estimated reading time: 30-35 minutes
Satoshi left Bitcoin only half designed and useless as money. He created a consensus algorithm, but no economic algorithm. Bitcoin can track 21 million coins. That's great for 'peer-to-peer electronic cash'. But when that fixed supply of coins meets variable demand, the result is wild fluctuations in market price. That's terrible for cash. That's Zimbabwean cash. Without stable value, the conversation about peer-to-peer electronic cash got shouted down by a really boring conversation about price movements.
I first got interested in stablecoins because I thought it was a fun word to say — stablecoins, stablecoins, stablecoins — but the rest of the crypto community is really starting to pay attention. Even MIT Technology Review covered the trend. Lots of brainpower and capital is researching the question Satoshi left unanswered. Researching this article, I found at least 13 projects to create decentralised stable cryptocurrencies, four to create centralised ones tracking the dollar, and about 50 tracking the price of gold. There are some multi-billion dollar stablecoins in the waiting room of the near future.
This is not unexplored territory. Every government tries to stabilise its national currency. Sometimes, like in Zimbabwe or Weimar Germany, they fail spectacularly. Even their success is never perfect. USD and EUR are as respectable as can be, but their exchange rate moved 0.36% over the past 24 hours, and has moved within a 13% range over the past year. Currencies like the dollar inflate a few percent per year. Cryptocurrency should aim for imperfect stability that is good enough for the cash to be usable.
This stabilisation requires economic thinking with the same sort of deep, careful, intelligent paranoia needed to secure computer systems. Blockchain's strength isn't that no one tries to attack it; it's that it can withstand every imaginable attack. The same cat-and-mouse game happens in economics. In 1992, an economy-hacker from Hungary outsmarted the Bank of England's stability mechanism in an audacious attack that cost the United Kingdom £3.3 billion.
The Linden dollar (the currency of Second Life) is not really a cryptocurrency. It is certainly not decentralised. It lives in a database on the company's servers, much like World of Warcraft gold or whatever. Yet it's worth mentioning because it has traded between 240-260 to the dollar for years, is used for real trade, and is what got me going down this rabbithole exploring stablecoins.
This is a long post that will go into details about decentralised currencies that maintain a steady price by dynamically balancing supply and demand, like decentralised Linden dollars. But first we'll look at a simpler, centralised design.
- Centralised stablecoins backed by the U.S. dollar
- Centralised cryptocurrencies backed by gold
- Centralised cryptocurrencies backed by something else
- Decentralised stablecoins
Centralised stablecoins backed by the U.S. dollar
Tether (USDT) is the most popular stablecoin, and the first to reach a valuation of over a billion dollars.
It's as simple as can be: there's a bank account somewhere with x U.S. dollars in it. The Tether people issue x of their tokens, declaring (by fiat) that each token corresponds to one of their dollars. Very simple. Very centralised.
Will you remind me please why the cryptocurrency community is opposed to centralisation? Is it for purely rhetorical reasons? Or is it because humans with a lot of decision-making power invariably make corrupt decisions in their own self-interest and against yours?
Centralised monetary systems can refuse to give you your money. The banks of Cyprus did it. Paypal does it all the time. Tether can do it, saying, "redemptions will not be unreasonably denied, but we reserve the right to selectively deny redemption and creation of Tethers on a case-by-case basis." Other cryptocurrencies reserve no such right — which is the whole point.
I said that Tether tokens correspond to dollars that sit in a bank account. Let me correct that: Tether tokens correspond to dollars that allegedly sit in a bank account.
It is reasonable to expect that Tether will expose themselves to the scrutiny of a neutral auditor to reassure us that the dollars are there. But Tether fired their auditor before a report could be produced. Ars Technica did good reporting on the issue, including on how Tether increased their supply by $2.15 billion while at the same time claiming they were unable to receive wire transfers. Tether's whitepaper states they are registered in Hong Kong. However, the Panama papers show they are registered in the British Virgin Islands. On June 20, they produced an incomplete audit, and Stu Hoegner, Tether's general counsel, claimed that all cryptocurrency companies have difficulties getting audited. I reached out to Hoegner on June 20, asking how, if that was true, their competitor TrueUSD has produced eight audits in three months. Kasper Rasmussen, Director of Communications for Tether-affiliated Bitfinex, told me, "We cannot comment on TrueUSD's work or on the work product produced by Cohen & Company... However, we note that we have been pursuing audit engagements with some of the largest global accounting firms. We also note that the size of the TrueUSD funds examined is considerably smaller than Tether's bank balances."
If the situation is as bad as it seems, then they don't have the money in the bank account, and are minting money out of thin air backed by nothing but lies.
Unfortunately, this could be much worse than $2.15 billion. Tether has the highest trading volume of any cryptocurrency apart from Bitcoin. The BTC-USDT trading pair makes up a significant fraction of bitcoin trading. If people are creating fake Tethers, they could use it to buy bitcoin at inflated prices. This would drive the market price of bitcoin up. If they hold Bitcoin, they stand to profit hugely by such a market manipulation. They have motive, means, and opportunity, and a statistical study released on June 13 found the fingerprints of such manipulation.
Trust Token, Circle USDC, and Stably
Trust Token's currency, TrueUSD (TUSD), uses the same model, but is much less shady. If we're gonna throw out the whole idea of cryptocurrency and trust a central authority, let's at least find someone trustworthy, who submit to checks and balances. Trust Token regularly publish independent audits to verify their claims. They emailed me on June 18 to report that they'd completed a $20 million funding round, including backing from Andreessen Horowitz. TrueUSD is growing rapidly, and doubled its market cap from $40 million to over $80 million in the past month, becoming one of the 100 largest cryptocurrencies. They are still not nearly as popular as Tether, with about $83 million in circulation compared to $2.6 billion for Tether.
