Topic: EnCoin Proposal v4.0 - scads of technical details - now with a WIKI! - page 2. (Read 6135 times)

To join the tradenet and partake in securing the network, 30 reputation points are required.
This initial 30 points can be gained in two ways: making at least 2 mint blocks per day in a supplynet group to gain 0.25 permanent points (120 days assuming each node is powerful enough to meet the minimum requirement), or having an account balance of 1,000 ENC to get 30 current points. The balance can't go below 1,000 ENC until you get 30 permanent points or the current points will be removed. So the money can't just be transferred around.

Once 30 points are reached, the only way to gain more permanent points is by receiving transactions and making a transaction block (being randomly selected out of 8,730 possibilities per day) that is later confirmed by the rest of the network reputation. 30 points is the starting reputation required for confirming transactions, so the signature weight that it has is worth 0. Since the transaction selection is random and there will be some checks for honest behavior (e.g. not getting transactions only after producing a transaction block), if all of those nodes had 1,000 enc, they would all have to be receiving transactions and associated fees as well before they would start gaining real reputation.

So, say there are 100k merchants and 100k botnets join. First they need 100 million ENC or they need to spend 15 hours or so a day working on mint blocks for 120 days, or roughly $21.6 million in electricity ($0.12/kwh assumed).

Say the average reputation of the existing 100k merchants is 60. To gain reputation between 30-60, a minimum of 5 ENC is required in tx fees to gain 0.25 reputation per day, and it is only gained if they produce a valid transaction block. Since 25% will be refunded of that 5 ENC, it costs 3.75 ENC per node per day for a possible 2,182.5 reputation per day, but they are only getting half of that because the 100k honest nodes are getting the other half of transaction blocks. So 3.75x100k = 375,000 ENC to gain 1,091.25 reputation per day. Since 100k x 30 (60-30 for the signature weight) = 3 million, 3 million / 1,091.25 = 2,749 days to equal the amount of reputation currently in the network, assuming it never increases among honest nodes. So a cost of 1.031 billion (375kx2749) ENC to equal the reputation and be able to fork the network.

And all they can do is fork. They can't reverse transactions because the honest nodes will not approve it and the cloudnet peers will only transfer honest data. They could take over the cloudnet as well, but there is no 50% in the cloudnet, if there are honest nodes honest data will eventually get around. If they do fork the network, the client will see that only around half of the total reputation signed the last consensus block so they are easily aware that something has gone awry. They can't change any of the underlying network structure (such as awards or whatever) because the clients know that this data is invalid. They could change account balances, but when the client sees that a fork has happened via reputation, they can ask for the transaction log and see that one side of the fork doesn't add up.

The only thing that can really be accomplished is the delaying of transactions. There will be backups to backups to backups of who is supposed to create a transaction block so that too much time doesn't pass.

Anyone who doesn't make a valid transaction block will lose a lot of reputation, or if they make an invalid block they will go very negative and their account balance will be either inaccessible or simply destroyed/redistributed.

TD-3, TD-5, TD-6, ATK-2, and ATK-3 covers these details in the proposal.

In essence, unless you work on creating a large chunk of the coins in existence, you won't be able to subvert the reputation.

Can you give the executive summary on how you expect to secure the network if not through hashpower? "Consensus and reputation" sounds vague and inadequate.

Also, you seem to suggest a supercomputer cluster and a PC are equals in your network when it comes to minting. Whats there to prevent a supercomputer cluster owner (or more likely, a botnet operator)  to present himself as 100.000 individual users? What would that do to your "consensus and reputation" model?

Merged mining does cover most of the risk. This is a few posts up. And the risk is a few dollars of electricity from an average computer. Just like the initial bitcoin adopters. You just won't see 3 million % ROIs. 300%? Well within reason. And that is on top of being awarded bitcoins as well. The one who takes the real risk is the one who designs the software and potentially wastes hundreds of hours of coding time. Let's not blow a few dollars of electricity out of proportion.

