If stability is not why you are mining, I totally get that. I don't have a problem with competitive mining.

I'm just trying to sort out if anyone thinks stability is a feature. Is there a market for such a thing?

I think this is a similar idea to suggester's thread. However, after 18 months maybe sentiments have changed?

First Order (Monetary) Dynamics

So how can it possibly be done if it is not something that can be remotely monitored? Well it can't. But it can be increasingly approximated over time.

----

Basing GEM on electricity absent any GEM/$ exchanges.

1) All miners use the same current POW difficulty, as in bitcoin.
2) The initial mining proof-of-work (POW) difficulty starts with a wild-ass-guess, scaled so that (36,000,000 joules) mines roughly (1 GEM) on current best-in-class mining rigs. Hitting an exact target is not important.
3) Over time, the mean electrical value of GEMs will tend toward the electrical consumption of the most electrically efficient miner. (MHash/j)
4) Left alone, this electrical consumption would tend to decrease over time according to Koomey's law.
5) To return GEM mining to its original electrical consumption constant,
GEM offsets Koomey's law by doubling the proof-of-work difficulty every 1.57 years.
6) However, the difficulty jump is smoothed using scaled 10 minute increases with each new block.
7) Difficulty never declines.

----

Constraints:
a) If processor efficiency outpaces Koomey's law, there will be intrinsic inflation. Before: (1 Donut = 1 GEM)  After: (1 Donut > 1 GEM)
b) If processor efficiency falls behind Koomey's law, there will be intrinsic deflation. Before: (1 Donut = 1 GEM)  After: (1 Donut < 1 GEM)

Note: Extrinsic inflation and deflation (over/under mining) will be discussed in the next section.

This intrinsic inflation or deflation will tend to be predictable rather then wildly dynamic. If efficiency isn't following the current Koomey ratio, it is likely following a close cousin.

There exist no automated solution for detecting or remedying intrinsic inflation or deflation. Monitoring processor efficiency requires external human observation. In cases where technology makes extreme deviations from Koomey's predictions, I see only two possibilities. Either, everyone must tolerate the inflation/deflation, or, everyone must tolerate developers modifying the protocol ratio.

Consequences:
a) Mining rigs lose profitability over time. Since difficulty is monotonically increasing, eventually it will cost more in electricity for a given rig to mine then the resulting GEMs can be traded for.
b) Computationally speedy rigs mint faster. Electrically efficient rigs last longer. The electrically efficient displace the electrically inefficient regardless of computational speed.

GEM Basic Premise

One GEM always represents a fixed number of Joules. (Punny huh!) The price of the Joules used to create a GEM will vary over time and across locations.

1 GEM = (a fixed number of) joules = (a variable number of) $

For reference: (36,000,000 joules) always equals (10 kwh) which currently sells for about ($1)

The actual GEM/joule constant ratio will be an emergent property of the system. See, the "First Order Dynamics" section.


GEM Basic Goal

The goal is a digital currency where GEM value remains stable over time. Meaning, a future "basket of goods (BOG)" costs the same number of GEMs as it does now. Even if the $/BOG ratio changes over time.

If you delay gratification now, you receive exactly the same future gratification value. Nobody should save the value of a cup of coffee, expecting to exchange it for a steak dinner in the future. You are going to get a cup of coffee.

GEMs are not intended as long term investments. If you want to invest in a commodity that might appreciate in value, trade GEMs for BTC, or gold. If you want to earn interest on someone else's effort, you'll have to arrange your own GEM loans and negotiate your own terms.

If you can mine GEMs for $X today and sell them for more than $X in the future, knock yourself out. But consider yourself more an electrical futures trader than a GEM speculator.

[edit]
I'd like to start a discussion about stable crypto-currency. I'm just trying to get a sense if anyone is interested in non-speculative coins. As an example, I've extracted the concepts from EnCoin and simplified them for discussion into a system I call GEM. It is based on creating coins using constant electrical cost (in joules) over time.

---

I've been working for two weeks trying to understand the details of Etlase2's EnCoin proposal. Originally I thought the idea of basing coins directly on kwh was impossible. There are simply too many hidden variables to make it possible to monitor an arbitrary peer's energy consumption.