Circle are doing the same. You make a wire transfer to them, and receive their Ethereum-based USDC tokens. This is not a new currency, but an on-ramp and off-ramp between the dollar and Ethereum ecosystems. Stably is similar, but is not released yet. Let's move on. I don't really have any interest in centralised systems and I don't spill ink on them.
Centralised cryptocurrencies backed by gold
That same model — having off-chain assets, and issuing blockchain-based tokens that correspond to them — can be applied to gold. Like the dollar-backed tokens I just badmouthed, this is totally centralised. However, I think it's slightly preferable for three reasons. Firstly, it doesn't depend on the American Fed. Secondly, it is likely to slowly increase in value (until someone figures out asteroid mining), making it better to buy-and-hold. And thirdly, because I was a pirate in a past life so I just like gold.
There are many gold-backed cryptocurrencies (here's a good attempt at enumerating them), but this article is already monstrously long, so I'm only going into detail on my favourite one: DGX. Some use a fractional reserve model; most claim to be fully reserved. Below, we will discuss BitShares' currency stabilised against the U.S. dollar; they also have one stablised against gold.
Digix Gold Token
Digix Gold Token (DGX) is pegged to a shiny pile of gold in a strong room in Singapore. Like Trust Token, they have credibility, exposing themselves to the scrutiny of third-party auditors. Their whitepaper is not so much about gold as about a protocol for on-chain proof of off-chain assets. Custodians and auditors digitally sign their reports, and stamp them on the Ethereum blockchain, providing evidence in the public ledger that the gold is really there. Scans of receipts and paperwork are uploaded to IPFS, and the hashes are included in the on-chain audit trail. In short, the gold to back it really is there.
While the gold is centralised, governance is decentralised in a DAO. They are currently hammering out the details of the decentralised governance model and posting updates on Medium. The DigixDAO token (DGD) grants voting rights on proposals about running Digix (DGD does not pay dividends). DGX holders pay a small fee to cover the costs of storing the gold in the vault (though the appreciation of gold should easily cover this). The fees are detailed in CoinCentral's guide to Digix.
The proof-of-provenance protocol Digix have pioneered has huge potential. It gives solid mathematical verification of off-chain, real world assets. The same cryptographic protocol could be applied to silver, platinum, palladium, gemstones, art, real estate, or racehorses.
DGX has been live since May 2018, so you can send and receive it now. For instance, if you decided you really appreciate this article, you could send some DGX right now to 0x4F7960d249B4E4f6C69f822B80C40A66010EB74C. It is available on Airswap, Kyber, and other exchanges. At the time of writing, etherscan.io shows 50,300 DGX tokens on the blockchain, indicating 50.3kg of gold, equivalent to a rather modest market cap of about $2.1 million, or about four Good Delivery bars, which weigh 12.4kg and look like this —
Centralised cryptocurrencies backed by something else
Globcoin issues ERC-20 tokens backed by a basket of 16 things: the national currencies of the 15 largest economies (by PPP-adjusted GDP), and gold. This is somewhat like Trust Token's model, but with less dependence on the U.S. dollar. You can read their whitepaper here. The website and social media posts look a bit amateurish. It seems to be fighting (and losing) for the rights to the ticker symbol GLX with something called GalaxyCoin.
AAA Reserve is backed by several currencies, gilts (government-backed bonds), and AAA-rated credit investments. It is centralised in a New Jersey company called Arc Fiduciary Ltd. that require you to verify your identity before you buy their tokens. This is not what I would call a 'cryptocurrency'; it's a standard investment fund that has tokenised. It's only interesting because it diversifies beyond the dollar.
Centralised systems are boring. Let me get on to what I really want to talk about: decentralised stablecoins.
Bitcoin was supposed to be money, not a buy-and-hold investment. Excessive volatility has changed that. Crypto needs stability if it to is to mount a challenge to legacy currencies. Yet the soul of cryptocurrency is decentralisation, so tethering to a Cayman Islands corp, or trusting a private company, or circling back to Goldman Sachs, is not satisfactory. We need an economic system that can accomplish stabilisation in a decentralised way. To do that, the system must accomplish three things:
- Know its current market price. Is it too high? Too low? Oracle – besides being my playa name – is the word used for mechanisms like this, that feed external information into computer systems.
- If price is too low, the system needs a way to bump it up. Because price results from the ratio of supply and demand, and because demand is a squishy real-world thing no one can think of a way to algorithmically control, the way to bump up price is to cut supply. (This is sometimes called contraction, meaning contraction of supply. That's not my preferred terminology because my little brain gets confused about whether the supply or the price is being contracted.) Most stablecoins have a mechanism make it profitable at times for people to burn the stablecoins, or park them until the price rises.
- If price is too high, knock it down. The usual way to do this is by increasing supply. (This is sometimes called expansion.)
One of the most promising models for decentralised oracles is the Schelling point scheme. Of the stablecoins we're about to see on our tour, Kowala, Carbon, and Basis have confirmed they are using it, and others are considering it. Here's how the Schelling point scheme works –
- Every so often (every 5 minutes in the case of Basis, every 24 hours for Carbon), anyone on the network can submit their opinion on what the exchange rate is.
- Opinions are weighted by stake. The weighted median is the answer.