Give me a chance to respond, bro. See the other post, obviously, but to touch on bitcoin, yes bitcoin probably wouldn't be as popular without the built-in pyramid. However, the advantages of a cryptocurrency exist regardless of that pyramid, and I believe with time, people will see that these advantages are very real. And encoin will attract a different group, one that really wants separation from government-controlled fiat to a monetary system that is controlled by the people. Encoin does not need to grow to maintain value. It can grow as slowly or as quickly as necessary.


"Massive rewards" being the tune of a couple dollars per month. I can't control the price of electricity. But hashing power still costs a very significant amount of money. If you want to build that server farm, you have to plunk down the cash for those GPUs or whatever hardware you use. It will take months or years to begin to see returns. That's why I want the everyday GPU to set the bar. Something that is already a sunk cost in a computer. When people buy or upgrade their computers, they are generally not doing it so that they can mint coins faster. They are using it for its intended purpose of having a better computer. Someone who builds a rig or operation intending to profit on the coins must first pay off the costs associated with that operation. This could take many, many years, and in the mean time those upgrading their computers for fun will set the bar higher, lowering the profit margin of existing operations.

You never do. You simply take all the available information and make a wild-ass-starting guess. By for example saying the most efficient advertised GPU (etc) miner is X hash/joule. The most lowest known electrical cost where miners tend to be is Y. We want 1 ENC to be in the $Z range so (do a little math) the starting difficulty is (WAG).

At that point minters will mint what they want and stop when they want. If they want to mint ENC for $2 and trade them for $1 we don't try to stop them. However, the electric company eventually will.



There really is no "massive risk". The idea is that merchants want to trade in digital money. People want to trade in digital money. How many digital coins do we need? It's a matter of external value and coin velocity. The faster the coins circulate the fewer total coins are needed. If the external value to be traded exceeds the plausible velocity of the existing coins, the coin value rises. This stimulates the creation of new coins.



No one has to take the massive risk of generating 100,000 ENC worth of coins up front. Minters can start by generating 100 ENC. If there is a growing demand, they generate more.

I have some difference from Etlase2 on how cutthroat the competition should be. I think the most efficient should win. He likes to spread the wealth around a little more. He can explain why.

So it is nothing different than Bitcoin.   If efficiency is based on the network average than those who can undercut the network can reap massive rewards.  For example even at $3  Bitcoin is massively profitable (compared to conventional businesses) if you can find a location that has say $0.04 per kWh or less. 

Actually, early adopters will be rewarded. How, exactly, is up for debate, but I do have a discussion point on it.

DP-4) Encouraging Early Adopting
With ENC being difficult to create from the start (compared to Bitcoin), several options exist to encourage people to join the Network:

•   Low Initial Difficulty: By setting the initial difficulty on the low side, users who join within the first 10 CCPs will be rewarded with more coins while the difficulty catches up with standard hardware. The effect of these early coins will be continually reduced as more people join the Network.
•   Donations: Because of the low initial difficulty, it is hoped that users will choose to donate to services such as a “faucet” similar to Bitcoin, as well as for bounties to encourage people to develop products necessary for the advanced functioning of the Network. All donations will be voluntary.
•   Beta Testing Payment: As the Network is quite complex, many areas will have to be beta tested prior to official release. Since there is no block chain to build from, initial accounts can be given a strict, set amount as payment for beta testing the software.
•   Merged Mining with Bitcoin: Assuming the SHA2(SHA2()) hash function is used for creating coins, users of the Network could be paid in both ENC and BTC for the same work. If the Network begins this way, it will be intended from the start to eventually break away from merged mining to let the Network stand on its own. But in the beginning, this could absolve almost all risk in switching to ENC.


Merged mining is certainly the biggest boon. And low initial difficulty is not like the low initial difficulty of bitcoin, it will be nowhere near the same order of magnitude. It will be a starting point where the network will work to get to a stable cost to produce. I can't predict what type of hardware or how efficient that hardware will be from the start anyway.


