Смотреть поиск (2014) в Full HD качестве ОНЛАЙН
Posted by Vitalik Buterin on November 11th, 2014. Special thanks to Robert Sams for the development of Seignorage Shares and insights regarding how to correctly value volatile coins in multi-currency systems Note: Results of this research will likely go into either subcurrencies or independent blockchains One of the main problems with Bitcoin for ordinary users is that, while the network may be a great way of sending payments, with lower transaction costs, much more expansive global reach, and a very high level of censorship resistance, Bitcoin the currency is a very volatile means of storing value.
Seeing this concern, there is a growing interest in a simple question: Can we have the full decentralization that a cryptographic payment network offers, but at the same time have a higher level of price stability, without such extreme upward and downward swings? However, the problem of making a price-stable cryptocurrency, as the researchers realized, is much different from that of simply setting up an inflation target for a central bank. The underlying question is more difficult: To resolve the issue properly, it is best to break it down into two mostly separate sub-problems: Given a desired supply adjustment to target the price, to whom do we issue and how do we absorb currency units?
Decentralized Measurement For the decentralized measurement problem, there are two known major classes of solutions: As far as exogenous solutions go, so far the only reliable known class of mechanisms for possibly cryptoeconomically securely determining the value of an exogenous variable are the different variants of Schellingcoin — essentially, have everyone vote on what the result is using some set chosen randomly based on mining power or stake in some currency to prevent sybil attacks , and reward everyone that provides a result that is close to the majority consensus.
If you assume that everyone else will provide accurate information, then it is in your interest to provide accurate information in order to be closer to the consensus — a self-reinforcing mechanism much like cryptocurrency consensus itself.
Particularly, what if some medium-sized actor pre-announces some alternative value to the truth that would be beneficial for most actors to adopt, and the actors manage to coordinate on switching over? If there was a large incentive, and if the pool of users was relatively centralized, it might not be too difficult to coordinate on switching over. There are three major factors that can influence the extent of this vulnerability: Is it likely that the participants in a schellingcoin actually have a common incentive to bias the result in some direction?
Do the participants have some common stake in the system that would be devalued if the system were to be dishonest? Now, if there are two kinds of mining, one of which is used to select Schellingcoin participants and the other to receive a variable reward, then this objection no longer applies, and multi-currency systems can also get around the problem. However, we should not simply count on this incentive to outweigh 1. The 12-second block time would mean that there is almost no time for coordination.
The creator of the block can be strongly incentivized or even, if the Schellingcoin is an independent blockchain, required to include all participations, to discourage or prevent the block maker from picking and choosing answers.
A fourth class of options involves some secret sharing or secure multiparty computation mechanism, using a collection of nodes, themselves selected by stake perhaps even the participants themselves , as a sort of decentralized substitute for a centralized server solution, with all the privacy that such an approach entails. The incentive to vote correctly is that only tests that remain in the main chain after some number of blocks are rewarded, and future voters will note attach their vote to a vote that is incorrect fearing that if they do voters after them will reject their vote.
If Schellingcoin proves unworkable, then we will have to make do with the other kinds of strategies: Examples of such services include: Computation measured via mining difficulty Transaction fees Data storage Bandwidth provision A slightly different, but related, strategy, is to measure some statistic that correllates indirectly with price, usually a metric of the level of usage; one example of this is transaction volume. The problem with all of these services is, however, that none of them are very robust against rapid changes due to technological innovation.
Hence, trying to peg a currency to any of those variables will likely lead to a system which is hyperinflationary, and so we need some more advanced strategies for using these variables to determine a more stable metric of the price.
First, let us set up the problem. Formally, we define an estimator to be a function which receives a data feed of some input variable eg. Unfortunately, the problem with this approach is obvious from the graph and was already mentioned above: Note that there are an infinite number of versions of this estimator, one for each depreciation rate, and all of the other estimators that we show here will also have parameters. The optimal value for the compensated estimator is a drop of 0. The next estimator that we will explore is the bounded estimator.
The way the bounded estimator works is somewhat more complicated. Any growth outside these bounds it assumes is coming from price rises or drops. The theory is that cryptocurrency price growth to a large extent happens in rapid bubbles, and thus the bounded estimator should be able to capture the bulk of the price growth during such events.
There are more advanced strategies as well; the best strategies should take into account the fact that ASIC farms take time to set up, and also follow a hysteresis effect: A simple approach is looking at the rate of increase of the difficulty, and not just the difficulty itself, or even using a linear regression analysis to project difficulty 90 days into the future.
Here is a chart containing the above estimators, plus a few others, compared to the actual price: Note that the chart also includes three estimators that use statistics other than Bitcoin mining: We can also split up the mining-based estimators from the other estimators: Of course, this is only the beginning of endogenous price estimator theory; a more thorough analysis involving dozens of cryptocurrencies will likely go much further.
