On Wed, 28 May 2003, Einar Stefferud wrote: > Hello Dave Morris --- > > It would be helpful if you would explain how this payment system of > yours might actually work in real life. One model exists in the postal service operated 'by' each country. Stamps exist, procedures exist for sharing revenue or whatever when, say, a US Stamp is used to get mail delivered in Germany. Using the new protocols, my MUA would drop the mail in the local post box ... the SMTPnew server I use to send mail. That server would need to authenticate me and verify that I have money or credit available. That server would be resonsible to a local epostal clearing house for payment and would initiate delivery transfer of the post paid email. Could be that an electronic stamp token is included, generated with the servers private PKI key. It is possibly reasonable in the current network to insist that all mail delivery be point to point, but if not, intervening MXlike servers just move the bundle along. The final destination SMTPnew server verifies that the proper postage is attached, mostly a PKI decryption exercise. The stamp tokens are recorded in a database. Periodically, stamps would be bundled and sent to the epostoffice, perhaps along with cash, to get the local epostage meter recharged. Depending on the epostal cost structure, perhaps each received stamp would be worth 1/2 of a to send stamp. Based on local SMTPnew operator policy, end users might get 1/4 of a stamp credit for each received email. > Perhaps like TELEX worked before it died, with settlements between > the first posing ISP to the last receiving ISP, with "settlement" > payments spread across all ISPs in between. As implied above, one or a few organizations per nation would provide clearing services. > > Of course this leads to bilateral agreements among al the thousands of > ISPs, and collective agreements among the mass of global ISPs. No, only between the ISPs (where ISP means SMTPnew operator) and the chosen epostal service. And then tiered between epostal services. Millions of businesses today use postage meters, humble folks just by stamps. > Now, consider the cost of such arrangements, to cover the frictional > costs of just being in business, plus the required profit margins that > accrue to any such massive payment shuffling. Of course, there is a cost. And if you will, friction. A new PKI based trust system will also have added 'frictional' costs to create and support. My general approach provides a funding mechanism to pay those costs on a per use basis. > > Everyone here advocating payments do not seem to understand the > overhead costs of collecting and distributing the money. I think I understand the costs quite well. One of the protocol design challenges will to be minimize costs. Probably by appropriate tiering of responsiblity and granularity of transactions recharging meters. > Be careful of what you wish for! -- You just might get it! Hooray!