Actually, I was referring to koomey's and moore's laws. Most of the "kilowatt" coins only cover moore's law.


Because some of the rules governing difficulty and awards are based on the assumption that an average peer is one person with an average computer. I don't want botnets or supercomputers to be able to monopolize the supply or increase the difficulty for everybody else. In fact, I want to make it unprofitable for them to do so. I can't guarantee that, but I can make it a lot more difficult with design decisions. I believe it is better for the adoption of the currency if anyone can "get into the game" and make a few coins if they want to. It shouldn't be a constant competition of who can spend more money to make a faster rig to take coins away from you like it is with bitcoin. I don't want there to be a $1000 start up cost just to be competitive. That is my opinion, and it is reflected in those design decisions.

How do you know who is the most efficient miner, what their efficiency is, and what their electrical costs are?

Why would an early adopter take the massive risk of ending up with worthless alt-coin for no reward?  Or no reward that provides appropriate compensation for the risk taken?  Had Satoshi done that likely we wouldn't even have Bitcoin (as flawed as it may be) right now.  At some point a project need to get to the point where rubber meets the road and risk needs to be compensated.

I was mildly interested in the idea but the more I learn the more it seems to be some socialist fantasy land where people take risk for no reward and nobody is able to undercut competitors or profit too much. 

Yeah, Encoin is different like that. People are rewarded for producing/trading goods and services. The coins are designed just to facilitate this exchange. They are not designed to be goods themselves.



Encoin doesn't need to collect or even try to deduce this information. The system starts from a wild-ass-guess (WAG) as to what the difficulty should be. Then it stabilizes on an ENC = X kwh relationship based upon the most efficient of competing minters. After that point it should trade in a narrow range.

Wait

1) Not rewarding early adopters is the stupidest thing I ever heard.  Think Intel would have took the massive risk in making first personal computer microprocessor had it not been for at least the potential to be rewarded for the massive risk they were taking.

2) How exactly are you going to know electrical cost and efficiency of every miner? 

I said a few things to help people understand basic concepts. They might not exactly agree with your latest proposal. I suggest you discuss the two concept below based on how Encoin aspires to be different from my rudimentary GEM principles.


This was difficult for me to understand because you mixed intrinsic inflation (variances from Koomey's law) from extrinsic inflation (variances in demand).

Can you explain the basics?


You care more about leveling the playing field than I do. Perhaps you should explain why you do. And how you compensate against efficiency.

Remember that peers are randomly assigned to SNGs:

TD-9) Mint Block Awards
A Mint Block is awarded to a SNG when a peer finds a hash value that beats a target hash value for a given problem. Peers are then paid out based on how close they were to the winning value.

To encourage SupplyNet peers to use “honest” hardware (everyday GPUs), there are five restrictions to a Mint Block Award:

1.   To be paid in a MB, a peer must have found a minimum target value (TBD – probably 1/3rd or 1/4th of the target value). This prevents very efficient, low-output machines such as FPGAs or ASICs from making it in to most blocks.
2.   A minimum of 20 peers must be paid out. A supercomputer/botnet/4x GPU system will have to wait for other peers before submitting a block. It will also be subsidizing those other peers as the payout structure is fixed.
3.   The payout structure is one where 65% is awarded to the top 50%, 35% is awarded to the bottom 50%. Individual payout structures within those halves is TBD, but 1st place will receive a bonus to incentivize competition.
4.   A minimum amount of time must pass before a SNG may submit a winning MB (likely 1/3rd of the average time to create a block based on the last 10 CCPs – see TD-1). This prevents a supercomputer from being able to adjust the difficulty too much (see ATK-1). It also allows more peers time to find a minimum target value.
5.   If a maximum amount of time (in coin-hours, likely 3 times the average) has passed before submitting a block, the coin award will be reduced.