The best estimators may well end up using a combination of different measures; seeing how the difficulty-based estimators overshot the price in 2014 and the transaction-based estimators undershot the price, the two combined could end up being substantially more accurate. Something like Bitcoin, if it becomes mainstream, will likely be somewhat more unstable than gold , but not that much more unstable — the only difference between BTC and gold is that the supply of gold can actually increase as the price goes higher since more can be mined if miners are willing to pay higher costs, so there is an implicit dampening effect, but the supply elasticity of gold is surprisingly not that high ; production barely increased at all during the run-ups in price during the 1970s and 2000s.
The price of gold stayed within a range of 4. The other issue that all of these estimators have to contend with is exploitability: Mining difficulty, however, is much more difficult to exploit simply because the market is so large.
If a platform does not want to accept the inefficiencies of wasteful proof of work, an alternative is to build in a market for other resources, such as storage, instead; Filecoin and Permacoin are two efforts that attempt to use a decentralized file storage market as a consensus mechanism, and the same market could easily be dual-purposed to serve as an estimator.
The simplest approach is to simply issue them as a mining reward, as proposed by the Japanese researchers. However, this has two problems: Such a mechanism can only issue new currency units when the price is too high; it cannot absorb currency units when the price is too low. If we are using mining difficulty in an endogenous estimator, then the estimator needs to take into account the fact that some of the increases in mining difficulty will be a result of an increased issuance rate triggered by the estimator itself.
If not handled very carefully, the second problem has the potential to create some rather dangerous feedback loops in either direction; however, if we use a different market as an estimator and as an issuance model then this will not be a problem.
The first problem seems serious; in fact, one can interpret it as saying that any currency using this model will always be strictly worse than Bitcoin, because Bitcoin will eventually have an issuance rate of zero and a currency using this mechanism will have an issuance rate always above zero. Hence, the currency will always be more inflationary, and thus less attractive to hold. However, this argument is not quite true; the reason is that when a user purchases units of the stabilized currency then they have more confidence that at the time of purchase the units are not already overvalued and therefore will soon decline.
Alternatively, one can note that extremely large swings in price are justified by changing estimations of the probability the currency will become thousands of times more expensive; clipping off this possibility will reduce the upward and downward extent of these swings. For users who care about stability, this risk reduction may well outweigh the increased general long-term supply inflation. This approach can be described as follows: Vol-coins are an actual currency; users can have a zero or positive balance of them.
Stable-coins exist only in the form of contracts-for-difference ie. Because stable-coins are a zero-total-supply currency ie. However, the mechanism has some rather serious fragility properties. At the end, the stable-coin could easily end up being worth nothing at all. Note that BitShares has now moved to a somewhat different model involving price feeds provided by the delegates participants in the consensus algorithm of the system; hence the fragility risks are likely substantially lower now.
SchellingDollar An approach vaguely similar to BitAssets that arguably works much better is the SchellingDollar called that way because it was originally intended to work with the SchellingCoin price detection mechanism, but it can also be used with endogenous estimators , defined as follows: Vol-coins are initially distributed somehow eg.
Users may have only a zero or positive balance of vol-coins. Users may have a negative balance of stable-coins, but can only acquire or increase their negative balance of stable-coins if they have a quantity of vol-coins equal in value to twice their new stable-coin balance eg.
This prevents situations where accounts exist with negative-valued balances and the system goes bankrupt as users run away from their debt. This mechanism is of course subject to the limits described in 2.
The system keeps track of the total quantity of stable-coins in circulation. If the quantity exceeds zero, the system imposes a negative interest rate to make positive stable-coin holdings less attractive and negative holdings more attractive. If the quantity is less than zero, the system similarly imposes a positive interest rate. Note that there are still fragility risks here. Second, if the vol-coin market is too thin, then it will be easily manipulable, allowing attackers to trigger margin call cascades.
Another concern is, why would vol-coins be valuable? Scarcity alone will not provide much value, since vol-coins are inferior to stable-coins for transactional purposes. Now, consider a strategy where a user tries to hold on to a constant percentage of all vol-coins.
Seignorage shares is a rather elegant scheme that, in my own simplified take on the scheme, works as follows: However, the simplicity comes at the cost of some degree of fragility. To see why, let us make a similar valuation analysis for vol-coins. The profit and loss scenarios are simple: Thus, here lies the problem: In exchange for this fragility risk, however, vol-coins can achieve a much higher valuation, so the scheme is much more attractive to cryptoplatform developers looking to earn revenue via a token sale.
Note that both the SchellingDollar and seignorage shares, if they are on an independent network, also need to take into account transaction fees and consensus costs. Fortunately, with proof of stake, it should be possible to make consensus cheaper than transaction fees, in which case the difference can be added to profits.
Ultimately, however, some degree of fragility is inevitable: Even sidechains, as a scheme for preserving one currency across multiple networks, are susceptible to this problem. The question is simply 1 how do we minimize the risks, and 2 given that risks exist, how do we present the system to users so that they do not become overly dependent on something that could break?
Conclusions Are stable-value assets necessary? There would then be multiple separate classes of cryptoassets: If that were to happen, and particularly if the stronger version of price stability based on Schellingcoin strategies could take off, the cryptocurrency landscape may end up in an interesting situation: The true cryptoeconomy of the future may have not even begun to take shape.