I have recently adjusted my opinion from "impossible" to "plausible within constraints". I would like to discuss these constraints and the resulting economic and social dynamics.

However, I can't in good conscience suggest anyone go read the EnCoin proposal until the the next version is released. Etlase2 is a bit hostile to some of the existing bitcoin design decisions. Reading philosophy mixed with economics only serves to obfuscate the points I want to talk about.

As such, I want to distill his underlying concept (as I see it) into its most important dynamics. To reduce the learning curve, I'll frame my discussion around a theoretical bitcoin code fork (and new block chain) that I'll call GEM.

---

I want to put all the GEM dynamics into one thread. But, I don't want to cause an argument cluster-fluck. As such, I'll post the outline of where I'm going, but I'll post the detail sections one at a time. Meaning, if discussion survives the first, we can move on to the second. If not, discussing the second dynamic would be pointless.

=======

GEM Basics
Premise
Goal

First Order (Monetary) Dynamics
Basing GEM on electricity absent any GEM/$ exchanges.

Second Order (Economic) Dynamics
Springs and Shocks to smooth GEM value convergence.

Third Order (Social) Dynamics
Variation from the existing Bitcoin protocols.
- 10 Minute Timekeeping
- Block Chain Consensus
- Network Partitioning and Reconciliation

Fourth Order (Moral) Dynamics
Detecting Fraud
Outing Fraudsters

=======

OK, then.

If you wouldn't mind PMing me when either the white paper or code is released. I would greatly appreciate it!

Yes, very odd language. I'd be happy with a white paper explanation for starters at least.

Does that require trolling or not trolling?

Has any description of the algorithm for preventing 51% attacks been posted anywhere? I can't seem to find anything but the claim in their forum. Am I looking in the wrong place?

If anyone has seen even a cursory description of that mechanism please post a link. I'm not expressing skepticism. I'm just genuinely interested.

I'm just curious. Do you have a link? I'm searching the solidcointalk.org board and I see that claim. But I can't seem to find an explanation.

I agree with you here as well. Any idea I would propose would contain a smaller "trusted" core of peers with the basic characteristics you designed into your "TrustNet/FreeNet/CoreNet" entity. I just think it could simplify the (transaction union/reconciliation procedure) that you go through in creating Primary Blocks.


Fair enough!

I think we are in, what an old professor of mine used to call, violent agreement. I think we are saying very much the same thing, but using different words. Inflation for one tends to be a maddening word for me. Normally I use it to mean goods prices are inflated against ENC. Meaning bread used to cost 1 ENC and now bread costs 2 ENC. But talking about buying and selling ENC using dollars the terms get more complicated.

In my mind if,
before: 1 ENC = 1 loaf = $1
after: 1 ENC = 1 loaf = $2   Then we are stable and have met our goals
after: 1 ENC = 2 loaf = $2   Then goods pricing, with respect to ENC, has deflated. Meaning mint more coins. Discourage hoarding.
after: 1 ENC = 1/2 loaf = $1  Then goods pricing, with respect to ENC, has inflated. Meaning destroy coins. Encourage hoarding.

But I think we are both pushing in the same direction, no matter what definitions we are using. So much so, that if I didn't reply to something in the previous post, assume we are in agreement.


I agree with everything in that section, but I'm confused by your use of "lower on occasion".

Wasn't the plan to lower deterministically (and smoothly) based upon the presumption that Koomey's law will hold? I think that is what you are saying. If so, I agree.

But on first read "on occasion" sounded like, "If things get wonky..." 


If you mean, "When the award lowers over time" then I totally agree with the rest of your statement. As far as I have been able to figure, people are required to pay attention and make decisions in their own self interest. Otherwise, they pay their own consequences with the electric company. (Like a gambler losing in Vegas)

However, THE GENIUS OF YOUR ENC=KWH IDEA means nobody else has to care.

If the award reduction/difficulty increase keeps everyone else aligned with Koomey's law. (for argument 1 ENC = 10 kwh) Then the run away minter is now minting at 1 ENC = 11 kwh. If he continues to do so causing an ENC glut and price inflation then he is compounding his loss because his 1 ENC buys < $1.