Farms will have a massive set-up cost that will take many, many years to break even. Maintenance, etc. means it may never be insanely profitable. Regular GPUs will keep upping the difficulty over time, so this server farm will have to keep upgrading to keep up. SupplyNet members need to take part in the network as well, so if a server farm is split among 100 SNGs, it must send/receive 100x the data. Perhaps this will never be a big concern, but it might be.

When the award is dropped as I described in TD-2 on the last page, more efficient miners can realize a greater profit by increasing their output. They can "game" the system by not raising their output, but then they are reducing their potential profit. And everyone has to agree to game the system this way, or the difficulty will still rise.


*shrug* This is really no different from bitcoin. If demand is too high for cheap electric economies to keep up with, then the opportunity opens for those who pay more for electricity. It really is not intended to be a profiting venture, it is meant to pay minters for their services and allow those who want to store wealth without worrying about the machinations of fiat the ability to do so. As well as being a nifty way to send money wherever you want.

The system is bounded by the most efficient producer in terms of electrical efficiency (times) electric costs. So those using efficient hardware in low energy cost areas will run other "minters" out of minting.

I can't answer the "what if electricity prices tend toward zero question." In the GEM system I laid out that contingency requires manual intervention. EnCoin handles it differently. Etlase2 will have to explain his concepts there.



All alternatives compete, but they have the same bound, electrical efficiency (times) electric costs. Personally, I don't care who creates the new coins. I only care that they get created in a timely manner.

However, unlike Bitcoin, minting isn't a continuously running process. Minters will mint until the trade value of the coins is below their electrical cost to create them. Then they stop. When the trade value of the coins increases again, all minters race to create new coins until falling values force them to stop again.

Are you referring to my post just above? because this "The more people that want bitcoins, the more they cost to create." was like the 4th sentence.


Free electricity is never free. Someone is always paying for it and that is all that really matters. Effort was made to produce the coins. And encoin discourages high-end rigs by using a payout structure. No matter how awesome your 8x gpu rig is, you still have a maximum award per block and a few other rules that basically means you want to be only as efficient as necessary. Being too far above everyone else just means you're wasting electricity. You could divide up your rig into multiple peers on the network, but you still have to account for the high initial costs of these rigs when the profit per month will be quite low. You are no longer competing for a fixed amount of money, a thousand other people could join the supplynet when the price is high and take all of your hopeful profit right out from under you.

Encoin is the opposite of Bitcoin. If you like the idea of a fixed number of coins that grow in value based on population growth, you will hate Encoin. There is no early adopter premium in Encoin. Nor does ENC grow in value based upon increases in the trading economy nor number of users.



Supply is constrained by generation cost bounded by trading value. It doesn't make sense to spend $X in electricity to generate an ENC when it is worth $X in trade. This will stimulate new minting which brings the value back down to $X.

Quote from: P4man on October 31, 2011, 06:23:56 PM

That makes no sense at all, if there is one thing you dont need or even want to be stable its hashing power. You want it to grow with the value of the block chain.

The transaction record is not secured by hashing power. There is no concept of a hidden 51% attack that can change history.

Or large enterprise setting up massive farm in an ultra low energy cost country?
What about the effects of falling energy prices?  
Say a breakthrough in solar power bringing cost below current grid parity?

With electrical costs varying from free/stolen/unmetered to as low as $0.01 per kWh up to $0.48 per kWh any price point selected results in massive profit for some and non-viability for others.

At least existing systems allow alternatives to compete.  For example Venezuala has ultra low energy costs (subsidized) but highly developed countries like Finland have high standards of living meaning that FPGA are more viable alternative. 

What about those that have free electricity or steal coins or steal electricity ( broken power meter BS ) etc. How will EnCoin sort that aspect out ?

No time to read it all, Ill read it later, but from what little I read, you seem to think production cost has anything to do with price? BTC price doesnt adjust to mining cost, its the opposite. Whatever scheme you come up with that regulates the production cost will have no effect on the value, the only thing that will remain stable is hash power for a given electricity cost. That makes no sense at all, if there is one thing you dont need or even want to be stable its hashing power. You want it to grow with the value of the block chain. And if you regulate neither supply nor demand in, there is nothing to prevent the exact same excessive speculation that is causing btc volatility.

