bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 3쪽 중 2쪽
o-knowledge proof using a variety of cryptographic means, including [10, 3, 5].
2.2.3 Personalized Pre-T rusted Peers
Second, in order to break malicious cliques, and to ensure convergence of the power method and
uniqueness of the principal eigenvector, EigenTrust introduces the notion of pre-trusted peers, a group
of peers that are active and assumed to be universally trusted. This ensures that the graph is acyclic
and strongly connected (and that the matrix is irreducible and that the problem is well-conditioned).
However, it requires the system to define a set of universally trusted peers, and concentrates outsized
power to confer reputation in those pre-trusted peers.
We can address this through personalization. Rather than computing a single global reputation
vector, the system can compute a personalized global reputation vector for each peer, that gives the
reputation score of each peerj in the network from the point of view of a single peeri. To compute
personalized EigenTrust for peeri, one can simply perform a traditional EigenTrust computation, but
use the contact list of peeri as the set of pre-trusted peers.
This is far more computationally expensive than a single EigenTrust computation; however, we
apply many of the computation-saving techniques that enabled personalized PageRank [13] to a per-
sonalized EigenTrust computation.
2.2.4 Practical Implications
For the social payments case, in which people text money to friends, the address-based encryption
scheme suffices as a lightweight identity proxy, allowing people to send money directly to people’s cell
phone numbers, even if they have not signed up for a wallet.
As people are interested in using the protocol to pay people outside of their direct circle of contacts,
it is useful for a user to be able to aggregate the trust signals of those in their network to make purchase,
payment, and credit decisions, and to mitigate bad actors.
Further, a reputation scheme as we described enables a more robust identity scheme. Most identity
schemes are based on attestations from others, and it would be useful to be able to weight those
attestations by the reputation score of the attestor.
3 Stabilizing Value
Perhaps the biggest hurdle to the use of cryptocurrencies as a means of payment is their volatility.
Consumers are unlikely to want to buy a volatile cryptocurrency to spend it, since the purchasing
power of their accounts would fluctuate widely with market demand for the currency. Merchants who
accept cryptocurrencies are likely to convert to fiat upon payment, because their business model does
not involve speculating on cryptocurrencies. And the most successful cryptocurrencies today are not
just volatile but deflationary – their success leads to their price rising; as a result, prices denominated
7
in the currency fall. Rational behavior would be to use such currencies as a store of value rather than
a medium of exchange, and in practice that is what has happened.
Stable-value cryptocurrencies would bring a number of benefits to the cryptocurrency ecosystem.
Forone, stablepricesremoveaconsiderablebarrierforusingcryptocurrenciesasamedium-of-exchange;
salaries, prices of goods, fixed obligations, can all be set in a stable value cryptocurrency without
requiring either party to speculate on the future value of the currency. Further, financial contracts are
more easily built with a stable value coin, because the issuer can separate the function of the contract
from the price risk of the currency in which it’s denominated.
While a single stable-value currency would be helpful, a thriving cryptoeconomy is best-served by
a family of stable-value currencies, much as it is well-served by the family of variable-value crypto-
assets that we have today. Certainly a cryptocurrency pegged to the US Dollar has several uses, from
social payments in the US, to user-initiated dollarization in hyper-inflationary markets, to the efficient
settlement of high-frequency crypto-asset trades. At the same time, a cryptocurrency pegged to the
Euro would also be useful for many purposes, as would a cryptocurrency pegged to the price of a
basket of goods in Greece, as would a cryptocurrency pegged to the price of a barrel of oil, or housing
in San Francisco. Stable-value local, regional, and utility currencies allow people to hedge price risk in
their lives by denominating a portion of their personal economy in currencies that are stable vis-a-vis
the price of the goods they regularly use.
