[StableCoin] Welcome and Introduce Yourself...

[Etlase2]

Right now, people say that a lot of their coins are in cold storage, but we have no way of verifying the same. They can appear anytime in the nearest exchange and drive down the price to whatever the market will bear at that time. So stablecoin needs to have a way of clearly placing coins in cold storage in plain sight, with clear indicators that during the next few hours, days or so, more than X number of coins or Y percentage of coins, simply cannot be traded. Thus, not only is the money supply (M1, is it?) in plain sight, but money placed in short term and long term locked coins (bonds?) is also in plain sight.

To what real end? Something I've learned over going through the process of doing the whole stable currency idea is that you can't brute force the price to some acceptable spot, you've got to let it run free a little. Your view here is colored by bitcoin's heavy one-sidedness that comes along with its currency distribution scheme, and I think your fears would have very little place in a currency where the supply of new money was not so heavily restricted.

If hoarders hold back in an attempt to increase the price, minters create new currency for a profit and bring the price back in line. Hoarders are no better or worse off, though the value in the economy has arguably increased (hoarders chose not to revert back to fiat). Any rash of selling below the cost to produce coins is a lot less likely, because everyone has an idea what currency can be created for. It's a braking mechanism. Bitcoin's cost to produce tends to move with the price of bitcoins; a stable currency's cost to produce should try remain as stable as possible. So while bitcoin can lose or gain hundreds of % in a day and market movers can take advantage of fear, anyone manipulating a stable base cannot take advantage of that fear.

PS - An interesting side effect of the Decrits proposal, where people purchase 1 year shares for a meaningful amount of currency and receive a portion of transaction fees as payment for helping to secure the network, during periods where there is mild inflation due to whatever reason, buying shares for a small but guaranteed return will temporarily lock money. Periods where price inflation is experienced are even more beneficial times to purchase shares because the tx fee is a percentage of each transaction. Thus, inflation = more fees. It encourages temporarily removing money from the economy, and this locked money is an easily verifiable thing and part of normal network activity.

[Red]

Really interesting ideas blogospheroid! The group is really getting larger!

Just for your reference:
Impaler is working on Freicoin. It is demurrage based. He is also kicking around ideas on locking coins.
Glad he mentioned it. I'd forgotten Etlase2 was fleshing those ideas out as well
I'm glad you mentioned incentives. That's one of my favorite topics! Nobody ever wants to discuss "manipulating" the users.

[blogospheroid]

To what real end? Something I've learned over going through the process of doing the whole stable currency idea is that you can't brute force the price to some acceptable spot, you've got to let it run free a little. Your view here is colored by bitcoin's heavy one-sidedness that comes along with its currency distribution scheme, and I think your fears would have very little place in a currency where the supply of new money was not so heavily restricted.

I cannot deny that bitcoin's structure influenced me on this. And I know that the price will run free, even if only a certain percentage of the coins are transactable at any given moment. The question is, does it add a little more stability or not?

If hoarders hold back in an attempt to increase the price, minters create new currency for a profit and bring the price back in line. Hoarders are no better or worse off, though the value in the economy has arguably increased (hoarders chose not to revert back to fiat). Any rash of selling below the cost to produce coins is a lot less likely, because everyone has an idea what currency can be created for. It's a braking mechanism. Bitcoin's cost to produce tends to move with the price of bitcoins; a stable currency's cost to produce should try remain as stable as possible. So while bitcoin can lose or gain hundreds of % in a day and market movers can take advantage of fear, anyone manipulating a stable base cannot take advantage of that fear.

I'm sorry if you've answered this one before, but what prevents this from becoming a mordor coin, burning up resources indiscriminately?

PS - An interesting side effect of the Decrits proposal, where people purchase 1 year shares for a meaningful amount of currency and receive a portion of transaction fees as payment for helping to secure the network, during periods where there is mild inflation due to whatever reason, buying shares for a small but guaranteed return will temporarily lock money. Periods where price inflation is experienced are even more beneficial times to purchase shares because the tx fee is a percentage of each transaction. Thus, inflation = more fees. It encourages temporarily removing money from the economy, and this locked money is an easily verifiable thing and part of normal network activity.