If I got that part right, there is no point in reading the rest.

I'm sorry, but you are wrong. It has nothing to do with short term fluctuations, it has to do with long-term growth. There is a very limited supply of money, and a large percentage of it has already been produced for less than what later coins will be produced for. The more people that want bitcoins, the more they cost to create. This does not mean 1 BTC costs 1 BTC to produce on average. If 7.5 million BTC cost $2 to produce on average and bitcoin doubled in popularity and the award halved, and a BTC now costs $8 to produce on average between 7.5 and 15 million, that means the average cost to produce was $4 but currently costs $8 to produce. That value goes to those who mined or bought coins when they were $2 to produce. This can be brought out to extreme examples. It is meant to be deflationary, it is meant that later coins will cost more to produce assuming a growing economy. It definitely does not mean that 1 BTC will cost, on average, around 1 BTC.


Instability in the value of BTC. The halving of the hash rate because of the price halving does not suddenly refund those who didn't sell their BTC when the hash rate and value was double. I want a system where people can actually confidently store their wealth rather than having it be an investment vehicle that is, as far as anyone can tell, just as likely to go up or down. And it is much more likely that early adopters are the market force rather than any combined thinking of the people. This is power removed from the people in a manner very similar to government-backed currency. I do not believe that bitcoin is an improvement at all in this respect.


Rather than pooling resources together in large groups like bitcoin, small groups of peers are randomly assigned together to create coins (supplynet groups). There is no limit on the amount of groups there can be. Each group only competes within itself to grab a share of the mint block award which is proposed as either 4.5 or 6 coins. Based on the mint block rules described in TD-9, it is very un-advantageous to use a botnet or a super computer or even a 4x GPU rig to try to mint coins. It encourages everyday GPUs which means less overhead in creating coins; just electricity, not a huge investment in hardware.

Based on how quickly coins have recently been created, difficulty increases in a manner similar to bitcoin, except the difficulty is on a per-coin basis. However, after a certain time period has passed, the award will drop to 4.5 coins. This means more efficient machines can run at full power and still be profitable while less efficient machines will now make less money. This again increases the difficulty to account for improvements in electrical efficiency. When the award is raised again to 6, the difficulty will multiply back up now taking in to account this efficiency increase.

So it is not just based on hashing power = coins, it takes into account the improvements in hashing power over time plus the improvements in hashing power based on a restricted supply.

The example from TD-2 (assumes no improvement in hashing power based on hardware upgrades):

EXAMPLE:
1.   The current difficulty for creating coins is a value of 100 which causes the average coin to be produced in 50 coin-hours.
2.   The Network originally had 100% of computers producing coins using 150W of electricity to produce a coin in 50 coin-hours, 50 * 150W or 7.5kWh per coin.
3.   50% of the computers producing coins now use 125W of electricity while 50% continue to use 150W, while both produce coins at the same rate.
4.   The cost to produce a coin in 50 coin-hours is now 50 * 137.5W or 6.875kWh.
5.   When the Mint Block award drops to 4.5 coins, the difficulty drops to 75 as well and the client scales back computer output to 75% so that 125W computers are using 95W and 150W are using 115W.
6.   By using the client calculator, those using 95W can easily see that they will make more coins at a profit by increasing their output to 115W (17%). Those using 150W computers cannot profitably increase their output because they are only getting 4.5 coins where they were once getting 6. The client can even be used to automatically increase this power output based on the market price supplied by the user (or by having the client contact a site in lieu of this).
7.   Now the original 125W computers are producing coins at a 17% faster rate than their 150W competitors. This will cause the difficulty to increase by 8.5%.
8.   When the block award returns to 6 coins, the difficulty will be 108.5, or 54.25 coin-hours to make the same coin as before. 54.25 * 137.5W ~ 7.5kWh.

kWh is only used as an example, but the cost per kWh is important as well, and this still works. Even if fusion comes around, it still works. People can pour more electricity in in times of short supply to make money. It's all market-based. The coin won't have a set kWh value, it will have a market value and it is up to the minter if they can profitably make coins.

