Anonymous points out problems with managing and using digital cash:

(1) customer reads ad about cool net.service (a) they contact directly (but this ruins privacy) (b) they contact independent distributor of PGP key and anon-remail address generating software. (but how does customer trust _them_?) (2) vendor sends key & address generators (via e-mail or floppy), and physical-mail-security instructions (3) customer sends in money order (from mail drop or without return address!) along with chosen anon e-mail address and public key. (4) vendor sets up account and e-mails the d-cash. (5) we still need a physical mail drop or bearer bank account for withdrawals, refunds, etc. of physical cash.

It's true that this is a lot of steps. This is one reason why we should push to make anonymous mail easy to use. If you start with an infra- structure where you can communicate securely and anonymously across the net, with return messages that don't reveal your true identity, you have a good start on steps 1 and 2 above. Another simplification would be to split the users of the digital cash into customers and vendors, along the lines of the triangle diagram Tim suggested. In my experience, I make many withdrawals from my bank, but often only two deposits a month as my paycheck comes in. Applying this to the digital bank model, customers would mostly make digital cash with- drawals to buy things on the net, with occasional physical cash deposits to keep their balance up. Customers would thus primarily turn real cash into digital cash which they send to vendors; the vendors then turn the digital cash back into real cash. Customers can remain anonymous even without physical mail drops, while vendors have less anonymity. In this model, a customer sees some new software being sold on the net. He normally keeps enough digital cash on hand for such spending, so he sends the cash to the seller, including one of his standard anonymous return addresses for the return software. If this lowered his stock of digicash below the amount he likes to keep around, he sends another check to the bank and gets another batch of digicash. (This would be analogous to carrying cash in your wallet, and when it gets low you stop by the ATM and withdraw more. How often do you find yourself depositing cash back into the ATM? I suspect customers of the digital cash bank would similarly not need to turn their digicash back into real dollars very often.) With an infrastructure like this, using digital cash does not have to be complicated.

Also, the issue of which parts of these schemes are *legal* is critical, but being completely overlooked.

I once posted some excerpts from the Code of Federal Regulations involving the tax requirements of barter agencies. These are organizations in which their members exchange their labor without the use of regular cash. Often they use some scrip as a substitute for cash, or they may just keep records in an accounting system. It appeared to me that virtually any form of digital cash would fall under this definition. Barter agencies are not illegal, but there are many rules about reporting transactions and members. It would definately not be possible under the current tax code for a barter agency to have anonymous members. Therefore it looks like anonymous digital cash would not be legal in the U.S. at this time. I don't know about Duncan's suggestion to use an offshore bank. Paul Robichaux wrote, regarding the NetCash proposal:

Collusion between the service provider & the currency house can produce a record of exactly what I bought, but I don't know that blinding can do much better.

Actually, blinding can do better. Collusion between the bank and the vendor can not break customer anonymity in a cash system using Chaum's blinding protocols. This is one of the things I found so surprising about the NetCash proposal. I am surprised they dare to call their idea an implementation of digital cash when it does not even provide this bare minimum of customer anonymity. This anonymity is why we call it "cash", as distinguished from other forms of money. The NetCash proposal is more like cashier's checks. Bill Stewart writes:

If you're sending non-accountable-by-sender cash, you need some way to get a receipt. If you're mailing a check or digicash, you have a way to repudiate the transaction or at least make a claim against them, or if you can go in to the bank in person for the transactions with cash.

This is a good point, and is another reason why blinding is so important. If you don't mind it being known in general terms that you are a customer of the bank, you can send a check with instructions to turn it into digital cash to be sent to your email address. There will still be no way that the bank can figure out when or where you spend that digital cash. And they are no more likely to just cash your check and pocket the money than any other mail order business would be. Hal Finney hfinney@shell.portal.com