the validity of their block depends on the network accepting those reports as truth or not. They can average various exchange and an error must be allowed.
If you know the prices for btc/usd, usd/oil, usd/gas, usd/rice, usd/corn, etc. You can make a target stable value to estimate dP.

None of these suggestions has a workable method of ensuring honest reporting. Instead, they use democratic reporting. This might end up being honest, but there is no incentive for honest reporting in the system. Given how much nonsense is talked in the forum, would you really want the masses to decide how much money is generated?

I would put more trust in a dictatorship or oligarchy as these governance systems provide better incentives for truthfulness. The dictator loses big if he fucks up his coin. In a democratic system, the losses are divided among many decision makers so no one has a significant stake in making good decisions. I think we should vest all power in a few oligarchs, but tie their hands using hard coded rules. If we must use a democracy, then we should use even stricter rules. I really don't think a democracy is a good way to go. One oligarchic option is vesting all voting power in the 10 largest account balances. These guys are the 10 oligarchs. If someone else wants to be an oligarch, they can purchase the necessary coin.