May 2, 2022
The Crypto Turing Test
A simple benchmark to measure the everyday usability of cryptocurrency
Alan Turing famously proposed an imitation game. This was a thought experiment as to when artificial intelligence may achieve milestones with respect to humans and our ability to accept the intelligence of robots. The Turing Test, as it came to be known, posited a person sitting at a computer screen interacting using typed messages with something else. An AI would pass the Turing Test if it could convince the person that it was a person rather than a robot. Much has been written about whether the test is reasonable or would tell us anything. But the spirit of the test was the notion that people could interact with computers without realising it. In other words, those interactions could be seen as natural.
Cryptocurrencies have now been around for almost 13 years. They certainly have achieved substantial interest as well as proof that they were possible at all. But when we will know if they have “made it”?
Here I am going to propose a Crypto Turing Test to provide that benchmark. One thing this definition has got going for it is that no one (really, no one according to Google) has proposed a Crypto Turing Test before. That’s pretty surprising (to me at least) as Turing completeness is a big part of the crypto project.
Here is my suggestion for a Crypto Turing Test:
A cryptocurrency that can be used to purchase a latte at Starbucks without the merchant noticing that the purchaser is using cryptocurrency rather than another means of payment will be said to have passed the Crypto Turing Test.
I picked the purchase of a coffee as it was an ordinary transaction. Crypto people have been obsessed with being able to buy everyday things. The idea of the Turing Test is to get something to the point where people “don’t think about it.” When buying a coffee, you don’t want to think about it. So if you have to, that is a constraint and an indicator that a payment method isn’t there yet.
Why Starbucks? Now there are examples of independent, individual cafes accepting Bitcoin. But Starbucks, precisely because it is ubiquitous, would be a whole other matter. I could have easily picked Mcdonald's or Tim Hortons but this seemed as good as any.
The other feature of this test is that no cryptocurrency yet has passed it. This is not for want of trying. Starbucks announced attempts in 2017 (iPayYou), 2019 (Flexa), and 2021 (Bakkt). That last one is not really a proper attempt. Bakkt will allow you to load your Starbucks card from cryptocurrency. You can do the same thing using PayPal today. But as soon as you do, you have dollars and are really paying with dollars. So I am going to put that in the “not passed” category.
You might be thinking at this point: are you asking for Starbucks to post a BTC price for a Venti latte? No, I’m not. The problem there is that it places too much weight on a cryptocurrency having a stable value and I do not want that to be a direct driver of the test. The idea is that Starbucks could offer a minute by minute BTC denominated price if they chose and still pass the test. However, this is not to say that doesn’t matter in practice. It just shouldn’t be a deal-breaker.
What I have in mind is something similar to what we already do when we purchase a Starbucks coffee in another country. I’ll take my Canadian dollar-denominated debit card to the US and pay the US price for a latte. The amount is taken directly from my Canadian bank account with an exchange rate determined by my bank.
To pass the Crypto Turing Test, I would have to be able to do the same with, say, Bitcoin. I would take, say, my Coinbase wallet which has a card associated with it and tap that card at a Starbucks merchant payment console. Coinbase would charge me a conversion fee to fiat currency that Starbucks accepts and send that currency to Starbucks. Alternatively, Starbucks could accept the crypto directly and not charge a conversion fee per se but instead for a BTC price that incorporates that fee. Or Starbucks may start holding crypto and using it to pay for coffee beans, cups, baristas or whatever. The point is that to get to “without the merchant noticing,” I put crypto up to the same standards that most government-issued currency has achieved.
Actually, you will notice that is not a purely governmental standard. Banks offer you accounts with digital balances that are not base money. The reason that works is that their payment instruments — e.g., debit cards — now pass a version of the Crypto Turing Test. This was also true of checks in the past. That took some work. Back in the day, it wasn’t automatic that merchants would accept checks. And nowadays they may also not accept them. They had their day.
Note that one thing that would have to be resolved is the issue of what happens when you convert crypto to make a purchase in fiat currency. In some jurisdictions, it is possible that you have to keep track of that conversion for taxation purposes. I’m going to suggest that even if you successfully make a crypto payment for coffee, if you have to involve your accountant before April the next year, that transaction hasn’t passed the test.
Hasn’t this test been passed in El Salvador? Last year, El Salvador, which for years used US dollars as the main currency, announced that Bitcoin would be legal tender. As all merchants in the country were required to accept Bitcoin for payments, this seems like it was a fast track to passing the Crypto Turing Test — at least in El Salvador. The government even launched an app — the “Chivo Wallet” — to make this easier. They even gave people who used the app $30 in starting credit.
On the face of it, this looks like the Crypto Turing Test was passed. There were many transactions that took place using that wallet. But I’m going to move the goalposts just a little and recognise that the Crypto Turing Test is a market test. It can’t just be a “yes you could do this” but “would people want to do this.” This paper shows that, in El Salvador, Bitcoin hasn’t reached the purchase acceptance threshold. Only 20 percent made transactions after spending their $30. Very few did so intensively. The most common uses were not commercial transactions but, instead, to pay government taxes and remittances.
On the merchant side, despite the law, only 20 percent of businesses accepted Bitcoin and those were really large corporations. One suspects that they were doing so simply because of the law.
Given this, I am going to file the El Salvador experiment in the “not quite there” category. I am not going to say that the Crypto Turing Test cannot be passed if the government makes a cryptocurrency legal tender. That would be silly. But I am saying that the use “without thinking about it” notion will require more than that.
One final open question is whether passing the Crypto Turing Test requires both merchants and purchasers to hold balances of cryptocurrency. This is what happens with foreign exchange and certainly any other successful currency has achieved systemic success in that, at least in a location, people conduct their buying and selling (and holding) business in that currency. If people are holding cryptocurrency and then merchants are accepting a converted fiat currency, while in principle that can pass the Crypto Turing Test, in practice it may not be possible without system level adoption.
