Thoughts on GeminiCoin

Today we learn that Gemini, the exchange owned and operated by the Winklevoss Brothers, has launched a “stablecoin:”

https://twitter.com/prestonjbyrne/status/1039145312630329344

Some thoughts.

1. This isn’t a stablecoin

First, this isn’t an “algorithmic stablecoin” à la Basis, Reserve, MakerDAO, et al., although those schemes will no doubt point to the Winklevoss solution as a vindication of their (deeply flawed) approach to dollar parity.

A stablecoin is legit if it is convertible. If there is no convertibility, there can be no parity. Viewed thus the algorithmic stablecoins are basically accidental Ponzi schemes where you need to introduce new buyers constantly in order for the scheme to perpetuate itself, otherwise it collapses.

The Winklevoss scheme, by contrast, is a coupon which is redeemable 1:1 for an actual deposit held with a trust company or an actual bank. In this way the Gemini “stablecoin” is in fact quite similar to existing interbank solutions such as Clearmatics’ Utility Settlement Coin, save that in Gemini’s case the stablecoin is publicly viewable by anyone who wishes to inspect the smart contract and Clearmatics’ system is presumably run over a VPN.

2. KYC?

Big question mark how this is going to be done although having seen how the Winklevoss brothers run their businesses over the past couple of years I am fairly sure they got their bases covered. I’m going to be lazy and just quote my own tweet here rather than explore this issue in detail.

tl;dr the Bank Secrecy Act and the Wu-Tang Clan have one thing in common, and that is that both of them ain’t nothing to f*** with.

3. HybridChains

The most interesting thing about this is structural: Gemini has built a public, permissioned application that effectively hybridizes the separate approaches of the coins on the one hand and the enterprise blockchain applications that have been developed on the other, to bring something to the marketplace that is unique and genuinely new.

I don’t count Tether as unique and new because of the huge question mark raised by public whistleblowers over their regulatory compliance. To do this correctly you have to do all of the regulatory legwork. Gemini appears to have done this. (…so has Paxos as well, it seems.)

Gemini’s experiment with a permissioned smart contract on a public blockchain shows that there is thought by them to be considerable value in leveraging the security and auditability properties possessed by blockchains without needing to pair them with an incentive mechanism such as a coin. Few banks would expose their own internal ledgers to the public internet. Gemini has no fear of doing this as long as this ledger is on a blockchain, meaning that tampering is very hard and they can stand up the application on the public internet, public-facing, with little fear of, say, a SQL injection attack or the like altering the records.

It logically follows that any “permissioned Ethereum” implementation a la Hyperledger Burrow or BlockApps has the same security and auditability properties running in a public setting as the main public Ethereum chain. So while Gemini is running this stuff on an ERC20 now, as Gemini wants to increase the complexity of the applications it is running – say, to run a financial instrument over the public internet – it may choose to simply deploy its own blockchains – permissioned, but publicly viewable – in future.

“Why not just use the banking system?” I hear you say. Well, the answer may not immediately be apparent because the financial products to which a GeminiCoin might be best suited do not yet exist: chiefly, fully-automatic products that need to be able to interact with suppliers and retail investors with zero supervision from a centralized authority such as a bank or a registrar, but which are readily and automatically auditable on request.

Imagine for a moment that we live in a world where asset ledgers run on their own chains that can be viewed by anybody (and transacted with by anyone), and let’s say you want to securitize a fleet of self-driving, battery-powered Amazon delivery trucks, not by pooling the trucks but by selling ownership in individual trucks that service particular neighborhoods, so people could invest in the Amazon infrastructure that they use on a daily basis.

Here, it might be possible for a holder of shares in a particular truck to run nodes on both the truck’s chain (TruckChain) and the GeminiCoin chain. The truck’s “administrator” would also run a node on TruckChain and the GeminiCoin chain. If the holder of a share in the truck (Alice) wished to sell the shares to Bob, Alice would simply post a transaction to the truck’s chain, or more properly an escrow contract, stating that the administrator of TruckChain was authorized to transfer Alice’s shares to the next user who who paid Alice a particular amount to Alice’s account on GeminiCoin chain. If Bob, who is also running a node of GeminiCoin, then does so and provides also a public key address of a keypair to be registered on TruckChain in the transaction to Alice on GeminiCoin,  the administrator’s node would check the proof and transfer the ownership of the shares in the truck to Bob.

There’s no need for this to be confined to GeminiCoin, either. Any liquidity/coin provider could provide this service, so long as the admin of TruckChain is willing to read the chain and Alice is willing to accept funds from it.

Completing such a transaction would happen automatically and in a matter of seconds with zero human involvement apart from the provision of payment and address details by Bob as the buyer of the shares. The low infrastructure overhead from automating the entire process is why you can securitize the truck individually instead of pooling, say, 30,000 trucks together and creating a single bond from the aggregate cashflows arising from them. Rather than tranching up the risk, as is done today and which can operate to conceal the real risk of the investment under the right circumstances, exposure here might (e.g.) be backed by Amazon’s receivables and backed out by a bog-standard auto insurance policy.

More transparent financial system ahoy.

That said. It’s good to see people figuring out that permissioning == more functionality. Many people have given me a lot of shit for many years over “permissioned blockchains” (see e.g. this WSJ piece from 2015), which I’m pretty sure Monax invented in 2014, on the basis that “without the coin providing incentives for users to secure the network, there is no purpose to the chain.”

Or this more recent quip from Fred Wilson:

Well, Fred, I can only suggest you chew on the implications of Geminicoin, publicly-auditable trust accounts with open APIs, and fully automatic securitization of self-driving truck fleets for a little bit. After you do that, then I’d be happy to hear from you why you think permissioned blockchains don’t have a future.

I certainly think they do. But then, we’ve known that for awhile, haven’t we.

4. Marmot pictures

To close, here are some pictures of marmots.