The variable money supply allows the supply to grow as the economy grows, something bitcoin is designed not to do because "printing money is bad." Well this money is actually backed by its difficulty to produce; *always*. Not easy to produce from the start and progressively more difficult to produce, in more ways than one, as it gets more popular and time passes. And the difficulty to produce will remain relatively constant over time.

This is possible because hashing power is not required to secure the network, but describing that is the bulk of the proposal. In a few, quick points:

* Every transaction has a mandatory fee of 0.25% (min 0.05 max 10).
* Merchants are encouraged to secure the network by joining the tradenet, in which transaction fees will be refunded beginning at 25% on a tiered scale based on their reputation.
* Reputation is gained by receiving transactions and producing a confirmed transaction block containing transactions in a 10 second window. It is harder to gain reputation (more tx fees and txs are required) as your reputation increases.
* Reputation is awarded only once a day, so all merchants compete for these transaction blocks (that are randomly assigned based on previously confirmed transaction blocks). As more merchants compete, it will be tougher and tougher and cost more and more (in tx fees) for an agency to subvert this reputation.
* Merchants are required to hold a certain amount of ENC to reach the higher fee refund tiers (and reputation), which both creates demand for the currency and a real cost for anyone trying to subvert the reputation. When a node subverts the reputation intentionally by confirming a bad transaction, they will lose the money in the account (somewhat up for debate, but if it is a hacked account they are going to lose the money anyway).
* If they subvert reputation by splitting the network, all they can accomplish is delay transactions until reputation reaches a minimum of 50% of the total reputation ever seen in a day (see ATK-3 for more info). Assuming they are trying to make their own network, the "cloudnet" will be able to send transactions between both networks, and both networks will kill the others' reputation. Average users will be able to see which network has your Amazons and Neweggs and such while the other has a bunch of unknown and shady "merchants." They can't change even basic network operations because even end-users won't use the network then because the consensus block will not be valid in the client. This, I believe, is a better option than having developers intervene and having to release a new client in the case of a >50% hash power take over. Even then, someone with >50% of the hash will always be able to reverse transactions to some degree in bitcoin. Once a transaction is level 2 confirmed in Encoin, it is not reversible. This could take as little as 20 seconds. And no hashing power is necessary.

The basic TL;DR description is that EnCoin is designed as a pure digital currency with coins (ENC) that cost a fixed amount of electricity to create. The system-wide goal is to hold this amount of electricity fixed over time. Even though, electrical costs will change and technologies will change. Hence, the slightly inside joke "1 ENC costs about 1 ENC to produce". This means if it cost your grandfather 10 kwh to create 1 ENC, it will cost you 10 kwh to create 1 ENC. This ratio remains constant regardless of adoption rate or demand for ENC.

The better way to visualize the goal, is that if 1 ENC bought 1 loaf of bread for your grandfather, it will likely 1 loaf of bread for you even in the face of fiat currency inflation.
(pre-inflation) 1 ENC generated at $0.10 kwh = $1 loaf of bread
(post-inflation)1 ENC generated at $1.00 kwh = $10 loaf of bread

The simplest way to understand the implementation concept is to think of bitcoin but with a monotonically increasing difficulty level. Each day the hash difficulty is raised a little to compensate for Koomey's law. Over time that difficulty will double about every 18 months. That way future increases in processor efficiency are offset by increased computational difficulty.

The most important difference from bitcoin is that this creates a stable valued coin that is created in differing quantities base upon demand. So if a million new folks want to trade $20 each worth of eggs, butter and cheese each month, they can buy or generate as many coins as they need to make it happen. There is never a hoarding issue because ENC values spikes are designed to be momentary.