I have only one small quibble with Perry's recent comments.
Why not just have people use a normal currency? Seems like you have some bizarre idea that the poor won't be able to afford the real currency, which is silly.
The question is not a bald one of access or no access, but one of quantity obtainable in aggregate and timeliness of such accrual. Experiments with LETS systems have shown that local economic activity does improve in depressed cities when a barter currency is introduced to supplement a paucity of the nominal national currency. The city I seem to remember is Manchester, England. LETS (I forget the acronym expansion) is a barter system with a virtual fiat currency. Originally it used just a ledger; later, PC's were used to keep the books. The currency was zero-sum; all accounts added to zero. Reputation was provided by making all aggregate balances public to the members of the system; you could decide not to provide services to anybody, particularly if they had a large negative balance. Another example of how a dearth of transfer instruments affects an economy was 16th/17th century Venice. Coin hoarding did become somewhat of a problem, and it affected the speed at which business could be done. This era was that of the rise of 'book-gold', or in modern parlance, fractional reserves. Perry is certainly correct that any commodity can be used as a backing. Recall, however, that promises are a commodity like any other. This is the unification of fiat currencies and gold currencies. It is also a basis for understanding that multiply backed currencies can and do coexist stably. Promises are not as fungible as gold is of course, which is one reason that LETS systems do not scale well, since the characteristic effort and communication needed to evaluate the worth of such a promise (even an averaged promise as in a LETS system) is far greater. One can understand the rise of options markets as an effort to increase the fungibility of the option promise, given that these markets are not merely communications systems but also have some capacities as guarantors and insurers (broadly construed). Digital money has two characteristics that pertain to these issues: denomination size and access. The smallest denominable amount is not limited in any virtual system (bank books included), whereas when in a strict commodity system the unit of transactability may be too high and cannot be infinitely subdivided. For example, gold Spanish doubloons (from which the English 'dollar' is a corruption) were too large for many transactions, so people made them small by cutting them into eight 'bits', whence the equation of 'two bits' with a quarter. But gold is not infinitely subdivisible, but representations are. The question of access arises as well. Just as a LETS system is a very econopunk system, digital money can be issued by any one person or any group. If no other backing is available, they can back it with their own time and talent. When these currencies can be easily traded with other currencies, the problem of access to a more dominant currency is alleviated. These promises, being limited to a particular geographic locale, are not fungible, but then lack of fungibility does not so much prevent exchange as present some market-priced impediment to it. Eric