3.1 Elastic Coin Supply and Shifting Volatility Risk
Several protocols have been proposed for a decentralized stabilized value cryptocurrency (for exam-
ple [17, 2, 1, 19]). While a full review of these proposals is outside of the scope of this paper, they
generally share two properties. First, rather than a deterministic coin supply rule (in which the coin
supply and growth rate are determined in advance, independent of exogenous information), they each
introduce an elastic coin supply rule, that stabilizes the value of the coin by adjusting the supply of
the coin to match the demand. Second, they each introduce a multi-asset ecology, in which one coin
is intended to be stable, while one or more complementary crypto-assets bear the risk of a decrease in
stablecoin demand (and receives a reward in the case of an increase in stablecoin demand). In essence,
they each shift volatility risk from the coin holders to the complementary asset holders.
The Celo protocol utilizes the same two key intuitions, with five novel features: (a) it introduces
a multi-asset tiered reserve that supports several local and regional stable value currencies, (b) it sets
expansion and contraction parameters that are tuned to the reserve ratio defined by the tiered reserve,
(c) it introduces a decentralized exchange in which the different local and regional currencies and the
reserve currency can be traded amongst one another without a central party, and that the protocol
can use to perform expansions and contractions, (d) it releases block rewards and other incentives in
the reserve currency, and (e) it has a governance mechanism in which long-term stakeholders in the
reserve currency are responsible for governing the assets held in reserve and the new local currencies
that are introduced.
3.2 Protocol Summary
At a high level, the protocol proceeds as follows:
1. The protocol establishes a fixed supply of assets, called Celo (also referred to as the Celo native
asset), a portion of which is distributed over time. From the initial asset distribution, a portion
is placed in reserve and diversified.
2. The protocol also establishes a means-of-payment currency, called the Celo Dollar, that is in-
tended to be pegged roughly to the US Dollar, that adheres to the following elastic coin supply
rule:
When coin supply needs to expand (when the price of Celo Dollar is above the peg), the protocol
creates new coins, as in [17, 1, 2]. But rather than distributing them to token holders, it uses
8
them to purchase a basket of cryptocurrencies4 at market rates through a smart contract. These
purchases get added to the reserves. This is analogous to a central bank expanding the money
supply by buying financial assets on the open market and depositing them in the reserves.
When the coin supply needs to contract, the protocol uses reserve assets to buy Celo Dollars on
the open market. This is analogous to a central bank selling financial assets on the open market
in order to contract the money supply.
3. The protocol has a variable rate transfer fee on Celo’s native asset, to encourage long-term
holding of the reserve currency. The proceeds from the fee goes to bolster the reserves, and the
rate is based on the reserve ratio – the lower the reserve ratio, the higher the transfer fee.
4. The protocol uses a proof-of-stake model for governance. The weight of a node in governance
decisions is dependent on the amount of Celo they stake5
5. Every time a block reward is distributed, an equivalent portion of Celo is released. If the reserve
ratio is substantially higher than the target reserve ratio, then the released amount is largely
allocated for incentives (e.g. to developers and users). If the reserve ratio is substantially lower
than the target reserve ratio, the released amount goes mostly to towards bolstering the reserves.
An analysis of the stability characteristics of this protocol is given in [7].
3.3 Shared Reserves
While a single stable coin would be useful for several purposes (for example in cryptoasset trading
and internet commerce), a more robust ecosystem would involve a family of local, regional, and utility
stable value coins. The benefits of such a monetary ecology has been discussed broadly, for example
in [9, 16, 15], but here we focus on one: a stable currency is only meaningful if it is stable vis-a-vis the
price of goods and services that are purchased using that currency. Using a global currency for local
transactions would introduce price volatility in regions where regional consumer price dynamics vary
from the global consumer price dynamics6.
From a protocol perspective, we are interested in two mechanisms here: (a) a governance scheme
that determines how the protocol makes decisions on introducing new regional stable coins, and (b) a
structure in which the introduction of a new successful stable coin increases the stability characteristics
of the coins in the family.
As a starting point, we can imagine a protocol where each new stable coin is independent – there
is a blockchain and reserve for each new currency introduced. In this scheme, the governance question
is straightforward – teams will independently choose to introduce new stable value coins outside of the
protocol, and people can choose independently to purchase the new coins and their complementary
reserve assets. Governance on this issue is determined outside of the protocol, by the market.