- People who guess near the median (maybe between the 25th and 75th percentiles) are rewarded with coins to incentivise normal answers. You could go further and punish bad answers. Carbon does this, saying in their whitepaper, "Anyone who bids outside the 25th and 75th percentiles will have their balances slashed. Anyone within the 25th and 75th percentiles receive a normal distribution of the loser’s balances, with the highest reward distribution at 50% and normally diminishing on the right and left respectively."
The rewards should cause more people to participate, which will make the system more resilient against attacks. People can't mess up the answer by putting in rogue data, because outliers will be smoothed out in step 2, and because they'll lose their money in step 3.
You may notice that an attacker with 26% of the stake could make outrageous submissions fall within the accepted band. (Carbon's normal distribution would help diminish this threat.) Vitalik Buterin said, "if large collusions are possible then it may break down, but it is likely close to the best that we can do". Buterin talks about how to make a Schelling oracle more secure here.
An alternative approach to decentralised oracles is the one taken by BitShares and DAI. Stakeholders vote on good data sources, and fire ones that give bad information. Again, the median of all inputs is taken to smooth out differences and ignore outliers.
The dual-token model — stablecoins and shares
Basis, Carbon, DAI, Havven, Reserve, NuBits, Reserve, and BitShares use variations on this scheme. One token is stable, and the other is used to whip it back onto the narrow track. Roughly, the idea is that supply of the stable token is dynamically increased and decreased. Supply is increased by paying free coins to holders of the volatile token, and decreased by taking coins off the market in exchange for volatile tokens.
All these dual-token stablecoins are influenced by a 2014 paper by Robert Sams called A Note on Cryptocurrency Stabilisation: Seigniorage Shares. Sams points out that we hold some coins for spending, and some in the hope of future returns. If too much of the currency is held for future returns, it stops being used to transact. (This has happened with Bitcoin.) The solution he proposes is to split the system into coins (good for spending) and shares (which give profit if the system grows).
His equations on page 3 of the PDF simply say that when the market price of the coin changes, the supply should change by the same ratio. If price equals demand divided by supply, then for any particular level of demand, there is some supply that makes the price equal to the target price.
Sams proposes that, when the market price is above the target, people who hold the profitable share-token could create coins at a preferential rate. And, by contrast, when the market price is too low and supply needs to be reduced, they can profitably destroy or park coins by turning them back into shares. These share-to-coin and coin-to-share conversions are profitable, incentivising people to use the system. (This profit is the meaning of the word 'seigniorage'; seigniorage is the profit made by minting money, as when a king turns a lump of silver into a coin worth more than the lump.)
I think of it slightly differently. There are one-token models like Tether, TrueUSD, and DGX. Then the dual-token model has two flavours: the kind where the stable token is created by locking up other crypto (e.g. DAI and Havven), and the kind where it is not (e.g. Basis, Carbon). Crypto-collateralised coins win confidence by being backed by a valuable cryptocurrency, whereas the other ones must win public confidence through charm alone.
People often initially think stablecoins aren't investable, and this is true in a one-token system. However, in a dual-token system, the volatile token can be an investment opportunity. Stablecoins will gain billions in market cap over the next 12 months, and these volatile tokens are going to make some people very rich.
This opportunity will bring capital into the stablecoin space, and help build a stable peer-to-peer electronic cash system.
NuBits described a system in their 2014 whitepaper, and launched it on a proof-of-stake blockchain called Peershares in autumn of that year. NuBits (USNBT) are the stable coins in the system. Custodians are market-makers that try to drive the price always towards $1. NuShares (NSR) are not volatile tokens used to incentivise stability, but rather governance tokens that grant voting rights.
Shareholders vote and elect custodians. Shareholders also vote to create new NuBits which are given to the custodians. The custodians are elected on the strength of their promise to spend their NuBits for the public good. This mostly means selling them on exchanges for exactly a dollar to control the market. Anyone who tries to sell NuBits at $1.01 will be trumped by a more appealing offer from custodians selling at $1, and conversely, anyone offering to buy at 99c will be outbid by custodians who'll buy your coins at $1. This is the chief stabilisation mechanism.
The second stabilisation mechanism is parking NuBits to reduce the circulating supply and drive the price up. Parked NuBits are locked away for a fixed length of time, and earn interest. Shareholders vote on how much interest, increasing the rate when more drastic price-correction is needed. No oracle is needed, as human beings are in charge of the process.
All this has a political flavour, and any free-marketeers reading will be fuming. Shareholders elect human custodians who promise to make beneficial market interventions, and the interest rate on parked coins is voted on, not algorithmically determined. This is a manual approach to stabilising the currency, akin to government intervention. It works as long as the custodians have enough reserve cash to plaster over their differences of opinion with the market. The parking mechanism can whip the price back up, but mostly, the custodians buy and sell to manually control the market. Here is what happens when they run out of coins:
That wasn't terrible. It held stability from Sept 2014 to June 2016, and again from Sept 2016 to March 2018. It worked until it didn't.
Ethereum makes it possible to replace custodians with smart contracts. These smart contracts would accrue coins from stability fees or minting, and sell them on exchanges at exactly $1, no more and no less. (The Fragments reserve we will look at later works this way.) This could be a useful adjunct to other stabilisation mechanisms, but the real lesson of NuBits is that we need more organic, market-based stabilisation mechanisms. Buying the market over to your opinion won't work forever.
Someone here at Crypto Insider says that BitShares are the Simpsons of crypto; every idea someone has, BitShares already did it. They did decentralised exchanges and stablecoins back in 2014, long before they were fashionable.