If enough people mint at a loss, it will temporarily trick the system into thinking the consensus is to add more money into circulation. Your proposed fee refund will begin dumping money into the market. This will temporarily increase inflation for the minter, but your "interest" compensates everyone else as an offset. This further compounds the runaway minter's losses.

But there is no way a runaway minter can profit. Like in Vegas, a time will come when he has to pay the piper. And that electric company piper is a real bitch!

---

"If the award lowers over time" makes it sound like, you still entertaining other concepts besides adjusting awards & difficulty towards Koomey's law? Are you considering another alternative.

Me too!

I think it might be possible to implement this economic model on top of the bitcoin code base. I'm kicking around in my head if the reconciliation/consensus stuff could be avoided by using varying speed block creation.

Someone wrote GeistGeld was experimenting with 15 second block creation times? I think the feasibility of that depends on how many peer nodes they have in the network. But if they could substantially speed up the rate without overwhelming the network, then their appears plenty of room for flexible block creation rates.

No, because this is not a trivia: Take bitcoin's code base, change one constant, recompile and offer a release. At least we haven't figure out if it could be done that way yet. Some ideas actually take work.

I'll keep working on this until we can simplify the idea into something easy to implement. If that becomes the case, I'll just do it. But if the minimal implementation is going to take several man years of coding, I can't commit that much time and energy for free.

Eitherway the work would go so much faster if you would send in that $8,000 already!

I'm pretty sure I'm in general agreement with you. Not sure I understand all the mechanisms you have for paying interest. But I agree you are pushing the system in the right direction.

The hardest thing about monetary policy is that you are optimizing against a hidden variable. Nobody really knows what the break even point is for ENC. Once you have a $/ENC exchange, individual decisions becomes easier (I know what my cost, and immediate return would be). But still if you wanted to know what the system's optimal ENC price should be, there are too many hidden variables to get an accurate calculation.

That's what led me to try and detect what the "human" consensus on the money supply was (too high, just right, too low). Technically, I think you could get away with detecting just two states (too high, too low). In that case the stable (just right) state would be when the system was oscillating between too high and too low. However, in my personal analysis, I always tend to think about the optimal stable state first. Then I look at how the other two differ from optimal.

I'm pretty sure you are working on the same detection issues. It sounds like you are detecting money supply is "too low" by how many coins are minted. Or perhaps a ratio of minted coins/traded coins over a fixed period. Your money supply is "too high" as that ratio tends toward zero.

I'd like to propose is ratio for discussion. (minted coins/trading fees) Where minted coins is the total number of minted coins in a given period.  Trading fees is the total number of destroyed coins in the above period. Meaning total number of traded coins times whatever fixed fee you decide on.

(total minted/trading fees) < 1
If this ratio is zero, there is a 100% consensus that there is too much money currently available for circulation.
In this situation, I propose you attempt to do two things.
1. Encourage potential buyers NOT to buy. (Hoard coins until prices recover)
2. Destroy the coins of panicking actual buyers. (Your coin destroying fee)
3. Encourage potential sellers to sell NOW! (Stop hoarding goods)

Psychologically, I suggest charging your *panic* fee ON TOP OF the price the seller is asking. As opposed to requiring the seller to hide that included fee in his price. The former tends to make clear to *panicked* buyers that prices are *already* considered too high. It also makes it clear to sellers that NOW is a great time to actually close the deal. The later tends to encourage sellers to inflate their prices (which is what we are fighting against). And if they think prices are moving higher, they are further encouraged to hoard goods.

I think you are saying, "Instead of actually destroying these coins, they go into a system managed hoard for later re-distribution." If that is what you are saying, then I agree. Hoarding those coins is monetarily indistinguishable from destroying them.


(total minted/trading fees) = 1
If you are minting exactly what it take to replace your fees then there is a 100% consensus that the money supply is "just right".
Who should receive the newly minted coins is NOT a monetary policy decision. It can be adjusted to encourage other behaviors as you see fit.


(total minted/trading fees) > 1
This ratio is unbounded, so it has no way to express 100% consensus that the money supply is "too low". But the farther this ration moves above one, the closer you get to that consensus.
In this situation, you have too many goods trading for the existing amount of coins. So I propose you attempt to do two things.
1. Encourage, hesitant buyers to buy NOW! (Stop hoarding coins)
2. Discourage sellers from panicked selling of goods. (Hoard goods)

As this ration tends above 1, your fixed transaction fee tends to fight against your monetary policy goals. You are destroying coins during a period where everyone is in agreement that more coins need to be minted. The consequences of this should be minimized whenever possible.