Decrits sounds like a very interesting proposal. I need to read it in more detail.

[Etlase2]

I cannot deny that bitcoin's structure influenced me on this. And I know that the price will run free, even if only a certain percentage of the coins are transactable at any given moment. The question is, does it add a little more stability or not?

Whether it does or not, you have not come up with a way to incentivize it, and thus no one (rational) will do it.

I'm sorry if you've answered this one before, but what prevents this from becoming a mordor coin, burning up resources indiscriminately?

It's all good, I prefer discussing this stuff in threads rather than the numerous questions I get over PMs because then I don't have to repeat myself!

In the other StableCoin thread - https://bitcointalk.org/index.php?topic=178140.msg1890016#msg1890016 - Although I did paste it in this thread as well, there is more context there

Relevant part is at the end:

"In addition to all of this, we must eschew mordorcoin and give money away for free. Now we know for damn sure that new money is needed with the difficulties associated with minting. The written decrits proposal proposes giving away 5x of this block of coins to transactions and 5x to existing accounts, both randomly (blocks of accounts will be stored together, and this will be the starting point for awarding free money to them because awarding every account is infeasible in a network of any reasonable size). This area can be nitpicked, but I have come up with some very solid solutions.

In addition to that, because network expansion could easily outpace the time constraints set on Mint Blocks, each successive Mint Block within a defined period will increase the tx/account award further. The second block in a row will award 6x to each. The third will award 7x. This may max out at 10 each, or maybe it could go on continuously (perhaps after 20x the difficulty could start increasing say 3% per block; with so much new money in circulation [especially to transactors], if this causes mild deflation it would likely not have any problems being counteracted). So even in the worst case scenario of everyone instantly switching to ASICs and intentionally not raising the difficulty, designing and producing those ASICs will never be profitable. Plus the inflation has to catch up and raise the average amount of transactions over an extended period, or the ASICs will simply run into a wall where they can not mint for extended periods."

Decrits sounds like a very interesting proposal. I need to read it in more detail.

Now I will be repeating myself, but that proposal is over 9 months old and it isn't fair to go back and change it all to fit my current ideas because the discussion will look weird. But it covers the major points well enough. I have 40-50 written pages of tweaks and ways to make it work better as well as multiple ideas on how each facet could work. Things that I would like to eventually flesh out publicly to reduce the chance that I missed something. I think one more proposal will be in the works and then it's time to finally start writing code, and hope some more will join up.

[blogospheroid]

The question is, does it add a little more stability or not?

Whether it does or not, you have not come up with a way to incentivize it, and thus no one (rational) will do it.

I agree. The incentivizing aspects need to be discussed.

My first thought is that similar to what you suggest for distributing new coins to transactions and accounts, it can be modified to transactions and locked coins, maybe weighted by how long they keep it in freeze.

[Etlase2]

But if you are trying to lock coins during inflation, there probably shouldn't be free coins being given out. I suppose if you could determine, somehow, that there are too many coins and then later reward those who locked money. But this seems convoluted to me. Like I said, stop worrying about a little bit of inflation. If there is more than a little bit, there is something wrong and locking coins isn't going to fix it.

[wingding]

A coin with predetermined difficulty

Imagine a coin that is exchanged at $100, and mining it costs somewhat less. Different from bitcoin, the difficulty is not adjusted according to block production. Hence, if an increased market demand takes place, the price will go up, but so will also mining, which in turn increase the supply and work against the price increase. In the case of a price decline, so will the profit from mining, and fewer coins will enter the market.

The mining difficulty could be predetermined to double each 18 months, following Moores law. I believe such a coin would show more price stability, and a value strongly correlated to energy costs. It would certainly have a different price dynamics than bitcoins.