It could be argued that BitShares have already solved the problem of a stable cryptocurrency. Their stablecoin, BitUSD, has been trading against the U.S. dollar since Sept 23, 2014. Over that time, its price has averaged $1.0240, with a standard deviation of 0.07896. For comparison, the euro traded against the dollar over the same period with a standard deviation of 0.05767, and the British pound with a standard deviation of 0.116. (Data is from Coinmarketcap and investing.com.) So although BitUSD tracks the dollar noisily, it proves that cryptocurrencies can be as stable as other currencies, avoiding the mad swings of Bitcoin.
The oracle system is simple and effective. Shareholders elect 'witnesses' to feed trustworthy information on exchange rates into the system. Witnesses who give bad data can be fired by the shareholders. The system takes the median price of the witnesses' opinions. It's the same system used by DAI.
The aim of BitShares is not to trade exactly 1:1 with the dollar, but to maintain a floor of $1 value, and generally trade slightly above. It maintains this floor in a very simple way: the system always lets you redeem a BitUSD for $1 USD worth of BTS. It would be silly for anyone to sell BitUSD anywhere at $0.50 instead of doing that.
New BitUSD are created when two parties agree on a loan contract on the BitShares platform. (As we will see, DAI and Havven's nUSD are created in a similar way, as debt on crypto loans.) Alice locks BitShares (BTS) into a loan contract, Bob pays for them and receives newly-minted BitUSD. The price Bob pays is slightly more than the oracle's price. Bob pays this premium in exchange for the safety of BitUSD against volatility, and the peace of mind of knowing that he can cash out at any time at the oracle's price: he can do this because the software gives the BitUSD holder (Bob) the power to terminate the loan at any time (in other words, to margin call it), and receive $1 USD worth of BTS at market price. The system will pay Bob from the open loan with the lowest BTS:USD rate.
If Alice puts $100 worth of BTS (at the oracle's current price) into a loan, and cashes out when BTS is worth $50, she receive twice as many BTS. In this sense Alice is shorting — betting that the value of BTS will fall.
From that brief description of how the system works, we can see how it creates stability. It can't go below $1, because the system will always pay you a dollar of BTS for it. It can't go much above, because people can generate new BitUSD by locking up $1 worth of BTS and selling it, with a small premium. (This premium is the reason why BitUSD generally trades a few cents above the dollar.) People opening short positions will compete with each other to sell it as close above $1 as possible.
(The Basis whitepaper, by the way, totally misunderstands the BitShares system. It says that short positions on BitShares last 30 days, whereas BitShares website specifically says, "Short positions never expire".)
BitShares clearly inspired certain aspects of DAI, which uses the same oracle system and creates and destroys coins by a system of loans collateralised by cryptocurrencies. Let's look at DAI in more detail —
DAI is the current leader of decentralised stablecoins. It launched in late Dec 2017, and is actively trading on many exchages, including Airswap, Radar Relay, and Kyber. You can store it in Metamask, Trust Wallet, MyEtherWallet, etc., or you could send it now to 0x4F7960d249B4E4f6C69f822B80C40A66010EB74C if you found this article valuable. DAI is true to the spirit of decentralisation, governed by a DAO, and has maintained stability quite well so far, with a standard deviation of just 0.0117 against the dollar, although for only about six months at a modest (though growing) market cap.
MakerDAO dates back to 2015, and has kept an impeccable reputation, with reassuring activity and community engagement on Reddit and an active Medium blog. The MakerDAO Medium explains how the DAO votes on the codebase governance — one MKR token, one vote. (Note that just two wallets hold the majority — 61.7% — of MKR tokens. Founder Rune Christensen clarifies, "The largest account is the developer fund multisig contract, which is used to grow the system and fund charity. The second largest account is actually the voting smart contract, so it doesn't represent one wallet, but rather many wallets that are actively voting.")
It uses a similar system of oracles to BitShares: "MKR voters choose a set of trusted oracles to feed this information to the Maker Platform". There is another security feature: the oracles can only change the price at a limited rate, so even if an attacker gained control of the oracles, they could not change the price by 99% in one second. This brake gives the good guys time to detect shenanigans.
Like other coins we're looking at, the DAI system controls the price of the stablecoin by controlling the circulating supply, and controls the circulating supply by economic incentives.
New DAI comes into the world when people go to dai.makerdao.com with a Web 3.0 browser, and lock their ETH into a smart contract. (This is called a 'collateralized debt position', or CDP, but don't worry about the jargon. It plays the role of the seigniorage shares in Robert Sams' proposal.) You can draw new DAI out of this contract. It is like taking out a cash loan form a bank, and locking your home into the contract as collateral. You can later close the loan by paying the DAI back into the smart contract (which results in the DAI vanishing into the ether).
That's how DAI is created and destroyed. As we've seen, the way to stabilise price is to tempt people into creating coins when the price is too high, and into destroying them when the price is too low.
That's what the DAI system does.
If the price is too high (say $1.01), the system needs to increase the supply. So the system makes the terms of the loan more appealing. At these times, I can deposit (for example) $9 worth of ETH into the smart contract, and get $10 worth of DAI back. That's a good deal. People like good deals. People do it. More DAI squirts out of the smart contract, into the marketplace. Supply increases, price falls.
Conversely, if the price is too low (say 99c), the system wants to tempt people to close their loans. So it allows people to close their loans at a favourable rate, say paying off a $10 loan with $9 worth of DAI. Again, that's a good deal, so it naturally makes people take their DAI out of circulation. Less supply, higher price.