This is a good time to re-distribute all fees. Both the system's previously hoarded fees, AND the current transactions fees. Who receives those re-distributed fee coins is NOT a monetary policy decision. It can be adjusted to encourage other behaviors as you see fit.



I think I'm generally agreeing with this, as stated above.


I'm not sure I understand this enough to comment.

EDIT: LMAO, I really didn't understand what you meant when I started typing below. (Basing ENC reduction on time.) I thought you were talking about the logarithmic reduction to compensate for increasing CPU efficiency. Now that I read what I wrote, and re-read what you suggested. It seems like I just re-wrote exactly what you said. If that is the case, I guess I do understand it enough to comment!


But I wrestled with a similar sounding conundrum:
If (total minted/trading fees) greater than one, persists for multiple periods in a row. It seem like coins should be easier to mint. That is what I was getting at when I added the +2 level to my system.
If (total minted/trading fees) less than one, persists for multiple periods in a row. It seem like fees should be higher. However, both of these lead to more abrupt final transition states.

I hadn't considered the above ration when I was thinking about that though. Perhaps things could be smoothed by considering the ratio rather than time? [edit: ]

Say perhaps, the number of coins awarded per block is in proportion to (total minted/trading fees). Does that tend toward stabilization or toward more dynamic swings?

Fee's are more difficult to adjust because they affect the ratio itself. But you could add a "surcharge" as the ratio moved below one toward zero. Perhaps a surcharge using the inverse ratio (trading fees/total minted) makes sense. (As minting moves to zero, the surcharge moves to infinity.) That would provide a strong disincentive to panicked selling!

I have so many different versions in my head right now. I would have to read your latest on fee and interest to be able to respond coherently. I know anything I say right now would sound hopelessly lost in the weeds.

I'll sell you 10,000 now from the initial pre-mine. Just send $10,000 to... Wait, you are right. You deserve a pre-adopter discount so you can get some serious return on your investment. Just send $8,000 to me!

I'm not exactly sure what you are tying to fix?

But my first guess you are trying to smooth out the function so that it does not change so dramatically by doubling in difficulty at 18 month boundaries. If that is the case, you want to reduce the ENC awards per block say every month, so you have an 18 step smooth exponential curve from 2 to 1. Then at 18 month intervals, you both double the difficult and double the ENC award/block and start down the curve again.

I can't give you the exact values for this because I forget how to calculate with logarithms. But in effect if you had Log base 2 graph paper. You plot one point at 1 on the bottom left, 2 at the top right and draw a line between them. Then you divide the line  into 18 intervals and read the intercepts. That is the old school, before calculators way.

So the answer is, if I understood the correct question, there is an easy answer that I don't know, but could probably look up.
If you are asking, "What if it MHash electrical efficiency doesn't follow koomey's law?" Then I really don't have any answer at all.



Honestly, I never bought that premise. I think (absent a currency exchange) the baker never sets prices based upon electrical costs. Instead he bases them on the cost of flour, eggs, rent, etc. delimited in ENC. He doesn't care what his customer's cost to acquire that coin was. The baker cares about the trade value of the coin when he goes to spend it.

I'm not saying koomey's law is the perfect final solution. But I'm saying koomey's law is perfect way for the proposal to introduce people to your concept. It is well known, and it makes a good concrete example of canceling a known tendency with a known function to create your proposed constant. No hand waving is involved. That is what you want at the beginning of a proposal. It is like an establishing shot in a movie. It helps people suspend their disbelief.

But it's fine to say "koomey's law[1]" and refer to a footnote that says, "See section 10.5 for a discussion of koomey's law and other possible alternatives for maintaining this constant."



I'm not missing it. I'm not disregarding anything you are saying. Certainly, the price will never get too the lowest cost minter's cost. But that is the whole system's limiting condition.

I'm using the phrase "tends to" in the mathematical sense, as with Big O notation. "The time it takes to solve the proof-of-work based upon  difficulty (D) tends toward O(2^n). Sure that's not exactly right. The exact calculation would be: POW time = setup time + (time for one hash) * 2^(D-1) + completion time. But that just obfuscates the point I'm trying to make.

