Airswap: a centralized exchange tragedy

the follies of attempting to design and ICO a controllable “decentralized” exchange experience

This week, a token exchange called Airswap will be completing a ~$32M token sale (42,000 ETH cap with over $20M already secured in presale). Airswap is a ConsenSys backed team building both a tokenized exchange and smart contract based token trade protocol system called Swap. If you asked Airswap what their product is, they might tell you its purpose is to reduce friction and increase efficiency in the token trading process through the services provided by the AirSwap Indexer and Oracle. However, this post shows in reality they are building an overly complex centralized exchange on top of Ethereum.

Lets introduce the parts of the Airswap Token (AST) ecosystem:

Oracle(s)

Airswap has committed to developing a number of Oracles for several inputs whose selection and values are determined by AST holders. The Swap whitepaper barely addresses Oracle design or implementation and mostly treats it as a black-box concept while proposing top level API endpoints. Recent blog posts do less to introduce the new Oracle types not in the whitepaper (token whitelist and AST lock amount oracles). Decentralized oracles have not been successfully developed or tested at scale so more than likely the Airswap team will end up using centralized trusted Oracles they can control. Even if AST is used for voting mechanisms the Airswap team is only offering 13% of tokens to whitelisted ICO participants and 40% of tokens will likely be released to the Airswap team in Q1 2018 (~6 months from ICO).

Indexer(s)

The Indexer is an off-chain service used for makers and takers to find each other, use the price oracle for guidance, negotiate, then settle on a price peer-to-peer. To place an “intent to trade” (order) in the Indexer a maker must hold and lock AST tokens (amount determined by an oracle) for 7 days. Since the “lock amount” oracle is centralized as explained above and the AST token is only divisible to 4 decimals it is possible large AST holders (or Airswap) could make it prohibitively expensive for new makers to enter the system via raising the lock amount and thereby forcing takers into prices provided by the price oracle (selection also controlled by large AST holders). The Swap whitepaper proposes an API that only finds and suggests “intents” but provides no transparency into all intents listed meaning an Indexer can trivially front-run either party.

It is worth noting that due to only having 4 decimals even if the AST token appreciates its lowest possible denomination is only 0.0001 (half as much precision as Bitcoin) — meaning if in several years AST’s value rivals ETH’s current value (~$300) then owning the smallest piece of the AST network will cost $0.03. While this might sound small now, a price point this high stifles network effects as new users coming onto the platform for market making and/or governance must pay increasingly higher $ or ETH amounts to acquire AST while large volume AST holders have no clear incentive to bring on new makers as they already control the oracle governance (lock amounts and prices).

The Airswap Token (AST)

The Airswap Token (AST) is proposed as a utility token with the function of allowing a holder to post trade intents to the Indexer, by locking a set amount of tokens for 7 days, and “voting” on oracles for governance. As explained in the AST ICO details post, less than 13% (the presale was “ heavily oversubscribed” and 70% of total ICO allocated tokens were sold) of the total token supply is publicly available for sale meaning >87% of tokens are maintained by the Airswap team (60%), presale investors (>15%), beta testers (no details — 2%) and advisors (10%). Yes you are reading that right…presale investors, likely mostly the VC firms listed on the Airswap website and ConsenSys, have received a larger stake of the “decentralized exchange protocol” voting rights than those being sold to the public (actual users). For a team that has already received funding (and is backed by ConsenSys) holding this portion of the total token supply in a decentralized oracle-driven voting protocol feels disturbingly centralized.

To add insult to injury the Airswap team has included a transfer killswitch to their AST token contract in the form of a Pausable smart contract such that the token contract owner (Airswap) can stop the ability of any Ethereum address being able to transfer AST tokens with a single contract transaction, forever. Put another way, if pressured by regulatory or other market forces, Airswap is fully capable of completely halting all transfer of their AST tokens between Ethereum addresses, exchanges, and effectively shutting down all Indexers and Oracles at will with a single Ethereum transaction.

Conclusion

Airswap claims they are building a better exchange by saying their design “sidesteps front-running and race conditions” while offering a “globally accessible, friction-free value network”. While these sound promising in theory (and in interviews), the Airswap team has clearly done little in the architecture of their token sale or oracle system(s) to prove these claims. Instead, they have piggy-backed off the ConsenSys marketing arm to create a deceptively over hyped centralized smart contract exchange with the potential and ability to shut down as such at any given time.

UPDATE #1 (10/9): Phil Daian (Airswap advisor) responded in this Reddit thread explaining the pause function is used as a bootstrapping feature and Airswap intends to nullify that contract owner (no timeline provided). Since Airswap is using their core technology for the ICO there is no reason (that I am aware of) for keeping this functionality in place once the token sale is concluded and AST tokens are tradable (Oct 17). Other points in this post have not been addressed by the team or elsewhere.