By opening and closing these debt positions, active participants make decent profit. This is what Robert Sams called 'seigniorage', i.e. the profit made by creating money.
The most common criticism levelled at DAI is that the collateral is ETH, which is very unstable. The system relies on people cashing out their loans for ETH, but if ETH becomes worthless, that won't be appealing. If no one wants to cash out their positions, excess DAI won't leave the marketplace, and the price will fall. This criticism has been made here and here by Preston Byrne. (It should be noted that Preston Byrne is generally cynical and angry about stablecoins. He has impressive credentials in economics, but often gets things wrong by being too negative, like saying that Basis uses centralised oracles, or that BitShares lost stability.) There are three counterpoints to be made to this —
- DAI maintained stability earlier this year when the price of ETH fell from $1200 to $400.
- There are plans to diversify collateral in future.
- The DAI system has one more safety net against this eventuality, described under the heading 'Automatic Liquidations of risky CDPs' in their whitepaper. If the ETH locked up in the the loans falls too far in value, the DAO can automatically liquidate the positions, selling off the ETH for DAI, which is then burned out of existence. (This safety net, too, was inspired by BitShares.)
There are more reasonable criticisms of the DAI system on Reserve's Medium and in the Basis whitepaper, but time will tell how successful DAI is. So far, it's holding, and we will be able to rest a bit easier when it implements diverse multi-collateral loans, rather than backing them with ETH alone.
Havven (pronounced 'haven')
Havven raised $30 million in Feb, the highest of any Australian ICO to date. They launched eUSD (a stablecoin backed by ETH) on April 11, then launched nUSD (collateralised by their own HAV token) on June 11. eUSD can be traded on Radar Relay, though I don't see nUSD there yet. You can buy HAV in any of these markets and switch it for nUSD using this converter. All these coins live on Ethereum, so you can store them in Ethereum wallets like MetaMask, Trust Wallet, or MyEtherWallet.
Havven is a dual token model, where nomins (nUSD) are stable and havvens (HAV) are volatile. I like the simplicity with which it implements the seigniorage shares system. Havven-holders get transaction fees from the network, and — very cleverly — HAV accounts that participate more in stabilising the price are rewarded with larger shares of the transaction fees.
As with other dual-token stablecoins, people mint new nomins by locking up their havvens. They have to stake more than $1 worth of havven to create one nUSD, to ensures a buffer of collateral if the price of havven falls —
Confidence in the stability of the nomin begins with overcollateralisation, so that the value of escrowed havvens is greater than the value of nomins in circulation. As long as the ratio of total nomin value to total havven value remains favourable, there is sufficient backing in the underlying collateral pool to ensure that nomins can be redeemed for their face value. The redeemability of a nomin for the havvens against which it was issued strongly supports a stable price.
Because every nomin on the market was created by staking havvens, each nomin has some corresponding havven backing it. The whitepaper calls ratio of the market cap of havven to the market cap of nomins the overcollateralisation ratio —
This can be re-arranged like this —
The number of havvens is fixed (100 million were created in the ICO). So for any particular price of havven (which floats up and down on the open market), there is a certain supply of nomins that gives nomins a price of $1.
The only variable the system needs to control is the overcollateralisation ratio. In other words, how many nomins to mint for each havven someone stakes. In other words, the terms of the loan. (If a lot of nomins are created for each havven staked, that incentivises people to create more nomins; if few nomins are created, people create fewer nomins, and are more encouraged to cash out nomins for havvens.) By tweaking the collateralisation ratio, the system controls the circulating supply of nomins.
The whitepaper lists "Fast decentralised oracles" as an area of research. They could probably learn something from Carbon and Basis's whitepapers. Project manager Garth Travers told me by email that they are temporarily using a weighted aggregate of data from exchanges as an oracle. The code that does this is currently closed-source "to protect it". He said, "It pushes the price to the smart contract every 30min, or when the price has moved more than 1%". This is an easy way to implement an oracle, but also the most vulnerable; an attacker who spoofed the feed would destroy the stability of the currency in minutes. They are drawing some fire on their Telegram for this. However, I do believe that they are working on a more resilient, decentralised oracle. Travers told me, "This is temporary while we are looking into approaches for a more decentralised solution."
They are planning to expand beyond the USD to other currencies in Oct-Dec 2018. It will probably be the euro, British pound, Japanese yen, or the national currency of Australia, the dollarydoo:
Basis, still in development, is the most fashionable stablecoin being discussed at San Francisco crypto cocktail gatherings. It has a very good slogan: "A stable cryptocurrency with an algorithmic central bank", and recently raised $133 million privately, from partners that include Polychain Capital, Pantera Capital, amd Andreessen Horowitz. They seem badass. Their FAQ says (paraphrasing a bit for dramatic effect), "Q: What if someone tried to attack it like Soros did to the GBP in 1992? A: The guy who did that for Soros is with us. He designed the defenses."
Like most of these other projects, it monitors the market price with an oracle, and when the price gets too low, makes it profitable to take coins out of circulation, and when the price gets too high, issues new coins.
The oracle described in the whitepaper is one of the best I've seen of any project. It is a decentralised Schelling oracle, checking the price every five minutes.
When the system needs to take coins off the market to increase the price, it allows people to buy Basis Bonds for less than one basis. (These are more like put options than actual bonds.) These bonds hold the promise of a $1 payout in the future. The stablecoins used to buy the bonds are burned.
To increase supply and lower price, it pays out new coins to people who hold Basis Bonds. The first to buy their bonds get paid first, with a payout of one stablecoin per bond.
