INTERNET-DRAFT Donald E. Eastlake 3rd Updates RFC 821, 854, 959 CyberCash Expires: 6 January 1996 7 July 1995 An Application Level Internet Payment Syntax -- ----------- ----- -------- ------- ------ Status of This Document This draft, file name draft-eastlake-internet-payment-00.txt, is intended to be become one or more Proposed Standard RFCs. Distribution of this document is unlimited. Comments should be sent to the author . This document is an Internet-Draft. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months. Internet-Drafts may be updated, replaced, or obsoleted by other documents at any time. It is not appropriate to use Internet- Drafts as reference material or to cite them other than as a ``working draft'' or ``work in progress.'' To learn the current status of any Internet-Draft, please check the 1id-abstracts.txt listing contained in the Internet-Drafts Shadow Directories on ds.internic.net, nic.nordu.net, ftp.isi.edu, munnari.oz.au, or ftp.is.co.za. Eastlake [Page 1] INTERNET-DRAFT Internet Payment Framework 7 July 1995 Abstract The Internet is becoming an increasingly commercial arena in which information is being bought and sold and payments are rendered for goods and services to be delivered outside of the Internet. Thus far, the protocols and format used for such payments have been ad hoc and proprietary. This draft proposes a uniform application level syntax to support such commerce. Specific specifications are given for how this syntax fits into the World Wide Web, FTP, Telnet, and SMTP. Acknowledgements The contributions of the following persons to this draft are gratefully acknowledged: Brian Boesch Phillip Hallam-Baker David S. Raggett . Eastlake [Page 2] INTERNET-DRAFT Internet Payment Framework 7 July 1995 Table of Contents Status of This Document....................................1 Abstract...................................................2 Acknowledgements...........................................2 Table of Contents..........................................3 1. Introductions...........................................4 1.1 Applications Level Applicability.......................4 1.2 Overview of this document..............................4 2. Price Tags..............................................6 2.1 Prices.................................................6 2.2 Payment System Strings.................................6 2.3 Price Tags.............................................7 3. Payments and Receipts...................................9 4. Use in the World Wide Web..............................11 4.1 Web Browser User Interface............................11 4.2 Anchor Embedded Costs.................................11 4.3 Page Header Price Tag.................................12 4.4 HTML Form Price Tags..................................13 4.5 Web Payments and Receipts.............................13 4.6 Payment Required Error................................14 4.7 Web Proxies...........................................15 5. Use in File Transfer Protocol..........................16 6. Use in Telnet..........................................17 7. Use in Simple Message Transfer Protocol................18 8. Protocols to Which Payment is not Applicable...........20 8.1 The Domain Name System................................20 8.2 The Finger Service....................................20 8.3 The Auth Service......................................20 8.4 The ECHO, DISCARD, and CHARGEN Services...............21 9. Security Considerations................................22 References................................................23 Author's Address..........................................23 Expiration and File Name..................................23 Appendix A: Initial Payment System Names..................24 Appendix B: Simplified BNF................................25 Eastlake [Page 3] INTERNET-DRAFT Internet Payment Framework 7 July 1995 1. Introductions Applications level Internet commerce requires a means to (1) indicate prices and acceptable methods of payment, (2) tender payment, and (3) issue a receipt acknowledging payment or indicate if payment fails. This document specifies a character string syntax for these three items. 1.1 Applications Level Applicability Payment facilities could be applied at a number of levels. This specification is concerned only with applications level data items and services. It does not in any way concern itself with network level packets or quality of service nor does it concern itself directly with transport level connections or quantity or quality of service except as these transport level measures impact application services. This proposed syntax is concerned with such matters as access to a web page page, storage of a file, initiation of a telnet session, or conducting an extensive WAIS search. These are generally user visible and meaningful data objects or tasks. Within most legal systems, the owners of such data objects and/or the owners of the facilities used to present such objects or perform such tasks are frequently entitled to require some recompense if they choose to do so. This document does not concern itself with the morality of such laws or requirements but merely provides a syntax whereby cooperating entities may speak at that level about prices and payments. There is no requirement that the "currencies" used with this syntax be the usually recognized national or international currencies. For example, some transactions could be denominated in frequent flyer miles or other private artificial unit. 1.2 Overview of this document Sections 2 and 3 below define a basic syntactic framework for price tags, payments, and receipts. Sections 4 through 7 specify a standard for inclusion of these items in transactions for the World Wide Web (HTTP/HTML), File Transfer Protocol (FTP), Telnet, and the Simple Message Transfer Protocol (SMTP). Eastlake [Page 4] INTERNET-DRAFT Internet Payment Framework 7 July 1995 Section 8 lists some protocols to which application level payment systems should not be applied. Section 9 discusses security considerations. Appendix A is an initial list of payment systems that