Token economics =============== Stablecoins Fully-collaeralized, redeemable coins issued by central entity AML/KYC compliant coins: need approval to hold/trade StrongholdUSD, AnchorUSD (Stellar); USDCoin, TrueUSD (Ethereum) Some coins require KYC to hold (stronghold), some to issue/redeem (TUSD) Former likely works better with FDIC insurance Latter easier. E.g., TUSD can only buy/redeem > $10,000 Less compliant coins harder for issuer to maintain banking relationships Tether USDT (Bitcoin omnilayer) In general depend on counterparty (issuer) for solvency, availability E.g., suspicions that Tether may not really be full-reserve Fear more centralized coins could be attacked/shut down Algorithmic - non-redeemable but more decentralized Growth-backed stablecoins (a.k.a. pyramid schemes) - basis Collateral-backed stablecoins - MakerDAO DAI Basis (basis.io) has three kinds of tokens: basis - supposed to be always worth $1 bonds - convert to basis or expire useless in 5 years shares - receive newly mined coins after bond holders paid off Oracle says if price is above or below $1 Above $1? Convert bonds to basis or pay to shares if no more bonds But don't sell bonds below $0.10 hard floor Below $1? Sell bonds (at a discount) to suck basis out of circulation Or use proceeds from shares and share sales to defend the peg Limitations Off-chain stabilization has at most 80% reserve funding, worse than Tether Vicious cycle if people lose confidence in coin Bonds are not fungible: each has unique queue position/expiration time Illiquid bonds will require higher interest rates As bond queue lengthens, bonds sell for less and less until $0.10 floor $0.10 floor can break peg, possibly for years until bond queue clears Ultimately, require sufficient growth to pay demanded bond premium MakerDAO Dai has two tokens: DAI - a stable coin pegged to $1 MKR - equity/governance token Collateralized Debt Position CDP = locked-up assets when issuing DAI Must lock up $1.50 in collateral to get $1 of DAI 150% is ongoing maintenance requirement, so probably lock up more The $1 is a loan to repay in DAI with 0.5% annual interest MKR holders can change interest rate and 150% collateral requirement MKR holders also elect oracles to specify asset prices If collateral dips below 150% loan, liquidate CDP with two auctions: Repay DAI by auctioning off new MKR tokens for DAI (that gets burned) Auction off collateral for MKR (that gets burned) Note must repay debt + interest ("stability fee") + 13% penalty Note interest and penalty reward MKR holders (like share buybacks) Remaining collateral returned to original CDP borrower Why would anyone want to issue DAI by taking on a CDP? To buy ETH on margin (can use recursively for more margin) Because you have some non-ETH asset that is illiquid Because you need cash but don't want to incur capital gains on your ETH Initially collateral is ETH, but intended for multi-collateral Would protect against big crypto crashes like this past week Each asset has own parameters (reserve ratio, interest rate, penalty) Global settlement = reset button held by committee elected by MKR holders Each DAI holder claims $1 of collateral, rest is returned to CDP borrowers Used to upgrade contract, deal with vulnerabilities or "irrational" markets What causes DAI's value to stay around $1? Global settlement (the nuclear option) liquidates DAI at exactly $1 Arbitrage in anticipation of eventual global settlement Both DAI buyers and sellers and CDP borrowers will drive price to $1 Limitations What if not enough CDPs compared to stablecoin demand? Will they really get diversified collateral, not just crypto? Augur (https://github.com/AugurProject/whitepaper): REP - new token for creating markets, reporting, governance, NOT trading ETH - used for actually trading predictions To create a market: - Define event time, designated reporter, outcomes, resolution source - Set creator fee - Post validity bond in ETH (forfeited if real outcome not in set) - Post no-show bond in REP (forfeited to other reporter if designated bails) Trading predictions, e.g., between two outcomes A & B: Shares created in complete sets: 1 ETH <-> 1 A + 1 B Order matching creates/destroys shares and matches buyers to sellers Reporting happens in 7-day periods (fee windows) Reporters purchase participation tokens using REP tokens At end of window, reclaim REP plus get share of reporting fees Incentivizes REP holders to know how to use system, check in once/week 7 phases of reporting (see Figure 2 of paper): 1. Pre-reporting 2. Designated reporting - 3 days for designated reporter to report Report on time? market creator gets back no-show bond, go to phase 4 Not on time? market creator forfeits no-show bond, go to phase 3 3. Open reporting - entered if designated reporter doesn't report on time Anyone can report, but first reporter gets no-show bond as stake So if their report agrees with outcome, will get REP tokens Note don't need to stake your own REP to be first reporter! 