(Note that in BitShares, Havven, and DAI, people needed to lock up a different cryptocurrency to create the stablecoins, whereas here that is not the case. This lack of collateral is the most controversial aspect of Basis.)
There is also a third token, called shares. These are "tokens whose supply is fixed at the genesis of the blockchain", presumably to compensate early backers and team members. Shareholders get paid newly-minted basis, "if there are no more outstanding bond tokens". So if the system needs to mint 1 million basis to dilute the supply, and there are 800,000 outstanding bonds and 20,000 shares, then the 800,000 bondholders get one Basis each (one bond always gets one basis), and each shareholder gets 10 basis. I don't see the need for shares as a stabilisation mechanism. I've been around enough token launches to know what it looks like when an unnecessary token is added to make more profit for the company and partners. The shareholders get ongoing rewards while doing nothing to help the ecosystem after their initial investment; contrast this to, say, Havven, which gives rewards only to people who correct the price.
One criticism of Basis is that it when it becomes unstable, it incentivises people to restabilise it with the promise of more Basis in the future. But if the system has become unstable, will people find that promise appealing? The system has to pull itself up by its own bootstraps: a consequence of not using another cryptocurrency as collateral when Basis is minted. This has been pointed out by Basis's rival Reserve, and by some guy called Vitalik, who said, "if the basecoin price goes down, then the mechanism pushing the price back up is to get people to buy basebonds, but basebonds basically just lock you into holding basecoin, and it’s not clear why people would want to do that". (Basecoin is an old name for basis.)
This danger is exacerbated by the first-in-first-out nature of the bond queue. The longer the queue becomes, the less appealing it is to buy bonds. The Basis team have considered this, and claim firstly that people are incentivised to buy quickly (because the first to buy get paid first), and secondly that their statistical models show that a five-year expiration on bonds cuts the queue to just the right length. Reserve rebut that if the mechanism to restore faith in the project takes years, people will have long lost faith and interest.
I find the duality of Basis's reasoning a bit strange. Shareholders don't get paid at all unless the bond queue falls to zero. Yet they anticipate the bond queue being up to five years long. Which is it? Will the queue be years long, or nearly zero? If they don't know, what does that say about the precision of their statistical modelling?
Carbon is very similar to Basis in a lot of ways. Their whitepaper describes a Schelling oracle that checks in every 24 hours (compared to every five minutes for Basis). Their oracle scales rewards so that they are greater near the median, and diminish the further from the median you get. It runs on a hashgraph rather than a blockchain.
Carbon is yet another dual-token system, with a volatile token called Carbon Credit used to stabilise a dollar-pegged coin called CUSD. When the oracle says the price is above the target, free coins are minted and distributed to holders of Carbon Credits. These coins find their way to exchanges, inflating supply and reducing price. (If Carbon takes off, these dividends could be significant.)
When the oracle says the price is below the target, the smart contract initiates an auction where people can buy Carbon Credits for CUSD. These CUSD get burned, reducing supply.
This system is very much like Basis and can be criticised on the same grounds: if people lose confidence in the project, they won't buy the credits and the mechanism to reduce supply will fail. But who knows, maybe it will whip up enough enthusiasm for long enough to become big enough that people just trust it. All money is a shared hallucination anyway.
Overall, Carbon seems like one of the more well thought-out projects, with good logic around decentralised oracles and economics. Hashgraph's intellectual property restrictions may come back to bite them, as a corporate board will be in charge of their protocol, and they will not be able to fork if they need to.
Fragments is another unreleased, Ethereum-based project. The CTO Brandon Iles has said on Telegram that there is no public code base yet, but the code will be open-sourced when more mature. It is very different from the other coins we're looking at. Whereas most current stablecoin projects are based on Robert Sams seigniorage shares proposal, Fragments follows Ferdinando Ametrano's Aug 2014 paper: Hayek Money: the Cryptocurrency Price Stability Solution (which Sams' paper criticises). Kevin Dowd has also argued for this approach to currency design.
The idea is simple, weird, and has a certain brilliance: if the market goes up, you get more coins, but each coin is still worth about the same. So if the price doubles, each coin in your wallet becomes two coins and each is still worth $1.
Fragments is not trying to compete directly with the other projects here. Their market positioning is different. CEO Evan Kuo said on Telegram, "We don't need to position as the type of coin that an exchange would need to de-list Tether for... We can enter as a speculative instrument with downside protection... almost as a hedging instrument... We will likely enter the market as a buy and hold hedge (a store of value token) as opposed to a medium of exchange token."
Like the other projects, it solves the volatility of units. I could list an item on OpenBazaar for the price of 23 Fragments, and people would be clear about what they're paying. (Whereas 23 dogecoins is a very variable measure of value.) However, unlike the others, it doesn't even attempt to solve the problem of fluctuating wallet balances. You could still wake up in the morning with half your money gone. The volatility is still there; it's just shifted from the per-unit value to the number of units. Ametrano's paper that inspired Fragments says,
Price stability, i.e. preserving the value of a currency unit, has always been the Holy Grail of monetary theory, and Hayek Money is finally limiting its elusiveness. Unfortunately the more ambitious goal of preserving the value of a wallet is a chimera, and it deserves to be clearly appreciated as airy-fairy wish.
The Fragments system doesn't just rely on the splitting mechanism for stability. It includes a smart contract called the reserve, which is a bit like a central bank, or an automated version of NuBits' custodians. This smart contract functions like a whale manipulating price, aiming to keep it at $1. It may seem heavy-handed to buy a market over to your opinion on the price, but their website argues that it can be done with a relatively light touch. The daily trading volume, to begin with, might only be a few million. We know that one whale can move a market; it doesn't require a monopoly. The reserve may only need to make up a few percent of trading volume to attract other traders towards the $1 target.