[Etlase2]

Imagine a coin that is exchanged at $100, and mining it costs somewhat less. Different from bitcoin, the difficulty is not adjusted according to block production. Hence, if an increased market demand takes place, the price will go up, but so will also mining, which in turn increase the supply and work against the price increase. In the case of a price decline, so will the profit from mining, and fewer coins will enter the market.

If only there were a link in my sig that goes into massive detail about such a system.

[Red]

@wingding, we've "imagined" that coin a number of times. See my GEM proposal linked in my introduction post. Etlase2's proposals are based on that concept as well.

@Etlase2, been looking at Ripple's underlying mechanics. They've implemented a lot of the stuff we've wanted to do.
1. Keeping a balance sheet instead of separate inputs and outputs. (Encoin style)
2. Distributed consensus building. For this they used the personal trust rules I described somewhere.
3. Instant irreversible commits based on that consensus. (Encoin)
4. No mining rewards. Peers validate the ledger because they have a long term self-interest in doing so. (My assertion)
5. No cpu power wasted on POW at all. (My assertion)

Edit: WRONG! Ignore this presumption
6. (unconfirmed presumption) Every Ripple peer is a stakeholder with a vested interested in stable valued XRP. They will use a centralized LETS-like process to "fairly distribute" some initial  XRP among early adopters. With the goal being to drive the initial XRP exchange price into a pre-planned trading price relative to some Fiat benchmark. After that Ripple will release XRP from their huge reserve in order to meet demand. This will be OK with a consensus of users because stability is a quality of every user's social contract with Ripple.

Neither of us was a proponent of the number 6 implementation. However, we were both in favor of the stability social contract idea.

[Red]

I've come up with an idea for the simplest (for me to create) InertiaCoin I can think of. I'm going to write about it in a separate thread. It is based on the "complementary currency" idea (a la NameCoin). It will use mixed-mining so there is zero extra PoW to be done.

It will use BitCoin's transaction fee to reward miners and discourage chain bloat. BUT, there is no preset mining reward. Over time mining rewards will vary with monetary policy to create price stability.

Monetary policy will be set in a completely distributed fashion. It will use a proof-of-stake "Vested Interest Voting" system. In other words, users who don't vote are presumed to be OK with the status quo. Those who want a change in the current coin creation/destruction policy, Vote. Voting can be done at any block change. Creating/destroying coins will be done similarly to my GEM proposal.

Any initial thoughts?

By the way, this is the Cheap and Sleazy version of my plan. I have a much more stable version. And a PeggedVersion. But before we get lost in details, I want to see if there is any interest at all. I think if I don't act on the idea immediately, Ripple will run the table on Coins.

[wingding]

I think that it is extremely difficult to have a new coin succeeding without strong power or support behind it. Even though it may provide more stability. When the fresh markets realize that that most bitcoins were premade before supply rate was halved, the hoarding/scarcity problem becomes evident, and new money will seek their way into alternatives like Litecoin and Ripple. I still think still the fork/twinchain/branch concept could achieve something, because it could be a healthy mutation of bitcoin, and not something entirely new.


 

[blogospheroid]

Another thought.

From Decrits, I got the idea of using transaction fees as a proxy for GDP. In a corner of the blogosphere, targetting a level path for the nominal value of GDP is being talked about as the correct way to target the money supply of an economy. i.e if by a certain time, nominal gdp is not as high as the level calculated by the growth path, then print money. If it is more, then withdraw money from the market by selling assets.

I, personally am in favour of such a policy for statist monetary policy. NGDP does look better than inflation targeting or money supply targeting.

So, a policy appropriate to crypto currencies could be to target a steady growth in nominal transaction fees. Have an initial booting up period where anyone who can bring in a certain amount of hashing power, brings it in and gets coins. It is a pure mordor coin during this initial period. After that (2-3 years, maybe, anybody have any ideas on how to determine when a crypto currency has stabilised?), the targetting mechanism takes over.