Again, I'm saying this is the best way to introduce the topic so unfamiliar people can grasp the system's dynamics. You can certainly put "electrically efficient minter[2]" and a footnote, "See section 10.6 on hashing speed vs hashing efficiency and how these effect total coins minted and the potential ROI of each minted coin.


I'm not saying your intentions weren't correct. I'm saying your execution confused me repeatedly.

(addition)
By the way, ROI in the absence of a $/ENC exchange becomes a much more nebulous concept. Sure nothing really changes, someone spends $1 in electricity but takes their ROI in 1.3 loaves of bread. But I would certainly leave that discussion to a later section.

Quite frankly, given your target audience, I don't think you are going to have any trouble convincing people to mint. They are already used to paying high electrical bills in hopes of higher future payback. In the absence of an exchange, I don't think the early minters/merchants would have a problem trading among themselves at (1 ENC = 1 loaf of bread). Then when an exchange appears, pricing their coins for sale at (1 ENC = 1.3 loaf of bread = $1.30), to those who don't want to mint. Over time, those two prices will tend to converge toward a single market price.

But even accepting all of that, I still have trouble wrapping my head around the ROI dynamics of that final single market place. I'm sure it is describable. I'm just suggesting it is not where initial discussion of your concept should start.

I'm going to respond to this tonight because I read this post earlier and I've been thinking about it already. I'll respond to the others later.

OK, I get what you are trying to say. (I Hope!) To paraphrase:

---
Each minter uses the same proof-of-work difficulty for minting.
As such, the value of ENC will tend toward the electrical consumption of the most electrically efficient minter. (MHash/ws)
Left alone, this electrical consumption would tend to decrease over time according to Koomey's law.
To return ENC minting to its original electrical consumption constant,
EnCoin offsets Koomey's law by doubling the proof-of-work difficulty every 1.57 years.
---

This makes total sense to me. I completely agree with this statement. I really hope it is what you have been trying to say.

Now a little needless bitching.


The part that needlessly confused me for two weeks was your use of averaging. While it is true that if one guy mints coins for 8 kwh and another mints coins for 10 kwh, the average amount of electrical consumption is 9 kwh. You seemed to also be saying that the sell price and value of each coin will reflect this 9 kwh average.

That seems ill conceived. In a marketplace, all the lowest price ENC sells first. Higher prices might not sell at all. Thus prices would tend toward the lowest cost, not the average cost.

I know now that is not what you were attempting to say. You were trying to say the more general, "If peers use more electricity prices will be higher. If they use less, prices will be lower." The use of specific numbers made it seem you were proposing precise targets.



This goes down as the worst sentence of the entire proposal! It made me think you were dismissing the likelihood of electrical efficiency increasing. All the talk about sunk costs, and the expense of upgrading made it seem like you were framing changes in overall efficiency as implausible. Having zero discussion of how you methodically change the difficulty level made everything seem even more hand wavy. "If in the unlikely chance that technology improves, we'll fix things by voting to change the difficulty." it was not very inspiring.

---

But anyway, I now understand the main trust of your proposal. (I hope.)

Perhaps my suggestion for phrasing it with help make your proposal easier for other doubters like me to wrap their head around.

So, given the above post about electrical efficiency.

In any system where stable coin value is a priority, (Exchange or no-exchange) as coin values approach stability, electrical efficiency becomes exponentially more important. If coins trade within 1% of minting cost, and I can mint for 2% less than you. Then I can mint and your can't. This is not so much a competition as an unavoidable axiom. (The efficient drives out the less efficient.)


Since I can't remove the axiom I'm attempting to use it as a feature.

The most electrically efficient state of the system, is where NO ONE has incentive to mint at all. This does not effect continuity of the network. Merchants continue to make money, clients continue to see the efficiency of the system as a personal benefit. It just means arbitragers don't mint, don't make profit, and don't extract value from the system in anyway.

Optimally, this would be the stable currency state. No new coins are required. No minting is done. No arbitraging is done.
However, I couldn't get my system this optimal. Maybe it is possible but I haven't notice that solution yet.