When fluctuations get too big for the reserve to buy them off, there are further stabilisation mechanisms:
When the oracle says the price is below $1 (the details of the oracle have not been worked out yet), the reserve offers to sell bonds for less than 1 Fragment, much like in Basis. If the price is very low, it will sell them cheap, maybe 0.5 fragments per bond. If a smaller adjustment is needed, it will sell them higher, maybe at 0.9 fragments. Fragments are the only currency accepted in bond sales. This takes fragments off the market, absorbing them into the reserve. This scarcity drives the price up.
When the oracle says the price is above $1, new Fragments are generated. First, outstanding bonds get paid off. Then three groups receive new coins: a portion to the reserve, a portion to the Fragments Foundation, and a portion to all wallet holders, who will wake up happy to find they have more coins in their wallet.
The reserve takes a bigger slice of the pie when the market has been jumpy, and when the swings are gentler, the walletholders get a greater portion. The developers are still tweaking these parameters: what portion should the Foundation, the walletholders, and the reserve get? How much should this ratio change in response to various levels of market volatility?
The reserve is constantly trickling out Fragments to the marketplace, at the price of $1 worth of ETH each. This accomplishes two things: firstly, it forces other traders to compete with this price, so it will be hard for them to sell Fragments for more than $1. Secondly, it builds a stockpile of ETH in the reserve. (The team plan to later diversify into other cryptocurrencies.) This stockpile gives users the peace of mind of knowing that their tokens can be redeemed for something valuable. In a situation where faith in the Fragments system was failing, and price fell to $0.95, the reserve would use its ETH to buy fragments at market price, keeping tokenholders happy, and keeping the market price at $1.
MinexCoin (MNX) is the worst stablecoin I have ever seen:
Their whitepaper is a mish-mash of obvious statements, vague plans, and bad maths. It was not planned to track the dollar, but to grow in value against the dollar by 47.2% per year according to page 9 of the whitepaper, but by 33.6% per year according to other parts of the same paper.
The paper describes an oracle that appears decentralised, but in fact, says nothing about decentralising the price feed. Instead it only smooths out differences in the MNX:USD prices that might arise from using, for example, LTC rather than BTC as an intermediary between MNX and USD. "In such a way we get unbiased cross-rate of MNX to USD" — but it is only as 'unbiased' as the source of the exchange rate data. Here's what it has to say about the data feed: "While assessing market conditions, MinexBank will invoke asset price at the main exchange, which is the exchange where MinexBank stores, buys and sells its assets. The selected
exchange will have to be the most secure of all currently in existence." This is pure centralisation, probably in the hopes of monopolising profits, and the people behind such a ham-fisted technical paper will certainly not be world-beaters at the complex and subtle art of cybersecurity.
It uses closed-source algorithms, which are not true to the spirit of decentralisation.
The actual stabilisation mechanism they describe is parking and intervention, similar to NuBits, but they were much less successful than NuBits, and I doubt this was ever anything more sincere than an excuse to have an ICO in June 2017, near the peak of the ICO shitstorm. They premined 50%, or 5.5 million, of these shitcoins (worth $25.8 million at current prices), and sold 2.7 million in the ICO (tokens were $6.18 in the ICO, though there were many discounts).
Kowala is another work-in-progress. When I read the whitepaper on their website, I found it full of holes. What was their blockchain or consensus mechanism? Were they using centralised price oracles? Perhaps worst of all, they described trading as a stability mechanism, with the market natuarally gravitating towards $1 because people know that $1 USD is the proper price for their coin. If this self-fulfilling prophesy worked, there wouldn't be failed stablecoins like NuBits. However, in preparing this article, the Kowala team kindly showed me a future version of their whitepaper, and they have addressed all these issues.
Regarding oracles, the whitepaper on the website goes into no more detail than saying they will "gather market information from endorsed sources". In fact, their logic around oracles is much better than that. They use polling, medians, and rewards for staying near the median, similar to a Schelling scheme.
As well as trying to stabilise price, they are ambitious about scaling, stating their "goal of processing 4,000+ transactions per second with a typical per-transaction processing time of one second". The unreleased whitepaper confirms that "to achieve ultra-fast transaction processing performance, we have also replaced the Ethereum consensus protocol with Konsensus, an implementation of PBFT derived from Tendermint." Tendermint is essentially a proof-of-stake consensus algorithm, where the more of the mining coin (mUSD) a node stakes, the greater a chance it has of being picked to propose a block, which others then validate. To keep this article focused on price-stabilisation, I won't get into Tendermint here. It is potentially quicker and more efficient than existing blockchains, but is less proven. Some good sources are the Cosmos whitepaper, and the official overview, which includes a diagram explaining the consensus mechanism.
Kowala's mechanism to increase the supply (and reduce the price) is to increase the block rewards given to miners. And to reduce supply (and increase the price), they impose added transaction fees, which get burned. I wonder if these measures will be drastic enough. They argue, "if just 15% per day (on average) of the kUSD coin supply is transacted via on-chain transactions, the burn amount will reduce the coin supply by over 8% in 30 days and over 41% in 180 days." But the assumption that 15% per day of a coin will be transacted is far from conservative. For comparison, 2.4% of DAI was transacted in the day before writing, and 5.8% of BitUSD (according to figures from Coinmarketcap). If there are not many transactions, transaction fees won't shrink the supply quickly enough to raise the price. Similarly, increased block reward might not be sufficient to drive the price down in the face of high market demand.