are or are planned to be usable via this syntax. Appendix B gives a semi-formal BNF-like description of the syntax. Eastlake [Page 5] INTERNET-DRAFT Internet Payment Framework 7 July 1995 2. Price Tags A uniform price tag format is needed to indicate when payment is due, how much, and what methods are acceptable by the seller. Such a price tag must include the specification of one or more acceptable payment systems (with a provision for payment system specific information where needed) and will almost always include one or more prices. Sections 2.1 and 2.2 below describe prices and payment system strings and Section 2.3 assembles these for complete price tags. 2.1 Prices Prices are encoded as character strings consisting of a number followed by a currency code. These currency codes are the three letter ISO 4217 codes, Internet Assigned Number Authority (IANA) registered four letter or longer currency codes, or private use currency codes starting with "x-". (ISO 4217 code normally consist of the two letter country code followed by a letter mnemonic for the major unit of currency.) Currency codes are case insensitive. One and two letter codes appearing in the place of a country code are reserved for future use. The number preceding the currency designation is the quantity of major units of that currency. It may optionally have a decimal point and additional decimal fraction digits for specifying minor unit or fractions of units. (Some currencies, such as US Dollars or British Pounds have minor units (cents and pennies). Others, such as Japanese Yen and Italian Lira do not.) Fractional digits may continue indefinitely after the decimal point but payment systems may define how many digits they utilize. Somes examples: 2.34gbp, 79ALL, 1.23456cad, 0.125usd which signify 2 pounds and 34 pence sterling, seventy nine Albanian Leks, one dollar and 23 and 456 thousandths cents Canadian, and one eighth of a US dollar. 2.2 Payment System Strings Payment system strings consist of the payment system name, a colon, and any payment system specific information (such as what account Eastlake [Page 6] INTERNET-DRAFT Internet Payment Framework 7 July 1995 within that payment system the payment should be made payable to). Payment system names are case insensitive, four or more letters long, and are indicated by a terminating colon. One to three letter codes occurring in the place of payment system names are reserved for future use. Payment system specific information must be encoded so that it contains no internal spaces or unusual characters as described in Appendix B. It is up to the named payment system to encode and decode any information it requires so as to fit within this syntax. Use of the base64 encoding defined in RFC 1521 is recommended. The payment system specific information, if any, appears immediately after the payment system name and colon and is terminated by white space or the end of the price tag character string. A registry of payment system names is maintained by IANA. Initial payment system names are listed in Appendix A. For experimental use pursuant to bilateral agreement between the parties involved payment system name starting with "x-" may be used. No names of this form will be officially registered. 2.3 Price Tags A complete price tag consists of a string of white space separated prices and payment system strings. There must be at least one payment system string present. Normally there will also be at least one price. However, there are circumstances under which the cost of a service in highly unpredictable and the seller is, in effect, requesting a payment system and account to which they can attempt to charge indefinite amounts. Under these circumstances, it is recommended that a price be listed which is a reasonable ceiling such that if costs exceed that, the seller which have to present another price tag; however, it is permitted to omit the price and list only a payment system in a price tag. Payment systems SHOULD provide a means for a limited amount of arbitrary seller information to be included in the payment system specific part of a price tag and be returned to the seller within a payment message based on that price tag. A price appearing after a payment system string applies only to that system. Putting a price before the first payment system specific information makes that price a default for every payment system specified. The default can be overridden by specifying a different amount for that currency after a particular payment system. Eastlake [Page 7] INTERNET-DRAFT Internet Payment Framework 7 July 1995 For example: 33.45all foocash:xxxx 22eTb barsys:yyyy 9.999ghC indicates that payment of twenty two Ethiopian Birrs via the foocash payment system or 9.999 Ghanaian Cedis via the barsys system or 33.45 Albania Leks via either system is acceptable. In cases where the cost of the service is not known in advance, the price can be an estimate, deposit request, or the like, with any overpayment refunded. Underpayment can be collected by requesting an additional payment from the client. In the absence of trust between the parties, frequent small payments may be required. Eastlake [Page 8] INTERNET-DRAFT Internet Payment Framework 7 July 1995 3. Payments and Receipts After encountering a price tag, either initially, during a session, or in conjunction with a "payment required" error, an application needs some method of tendering payment. This is done with a payment system string with the same syntax as described in Section 2.2 above. For example: foocash:29Uso+Oa/e92micHd4s3= The payment system used in the payment is selected from among those in the price tag as those are known to be supported by the seller. Payment systems will normally include in the payment system specific information some sort of serial or transaction number so that retransmission of a message containing the string will not result in duplicate payment. Normally the seller will provide a receipt for the amount of money actually collected or a message indicating payment failure or error. This will be via a receipt character string which is also simply in the form of a payment system