In the -1, 0, +1 system I proposed, in the stable state, some minting is required. However, it is collaborative (like you are proposing) and it does NOT involve arbitragers. In the stable state, only people with transactions will attempt to mint. If enough people attempt to mint, EVERYONE's tax is refunded. This is in effect a consensus vote that coin values are stable. It requires the combined tax rebate pool to spend only 1/2 the number of hashes as it would take arbitragers to create new coins.

The under valued state is the most efficient in my proposed system. If coin values are below stable values, nobody will mint at all. This serves two functions. Most importantly, it represents a consensus vote that coins are undervalued. Second, it removes all excess overhead from the system when overhead is the biggest hinderance to coin value recovery.

The over valued state is the only time Arbitragers attempt to mint at all. For maximum savings (to the system as a whole) arbitragers should mint as quickly and efficiently as possible. So if the most efficient minter can bring the value immediately back to stable himself that is optimal. If that is not possible other less efficient minters can step in to lend a hand.



The goal is that no monopoly of minters be allowed to drive the prices away from stability. (Tending prices toward zero or toward infinity.) That is how to best judge what I suggest. If we can show multiple solutions meet this goal, they can be ranked by how far they allow the stable prices to shift when drastic economy changes occur.

I realize now, that your goal is a variant of satoshi's. Create a coin that can be used in the absence of any exchanges. EXCEPT unlike BTC, you want ENC to be stable in value. Yes, this should have been obvious to me from the beginning. Let me explain why it was not.

Your initial premise was:
1) If coins creation could be made dependent on the consumption of a fixed amount of electricity, THEN (logical implication) coins would have a stable value absent any exchanges.

Yes, I recognized from the beginning that this statement is logically true. That implication does indeed hold. However, I recognized immediately that the antecedent "could be made dependent" was actually, False. While that does not disprove the consequent it doesn't help prove it either.

That was when I asked if you meant ENC = $ = KWH. Whatever your answer was, I interpreted it to be "close enough!"

So I substituted the original implication with:
2) If coins creation could be made dependent on the price of electricity, THEN (logical implication) coins could have a stable value. However, that *will require* an active ENC to $ exchange.


My substitution of premise (2) for (1) is the root of almost every argument we've had. In reality both assertions (1) and (2) are logically true. And in both, the truth of their consequent comes down to the truth of the statement "could be made dependent".

So the real question comes down to:
What made me "recognize immediately" that "could be made dependent" in (1) was False?

Well, technically, it is not false. If you made peer participation dependent on purchasing a specially made, tamper proof, efficiency matched, processing box that was cryptographically signed by a trusted supplier. Then you could assure that 1 ENC required 10 kwh. I dismissed solution out of hand, because it conflicted in philosophy with everything you wrote. But absent that, there is no way to remotely detect a processor's electrical consumption rate.


Everything in bitcoin is based around remotely monitoring (MHash/s). Hashes are monitored by knowing the difficulty level (2^(D-1)). Seconds are monitored by averaging block creation intervals. This gives you an accurate ratio, that you can use to recalculate difficult based upon a target interval of 10 min.

Note however, that monitoring the block interval accurately requires you to prevent parties from lying about block creation times. That is why bitcoin is winner take all. It prevents even collaborative lying. With block chains, the shortest intervals (most blocks) trump the longer intervals (fewer blocks). All of that is really about monitoring time.


Combined electrical efficiency (MHash/ws) cannot be monitored. You still understand Hashes. You can divide the number of blocks created (non-competitively, in a given interval), by the interval time. That give you a slightly sloppier measurement for seconds. But even given these, you have no way to monitor watt consumption (hence your 200W constant).

Side note: You proposed that block creation could cross PB intervals. That means you lose the ability to monitor seconds at all.

But even if you accept the above measurement of (MHash/ws) as being useful, changing difficulty (Hashes) does nothing to affect electrical efficiency (200W constant). You can only effect seconds. This leads to my example of the more electrically efficient driving out the less electrically efficient.

So, I presumed the (1) premise could not be yours, because the only solution to its antecedent was one you would completely oppose.

That is why I began thinking about (2) as the premise you were having trouble conveying to me. Really, it was nothing personal. I was just trying to figure out how you solved (what I called) the more important problems, that come later.