Cash

• Universal acceptance across a very wide geopolitical area, such as an entire country or an entire continent.
• Assured anonymity, such that payments are, if not theoretically impossible to track from point to point, at least infeasible enough in practice that it might as well be theoretically impossible.
• Ease of use.

Many casual definitions of cash seem to use the phrase "anonymous cash", which is similar to "free gift" -- the qualifier shouldn't be necessary, but the very term has been polluted by entities offering half-solutions that lack one or more of the desireable features of real cash. When we say cash, we mean (at least) all the properties above.

We are not necessarily talking about a micropayments system, e.g., one that is efficient for very small transactions. Such a system could be very useful for charging by-the-page or even by-the-byte for copyrighted content, but tends to be very difficult to implement in practice due to the overhead of the system used to manage it. Our problems are difficult enough without trying to be efficient enough for micropayments as well. For our purposes, we'll assume that any system that's efficient enough to make a one-dollar payment reasonable is efficient enough. Maybe we'll come up with a system that allows people to charge a millicent for something instead, but we may not be so lucky that we can improve the system by five orders of magnitude.

People have been trying to bring cash to the net for years, and the prospects don't look much better now than they did years ago. Efforts have been stymied by at least four major factors:

• Cryptographic export restrictions from various countries (still enough in effect in the US to hinder commercial deployment, and in general to create interoperability problems in almost any system likely to be deployed worldwide).
• Intellectual property fights leading to very slow deployment.
• Governmental resistance, due mostly to fears of illegal activity, money laundering, and tax evasion. (This latter seems especially peculiar, considering that income tax evasion would require that the employer also fail to keep a paper trail, and that sales tax evasion is already complicated by jurisdictional issues even in meatspace.)
• Lack of consumer demand, because most consumers either don't understand how little privacy they have with, e.g., credit-card systems, or don't care, or have been convinced that the effort required to fix the situation is too great.

And yet real cash has numerous advantages:

• Privacy. Many transactions that could be made on the net today are not made because consumers fear that their purchasing habits may be disseminated to unknown third parties. (This is particularly an issue when purchasing bits -- after all, since nothing physical is being shipped anywhere, there is no need, in theory, that the consumer need disclose any identifying information.)
• Bounded liability. Using something like a credit- or debit-card has its risks: the card number may be later misused, either by the other party to the transaction, or by some third party. This is a risk in meatspace transactions as well as online, of course, but many people worry more about it online regardless. Even though most credit card companies will refund all of a disputed amount (without even the $50 deductable legally required in the US), many people are still leery of disclosing credit card information online. Cash, on the other hand, is safer -- while the merchant may pocket the cash and fail to deliver the goods, at least that is all that might happen.
• Everyone is a merchant. Credit-card companies are notoriously reluctant to let just anyone accept a charge card. Yet the peer-to-peer nature of the network makes it natural that two ordinary people might wish to exchange cash without having to go through some troublesome middleman; this is done in meatspace every day.

A modest proposal

For example, imagine this scenario: You want to buy something on the net, but you don't want to tell a possibly unbounded number of people what you're buying. You walk into the local 7-11 and plunk down $10 in paper money. In return, you get a card that looks approximately like a credit card, except that it's really cheaply made -- probably a laminated piece of paper like a subway Metro card. It's worth $10 (or maybe $9.95, and the credit card company gets the nickel). You then go buy a book online, and supply the card number for payment, just typing it into the form on the web page like it was any random credit card. If you can't use up the last 75 cents on the card buying things, you bring it back to 7-11 the next time you buy a card, and those 75 cents are added to some more paper money to refill a new card.

This solves almost all of the above problems, immediately. It does not require any sort of cryptographic infrastructure that isn't already deployed; instead, it uses the preexisting credit-card database structure. It is definitely anonymous, because you paid for the card using something else anonymous, namely paper money, and didn't need to supply any other identification in the transaction, because paper money is honored everywhere without additional ID. (It is possible that two successive purchases on the card could be correlated, so it may be wise to change cards frequently to break such connections; similarly, if there is any possibility that refilling a new card from an old one might make an entry somewhere that links the two cards, perhaps it's better to abandon the last 75 cents on the card -- or hand it off to a stranger on the street in return for 3 quarters.)

What are the problems with this scenario? There are several:

• Credit-card companies want to do data mining. This anonymous system deprives them of one of the most valuable assets of the credit-card business, namely the detailed profiling of every single thing you buy with the card, and the subsequent sale of that information to eager buyers.
• It requires physical infrastructure, in the form of 7-11's (or whatever) that are willing to hand out cards. Obviously, you wouldn't want to get one of these cards in the mail, in return for a check or a credit-card payment -- you could be virtually assured that such a card is not anonymous and is being tracked to you by the issuing entity.
• It's not really peer-to-peer. It has the same centralized-acceptance problems that credit-cards do; there's no easy way to convince somebody who isn't online with the issuing agency what the card is worth, so it can't easily be traded hand-to-hand the way paper money can.

How can we possibly do this? That's something we need to figure out at the workshop.