If the transaction fees paid in a certain interval of time (maybe a week or fortnight) exceed the level target, then automatically lock up a percentage of the money in every account. Coins that can be used now, lets say block height 99, get locked so that they can't be used till block 149. Ownership is maintained, but usage is constrained
 
If the transaction fees paid are below a level target, then automatic unlocking of locked coins occurs, Coins locked till block height 140 get changed so that they can be used in block 90 and so on.

The level target for transaction fees could be a linearly increasing one, exponential or sigmoid.

[Etlase2]

So, a policy appropriate to crypto currencies could be to target a steady growth in nominal transaction fees. Have an initial booting up period where anyone who can bring in a certain amount of hashing power, brings it in and gets coins. It is a pure mordor coin during this initial period. After that (2-3 years, maybe, anybody have any ideas on how to determine when a crypto currency has stabilised?), the targetting mechanism takes over.

The initial bootstrap is tough. I was thinking 3 years. However, instead of pure mordor coin, the "early adopters" would get multiples of the coin award. Perhaps 10x for the first 4 months, then 9x, and so on until the last 4 months at 2x. This would hopefully be enough time to see a reasonable transaction volume and distribution of money among many people (while yes, giving it a slight pyramid distribution to encourage adoption that quickly flattens out). Then flip the switch and people no longer need to mint to get free money. Might be a big deal like the first bitcoin halve, and merchants would be encouraged to start getting in the game.

If the transaction fees paid in a certain interval of time (maybe a week or fortnight) exceed the level target, then automatically lock up a percentage of the money in every account. Coins that can be used now, lets say block height 99, get locked so that they can't be used till block 149. Ownership is maintained, but usage is constrained

I'm sorry but I think this is a terrible idea. You propose the exact opposite of what you should when the economy is expanding--you are strangling it instead of encouraging it. All to prevent price inflation temporarily if something other than expansion is happening (and cause deflation if it is expansion). Assuming the currency creation base is stable, the value of the money should continue to oscillate around that point. Hell, even without the idea of price stickiness, merchants in a system like Decrits would not need to constantly reprice because of DCR->fiat fluctuations if they could be fairly sure of the stable currency creation base. This will take time and market penetration though.

Messing with account balances or locking coins is going to leave a terrible taste in everyone's mouth, and I don't think it accomplishes anything worthwhile.

[blogospheroid]

I'm sorry but I think this is a terrible idea. You propose the exact opposite of what you should when the economy is expanding--you are strangling it instead of encouraging it. All to prevent price inflation temporarily if something other than expansion is happening (and cause deflation if it is expansion). Assuming the currency creation base is stable, the value of the money should continue to oscillate around that point. Hell, even without the idea of price stickiness, merchants in a system like Decrits would not need to constantly reprice because of DCR->fiat fluctuations if they could be fairly sure of the stable currency creation base. This will take time and market penetration though.

Messing with account balances or locking coins is going to leave a terrible taste in everyone's mouth, and I don't think it accomplishes anything worthwhile.

The thing that is being curbed is nominal transaction fees. Without a trusted Oracle, we have to rely on within-network indicators. Nominal transaction fees is a within-network indicator, that is relatively secure.

Any indicator that affects further results will be prone to abuse. And sorry, I believe the direction of the feedback you describe is not correct. If there are more fees being spent and the system encourages it with printing more money, then even more fees will be spent and even more money will be printed until the coin collapses due to hyperinflation. There needs to be a written down growth path for some indicator. For bitcoin, it is the monetary base itself. I'm discussing the possibility of using the nominal transaction fees as an indicator. One can try to model a good growth path that is high in the beginning and lower later, that is beside the idea of targeting a level of nominal transaction fees.

Hope I've made things a little more clear.

The way I look at it, there is a component of proof-of-work/mordor coin that has to be incorporated into every coin. The question seems to be how to minimise it and still maintain stability.

[Etlase2]

The thing that is being curbed is nominal transaction fees.