They do list one more stability mechanism, similar to 'parking' in the NuBits system or 'bonds' in Basis or Fragments. When the system wants to reduce supply (i.e. when the price is low), users can stake kUSD in a savings contract. Later, when the price is above target, this contract pays bonuses back to the address that sent the stake. This is smart and elegant, as it is a single mechanism that can decrease supply at times, and increase it at other times.
One decision does seem very bad: "each kCoin [meaning kUSD, kEUR, etc.] is implemented as a distinct, independent blockchain with its own tokens, smart contracts, mining community, etc." 51% attacks are a real threat to small blockchains.
Kowala has no show-stoppingly bad ideas, but they may be trying to reinvent the wheel a bit too much. I consider the economic mechanisms a bit less proven than most, but sometimes experiments work out very well. My chief doubts are that the small proof-of-stake blockchains will suffer 51% attacks, and that the fees and block rewards will simply be too weak as interventions. They claim to have done statistical modelling showing their system will maintain stability in the face of wild market conditions, but the details of that have not yet been made public. Time will tell.
Minor projects — Coin Payment Processor, RSRV, and Freedium
Coin Payment Processor should probably stop calling themselves 'CP Processor' before 4chan hears about it. They do have a whitepaper, but it is light on detail and heavy on bad English. I was able to glean that their coin will be created by locking up ETH (as DAI is). I am mentioning this project here in the interests of thoroughness, but everything about it looks messy.
RSRV have almost no information in public. I contacted them on June 16 but never heard a reply.
Reserve aren't saying for now how their protocol works, but they are starting to get some buzz, and have backing from Peter Thiel and Coinbase. They announced on June 20 that they had closed a $5 million seed round. Two sources have confirmed to me that the tech will be free of intellectual property restrictions when revealed, but for now they are being tight-lipped about how the protocol works.
A document I saw suggested it would use a market-making smart contract like Fragments: "The Reserve protocol automatically crowdfunds a smart contract called the Vault, which uses its funds to implement an automated exchange-rate peg between the Reserve token and the US dollar, or any other major currency. Users who contribute to this ongoing crowdfunding process are given Reserve Shares, which capture the value of the network if Reserve is adopted."
They say that Reserve uses a "hybrid design" where the coin is stabilised both by external collateral (as DAI is stabilised by Ether collateral) and their own volatile token (as nUSD is stabilised by HAV). "Reserve is a hybrid design that uses a Decentralized with external backing approach, supplemented by a Decentralized with self-generated backing approach for extra stability." If they are really combining the market-making of Fragments, the collateralised positions of DAI, and the simple dividend shares of Havven, this could be a very promising project.
Their docs say, "Reserve is different from Tether in that it is backed with assets on the blockchain, different from Maker in that it is backed through a standard reserve approach instead of by issued debt, and different from Basis in that it is not purely backed via its own future growth."
CEO Nevin Freeman told me on Telegram: "We are developing on Ethereum and watching other platforms to see how they develop. We think it makes sense to build a bridge between Ethereum and any other platform that gets widespread adoption so users can spend Reserve anywhere in the ecosystem."
They claim their currency will be "cheap and fast to transfer, easy to store, censorship resistant, private to transmit in limited amounts, and impractical for criminal use." It will be interesting to see how they intend to implement privacy for limited amounts, while disincentivising verboten use.
Unlike Preston Byrne, I don't think stability is inherently impossible. If it was possible for the dollar, it will be possible for some blockchain-based currency.
I am optimistic that one of the better projects in this article will achieve stability. I would not be surprised if Carbon, Reserve, Havven, DGX, or DAI become billion-dollar payment networks. DAI is doing great so far, with a sterling reputation, and plans to improve by diversifying collateral. If these particular projects fail, we will learn from them, and after more evolutionary pressure, we will arrive at stability through a decentralized, algorithmic mechanism.
Perhaps it will be a hybrid solution, combining a Schelling oracle, the variable fees and rewards that Kowala uses, the easy-to-use seigniorage shares of Havven, and perhaps the market-making reserve of Fragments. We would be getting pretty close to a perfect cypherpunk currency if such a system were built and privacy features like those of Monero were added. All my research didn't unearth such a project (if you know of one, get in touch with me).
Ametrano's paper on currency design (the one that inspired Fragments) argues that we shouldn't peg a stablecoin to the U.S. dollar, as its days of stability may be numbered. Ametrano's preferred benchmark is a basket of commodities. This idea has an appealing cosmopolitanism to it. Carbon's Medium, Basis' whitepaper (under the heading 'A Post-USD World'), and Fragments' roadmap all dream about evolving beyond the dollar.
But that comes later. For now, the U.S. dollar inspires confidence, because it backed by an immense amount of assets, some fissile. First, someone has to get a cryptocurrency to track the dollar and keep it as stable as the fifth labour of Hercules. Then privacy and scaling features can be added, and the economic logic can be applied to the gram of gold, the consumer price index, or Gross National Happiness.
Cover image from Maxpixel. Images from Wikimedia Commons, giphy.com , and the companies covered.
Never miss a thing and suscribe to our newsletter.
Conor O’Higgins has been a digital nomad for seven years. He learned the internet marketing game with Google and Facebook ads, but in the wake of the Snowden revelations, abandoned that to help marketing and communications for a more decentralized, encrypted internet. He is not a fan of current implementations of social media, but can be contacted through conorohiggins.com