However, this simplicity comes at a cost: the introduction of a new successful stable coin has no
stabilizing effect on existing stable currencies, and on the margins it has a small destabilizing effect7.
To address this issue, we introduce the idea of shared reserves. When the protocol introduces a
new stable value coin – for example, a stablecoin pegged to the Euro – the reserves for that coin are
the same reserves for Celo Dollars. When the supply of Celo Euros needs to expand, it expands using
the same mechanism as with Celo Dollars – the protocol creates new Celo Euros, and uses those to
purchase a basket of crypto assets for its reserves. When the supply of Celo Euros needs to contract,
the protocol uses the same mechanism as before: it sells reserve assets in exchange for Celo Euros and
retires the Celo Euros.
4Initially, the Celo native asset, and longer-term through a basket of cryptoassets via cross-chain decentralized
exchanges once available
5future versions of the protocol could be based on amount at stake and length of time remaining in their stake.
6For example, with Greece and the Euro, or with dollarization in Uruguay
7If the demand for the new stable coin is high enough, it could potentially cause a contraction in demand for
existing stable coins, reducing the value associated with the complementary assets of those existing coins, and increasing
uncertainty around long-term demand of the existing coins.
9
The protocol can make this process more efficient in the following manner: before selling the
reserves, it first looks to see if the supply of Celo Dollars needs to expand. If so, it creates Celo
Dollars, exchanges them directly for Celo Euros at the prevailing exchange rate, and retires the Celo
Euros. This is functionally equivalent to selling reserves in exchange for Celo Euros, retiring the Celo
Euros, and then buying reserves in exchange for Celo Dollars; it just disintermediates the reserves. It
only uses the reserves directly if the need for contraction of the Celo Euros is greater than the need
for expansion of all the other stablecoins supported by the protocol.
A shared reserve system must come together with a thoughtful method of governing decisions on
what new stable coins to introduce, and when to introduce them. If a new stablecoin is introduced
that has negative utility to the ecosystem, it can have a marginal negative impact on the stability of
the other currencies if the demand for that currency is high enough and volatile enough (for example,
a celebrity vanity stablecoin early on), or if the coin decreases aggregate demand for other coins
supported by the protocol (for example, the introduction of several duplicative regional currencies in
the same region with no differentiating features, causing confusion). For this reason, it is useful to
have a governance model that introduces a new stablecoin only if there is a widespread expectation
that its introduction would increase the aggregate demand for the family of coins over the long run.
We describe this governance model in Section 4.4.2.
It is useful to note that the shared reserve system does not require all new currencies to use the
shared reserve. In fact, for local or functional currencies, there are several reasons why it would be
useful to not engage in the shared-reserve model; we discuss these in Section 4.4.4. To support these
currencies, we also allow for new stabilized assets to be created with their own reserve; we call this
partitioned reserves. At a high level, the mechanism works in the same manner as the single stabilized
asset case, except that a third party can create the asset and initiate the reserve for that asset. For the
partitioned-reserve case, each reserve allocation is initialized at 25% Celo, 25% a local reserve currency,
and the remainder the same allocations as the shared reserve.
3.4 Price Discovery and Mechanics of Reserve Asset Purchasing
The Celo protocol is implemented as a fork of Ethereum. The cost of computation in the Celo network
is paid in Celo, just as Ether is used to pay for gas on the Ethereum network. Celo stable assets
are implemented as the equivalent of ERC20 tokens. One difference between the Celo protocol and
Ethereum is that while Ether itself is not compliant with the ERC20 token standard, Celo is. This
allows a decentralized exchange, through smart contracts, between Celo stable value assets as well as
Celo native asset, much like 0x [21]. This allows the automatic purchasing of reserves and distribution
of coins without cross-chain decentralized exchanges.
To determine the price of Celo stable currencies, the protocol will use a Schelling-point scheme
amongst stakeholders, with the weight of the a stakeholder’s vote dependent on the amount of Celo at
stake. One can imagine further augmenting the Schelling point scheme with price feeds from exchanges,
as determined through a governance scheme.