By reducing the velocity of money--a drastic measure. Are you familiar with the MV=PQ equation? Either P (price level) or Q (goods and services available to the economy) could be increasing that results in an increased V (velocity of money) assuming that the M (money supply) is unchanged. You are worried only about P and do not see that Q has an identical effect from the standpoint of transaction fees. Designing a system with some linear idea of how the economy is likely to expand is destined to be wrong. It is not a predictable process by any measure. Adjusting V will not make P stable.

Without a trusted Oracle, we have to rely on within-network indicators. Nominal transaction fees is a within-network indicator, that is relatively secure.

It may be secure, but for any argument you can make that it is useful I can come up with a counter-argument.

And sorry, I believe the direction of the feedback you describe is not correct.

Then what do you think will happen when people are cut off from trying to spend their money?

If there are more fees being spent and the system encourages it with printing more money, then even more fees will be spent and even more money will be printed until the coin collapses due to hyperinflation.

See the thing is, more fees being paid do not necessarily encourage the system to print more money, at least if we're talking about Decrits. Fees, on their own, have absolutely nothing to do with it. Only the willingness for a large group of people to invest time, hardware, and energy into minting new currency. For that to happen, Decrits must be worth more than their cost to produce. Even if a massively irrational actor decides to inflate the currency, all he has done has turned his effort into coins for other people. He cannot continue this behavior forever.

There needs to be a written down growth path for some indicator.

If you want instability, sure.

For bitcoin, it is the monetary base itself.

Case-in-point.

I'm discussing the possibility of using the nominal transaction fees as an indicator.

Even spam protection throws this off. I would imagine you'd need a minimum transaction fee, Decrits proposes 0.01, so that microtransactions do not tax the network. If microtransactions are popular, this would throw off your indicator by several times as it would see much higher fees than actual activity. "Too many microtransactions this week--LOCK ALL THE COINS!" You could argue to use actual amounts instead, but then you run into the significant problem of off-chain transactions/clearinghouses being left out of your indicator (which is also the case with tx fees).

The way I look at it, there is a component of proof-of-work/mordor coin that has to be incorporated into every coin.

It sounds like you think that Decrits doesn't have a proof-of-work component. It does. Maybe you're generalizing. But the whole hyperinflation thing seemed to be directed at it.

The question seems to be how to minimise it and still maintain stability.

I haven't seen you propose any way to minimize it, only how to hamfist a stable price based on incomplete information. What is your plan for this part of the equation? I've twice quoted how Decrits addresses this in this thread. I understand you're still developing your idea and I apologize for being harsh, but all of the concerns you state about what I've proposed are not really concerns and are things I have thought about and designed with in mind. You're using the same, weak arguments that everyone else does after getting an idea of what it is that is incorrect and not even bothering to quote the salient response points. I'd much prefer to argue my actual ideas rather than ones that look poor compared to yours; otherwise we go around in unproductive circles.

[Red]

Hey, let's take the arguing out of the "Welcome..." thread!

Make another thread that starts with [StableCoin] so everyone can find it.

I'm still coming to terms with Ripple. I'm going to make a new thread about that too.

[blogospheroid]

Hi Etlase2,

Do you already have another thread where this discussion can be taken, or should I start a new one?

[Impaler]

@wingding, we've "imagined" that coin a number of times. See my GEM proposal linked in my introduction post. Etlase2's proposals are based on that concept as well.

@Etlase2, been looking at Ripple's underlying mechanics. They've implemented a lot of the stuff we've wanted to do.
1. Keeping a balance sheet instead of separate inputs and outputs. (Encoin style)
2. Distributed consensus building. For this they used the personal trust rules I described somewhere.
3. Instant irreversible commits based on that consensus. (Encoin)
4. No mining rewards. Peers validate the ledger because they have a long term self-interest in doing so. (My assertion)
5. No cpu power wasted on POW at all. (My assertion)

6. (unconfirmed presumption) Every Ripple peer is a stakeholder with a vested interested in stable valued XRP. They will use a centralized LETS-like process to "fairly distribute" some initial  XRP among early adopters. With the goal being to drive the initial XRP exchange price into a pre-planned trading price relative to some Fiat benchmark. After that Ripple will release XRP from their huge reserve in order to meet demand. This will be OK with a consensus of users because stability is a quality of every user's social contract with Ripple.