4. Waiting for the next fee window - after 2 or 3 5. Dispute round - 7 days to disagree with tentative outcome from 2 or 3 To dispute, create a *dispute bond* Stake REP on some outcome w other than tentative outcome Can crowdsource dispute bond (many people staking) Is bond + stake on w >= 2 * stake on other outcomes? Success Is dispute bond >= 2.5% of all REP? If so, go to phase 6 Else, refund dispute bond other than w Add bond w to w's stake and go to phase 4 No success? return dispute bonds, go to phase 7 6. Fork - freeze parent, create a child augur universe for each outcome Halts new prediction markets for 60 days REP holders can move REP one way to one of the child universe Initial reporting stake can only move to corresponding universe One-way move, but 5% bonus if you move within 60 days Idea: REP in bogus universes should be worthless 7. Finalized - can settle shares directly with the market Note all tokens and bonds are tradable Don't have to wait weeks to get paid if can trade shares earlier What are the risks/limitations of Augur? Parasitic markets - betting on the outcome of Augur to avoid Augur fees If larger than Augur, reduces integrity of Augur Huge bet (bigger than market cap of REP) could incentivize forks Though could also increase the price of REP, providing some protection Resolution source could be hacked Self-referential predictions: Will a designated reporter fail to report? Forks are highly disruptive Ambiguous markets: Tradesports bet on DPRK missile landing outside airspace Missile did land outside airspace, but US DoD didn't confirm the event Contract had specified DoD confirmation, so those betting on missile lost Token Curated Registries (TCR) to maintain list of best X for various X E.g., best 10 restaurants, CS departments, adchain.com Three parties: Candidates - people who want to be included in the list Consumers - don't necessarily have tokens or want to use blockchain But if they make use of the list, it gives the list value Token holders - incentivized to make list useful to consumers So create a token for the list, and play the following game A candidate who wants to be listed deposits a token with application TCR has MIN_DEPOSIT parameter specifying how much it costs to be listed If accepted, can get refund at any time by withdrawing from list But token holders can challenge an application If candidate rejected, loses its deposit, which is given to challengers So revenues come from candidates at the margin Enough change to be accepted to apply, but still get rejected Why not challenge everyone? Empty list is useless, so token will have no value Token holder stakes deposit of MIN_DEPOSIT to challenge application or listing Upon successful challenge, token holders vote on candidate Votes are weighted by token holdings Votes held in commit-reveal fashion Vote in rational way because can't see how others voting Uses Partial-lock commit-reveal (PLCR) voting Commit stage: Submit commitment to vote {tokens, H(r, vote)} Locks up tokens so can't cash them out (and hence double-vote) But can use the tokens to vote in parallel on a different question Reveal stage: Everyone publishes their (vote, r), so can count votes If you don't reveal by deadline, your vote doesn't count However, can still rescue tokens Note: no penalty for voting wrong way Whoever loses vote (candidate or challenger) forfeits deposit Winner gets DISPENSATION_PCT (e.g., 50%) of forfeited funds E.g., at 50%, must have 67% certainty of winning challenge (1/3 chance of 100% loss, 2/3 chance of 50% gain) Remaining forfeited tokens distributed to majority voters by token weight Minority voters neither lose nor gain tokens What if MIN_DEPOSIT increases, old listing's deposit below gas of challenge? Special case lets anyone instantly remove listings below MIN_DEPOSIT Both challenger and candidate get their deposit back Why not just use ETH to curate registry? Kills incentive for list quality What are some limitations of TCR? What if high quality candidates don't place a deposit? E.g., Stanford won't pay to be included in "top CS department" lists Maybe token holders stake Stanford to make list useful? Related problem: how to bootstrap a list from empty Maybe air-drop tokens to potential consumers and candidates What if very few token holders vote (cost-effective bad listings)? What if people vote randomly on questions they know nothing about? Reward for winning vote, no penalty for losing vote Worse than random is "memeing"--e.g., vote low bit of SHA256(candidate) If many people do this, will always be in majority But will destroy value of token holdings