4 Governance and Incentives
A primary incentive mechanism in the Celo protocol is the distribution of block rewards, which are
allocated to the various contributors to the system – those who maintain the protocol (by selecting
validators, validating transactions, verifying users, and participating in the Schelling-point price dis-
covery mechanism), those who contribute to the robustness of the reserves, those who take on risk in
the case that there is a contraction, those who use the protocol as their means of payment, those who
invite others to use the protocol, and those who improve the protocol (by participating in governance,
and by making technical contributions to the protocol). We describe these below.
4.1 Maintaining the System
The system uses a proof-of-stake mechanism for selecting the validator set and participating in gover-
nance decisions. Both validator election and governance decisions are made through a bonded-stake
10
weighted voting scheme. Any Celo holder may put up a bonded deposit, which involves locking Celo
in a smart contract. Votes (for both validators and in governance) are weighted by the amount of Celo
in the bonded deposit. Any locked Celo should continue to be locked up for a certain time period after
voting, and in the future it would be helpful to weight voting by the length of time remaining in the
lockup. This incentivizes long-term holding of the reserve currency and aligns governance decisions
with long-term perspectives. Block rewards are distributed amongst those who participate in validator
elections and governance decisions.
Users don’t vote for validators directly. Instead, validators are expected to organize themselves into
groups and account holders vote for these validator groups. Just as anybody in a democracy can create
their own political party, or seek to get selected to represent a party in an election, any Celo user can
create a validator group and add themselves to it, or ask an existing validator group to include them.
Validator elections are held once every epoch, which corresponds to approximately once per day.
Validators, once elected, put up a slashable bonded deposit, participate in the consensus scheme,
send verification messages, participate in the Schelling-point scheme for price discovery, and receive
block rewards to cover their costs and as an incentive for their work to maintain the system.
4.2 Bolstering Reserves and Contracting Stable-Value Currency Supply
when Needed
Celo holders, through their purchase, give the initial value to Celo’s native asset and introduce other
crypto assets into the reserves. Further, Celo holders bear some risk in the case of contracting supply
or a dip in the reserves: transfer fees are imposed if the reserve ratio goes below the target reserve
ratio, and the value of Celo may go down if there is a contraction in demand for Celo stable currencies.
Celo holders may be rewarded for playing these roles in two ways: first, as there is greater demand for
Celo stable currencies, there will be more protocol-directed purchases of Celo, increasing the demand
for fixed supply. Second, if the reserve ratio is greater than the target reserve ratio, Celo holders
are rewarded with a portion of the block rewards (provided that they are participating in consensus
on transaction validation, sending verification messages when selected, and participating in Schelling-
point voting for price discovery). These rewards are paid in proportion to the amount of Celo at stake
8.
4.3 Increasing User Base and Usage of the System
Active users (people who use the payments protocol, participate in phone number validation through
the mobile wallet, and maintain a nominal stake in Celo) are rewarded through block rewards. In
effect, this reduces transaction fees for active users. (One can even imagine a scenario in which these
block rewards are issued by waiving transaction fees for a certain number of transactions in the stable
currency per unit time, implemented through part of the block reward going to paying transaction fees
of users, set at a rate to ensure a certain transaction speed, and prioritized based on the amount of
their stake.)
4.4 Improving the Protocol
And finally a continuously evolving protocol requires incentives, and a governance scheme, for improv-
ing the protocol.
4.4.1 T echnical Improvements
For technical improvements to the protocol, anybody may put up a bonded deposit to make a technical
proposal, with a proposed fee-for-implementation, on a regular cycle9. Proposals will be voted on by
long-term stakeholders, similar to the voting scheme with Dash’s masternodes [8], with their votes
weighted by the amount of their stake and notice period. Funds that are not allocated in a particular
cycle are added to the reserves.
8and in future versions of the protocol, the time remaining in the stake
9This mechanism can also be applied to other types of proposals, for example for marketing proposals.
11
4.4.2 Introducing Regional Currencies and Broadening the Reserve Base
Over time, it would also improve the protocol to introduce more stable value currencies, and to broaden
the reserve holding프로젝트로 돌아가기