Neither of us was a proponent of the number 6 implementation. However, we were both in favor of the stability social contract idea.

You make an interesting point with regard to removing PoW for simple zero-sum transactions.  I've been thinking about this lately and really the only reason BitCoin uses PoW is to control MINING.  As soon as you separate transactions and mining the need for ANY proof just disappears.  The reason is that all transactions are sent into the network as a cryptographically signed piece of data, they can not be faked by some 3rd party.  The only thing a malicious minor can do in BTC is to BLOCK transactions either selectively or in totality.  This should be seen as a design flaw as the network should be perfectly capable of detecting and rejecting this behavior but the PoW president actually gives this potentially devastating power to the Miner.

The network propagates the transaction message to all nodes and block creation is only necessary to 'cement' the transactions as permanent and to give some kind of time-step to the network which is long enough to tell us that a double spend has not occurred.  Even if we felt this block process was necessary theirs no need for Proof-of-Work, just get the nodes to periodically make a block and propagate that around until all nodes agree that is the FULLEST block of transactions within a time span.  Now you might have an issue with clients trying to lie about their time-stamp or some kind of lag induced time-stamp disagreement.  To solve this just require each transaction to incorporate the hash of the prior block, this puts an indelible block-height into the transaction and guarantees that no one can make time-traveling transactions.  Once a block has passed outdated transactions are considered dead and can never be included.  But transactions that fail to get into a block should now be unheard of because the network actually prioritizes their full inclusion.

Now clearly a Stable coin unlike Ripple needs a mining process (well a money creation process), so separating mining and transacting is just a first step to getting at a solution because it gives us lots more flexibility.  You can now slow down or turn off mining without making the whole system collapse.

[Etlase2]

Hi Etlase2,

Do you already have another thread where this discussion can be taken, or should I start a new one?

depends on what you want to discuss. if it is decrits or comparing your ideas to decrits, see sig

otherwise maybe this one: https://bitcointalk.org/index.php?topic=178140.0

[Red]

This should be seen as a design flaw as the network should be perfectly capable of detecting and rejecting this behavior but the PoW president actually gives this potentially devastating power to the Miner.

Exactly! Etlase2 and I have been saying the same thing for years!

The network propagates the transaction message to all nodes and block creation is only necessary to 'cement' the transactions as permanent and to give some kind of time-step to the network
...
just get the nodes to periodically make a block and propagate that around until all nodes agree that is the FULLEST block of transactions within a time span.  

Now you might have an issue with clients trying to lie about their time-stamp or some kind of lag induced time-stamp disagreement.  

Yup, Ripple has an equally brilliant solution to synchronizing time. Set the default block creation time to ZERO!

Seriously, as soon as there are any new transactions. ANY node can declare a block/ledger created. Then they start the consensus cycle. Once consensus is reached if their are any new transactions someone declares a block and they start again! Brilliant! The also included a (per address) transaction sequence number DUH! That way its trivial to tell if a circulating transaction is old or a double spend attempt.

To solve this just require each transaction to incorporate the hash of the prior block, this puts an indelible block-height into the transaction and guarantees that no one can make time-traveling transactions.  Once a block has passed outdated transactions are considered dead and can never be included.

I discussed this very point with satoshi himself as it related to bitcoin. I thought keeping the old transactions at all was a curious decision for an "anonymous" p2p system. We kicked around several ideas, but none really played out. You can all agree to "forget" but you have to presume someone is lying and saving old transactions anyway.

Now clearly a Stable coin unlike Ripple needs a mining process (well a money creation process), so separating mining and transacting is just a first step to getting at a solution because it gives us lots more flexibility.  You can now slow down or turn off mining without making the whole system collapse.

Yup! Now you are totally in sync with Etlase2 and I.