string. For example: barsys:8n7VtC2+uL341/ Payment systems will normally include in the payment system specific information of a receipt, in addition to an indication of how much the receipt is for, some type of serial or transaction number, so that retransmission of a message containing the receipt will not result in confusion. A null payment or receipt string is explicitly permitted in most contexts as a way for an entity to indicate merely that it is payment syntax aware. One or more payment system names in isolation are permitted in a payment or receipt context but only as a way to indicate that a particular payment system is understood. Any actual payment or receipt must have a colon and non-null payment specific information. Only one such full payment system string can occur in a payment or receipt. Depending of payment system details, a refund can be implemented in two way. It can be a payment message from the seller to the buyer, normally leading to a receipt from the buyer to the seller. Or the seller may be able to directly refund to the buyer's account or the like and simply send the buyer a receipt. In some payment system, both refund techniques might be available. In others, refunding may not be possible. The content and/or encoding of the payment system specific Eastlake [Page 9] INTERNET-DRAFT Internet Payment Framework 7 July 1995 information would normally differ between the price tag, payment, and receipt contexts but this is a matter only of concern to the payment system. Errors in formating or the like that are internal to a payment or receipt should generally be handled by being logged and/or reported by an error message encoded into a receipt. Errors in a price tag may be reported in a payment or receipt. Great care should be taken to be sure to avoid any situation that could result in an endless loop of receipts. Eastlake [Page 10] INTERNET-DRAFT Internet Payment Framework 7 July 1995 4. Use in the World Wide Web The World Wide Web is a rapidly evolving system for information interaction that is being increasingly used for commerce. It is particularly well suited for the inclusion of payment systems, especially any designed for efficient handling of very small payments which might reasonably be incurred on a "per web page" basis or the like. In the Web, a price is indicated by a COST parameter, as described below, which can occur within an anchor, HTML document header, or several places in a FORM. Payment can be included with any HTTP request using the "ChargeTo:" header. A receipt can be included in any HTTP reply using the "Receipt:" header. 4.1 Web Browser User Interface It is important that small payments be closely integrated into the browser user interface. An expected mode of operation will be one of many small payments so the overhead associated with each must be small. It is unacceptable for the user to necessarily interact with a separate screen or window to approve each small payment although a user who wishes to do so should have that option. The user should be able to establish some threshold (default perhaps around 0.1usd or equivalent) such that actions incurring that charge or less are semi-automatic. That is, no special approval action is required, although color coding or the like should be used to distinguish toll links from free links, an optional sound could be made when any money is sent, or other clues used to give the user a feel for what is going on. To avoid spending an unexpectedly large amount in small pieces, possibly a bank graphic or the like should be displayed to show how much cash is still available to the browser before the user will have to take action. The act of refilling the bank would be a more heavy weight operation requiring user interaction or, to get a default amount, at least user approval. 4.2 Anchor Embedded Costs A cost can appear in an anchor. This is a very strong hint that payment of the indicated amount should accompany the GET operation that occurs when following that link. Note, however, that it is ultimately up to the server being hit to determine if payment is adequate or to follow the course it chooses for different levels of Eastlake [Page 11] INTERNET-DRAFT Internet Payment Framework 7 July 1995 payment. The cost is given by a COST parameter in the anchor. For example: Great stuff for one thin dime! It is recommended that toll links be shown in a different color or type style from toll-free links. Browsers may wish to go further and indicate different cost levels, particularly costs above or below any "automatic approval" level the user has. When the user has their pointer over the link, the browser may wish to display the payment particulars in a similar way to that in which it displays the URL. (Such a display could be filtered to the currency and/or payment system(s) actually available to the user.) Notice that the cost, if any, indicated by the anchor text ("Great stuff..." above) could be different from the actual "COST=" parameter which controls the payment sent with the request. In turn, the "COST=" amount could be different from what the server really wants. Or the server may provide different data or services for different payment amount. Such variable payment schemes may be better handled with a FORM as described below. 4.3 Page Header Price Tag The cost for accessing an HTML page can be included in the header. For example: Mating Habits of the Red Breasted Geek 0.75usd 0.99cad cybercash:A8jne8W2/sw== ... An attempt to get such a document without payment or with inadequate payment should fail (see Payment Required Error section below). A second attempt with payment will be required. This could be done in a manner similar to an access restriction failure followed by a second attempt with access authorization information. Implementing page header cost requires that the HTML for a web page be partly understood by the server, at least through the head, but this is necessary to implement the "title:" and "link:" response entity header fields anyway. Eastlake [Page 12] INTERNET-DRAFT Internet Payment Framework 7 July 1995 4.4 HTML Form Price Tags A cost can be associated with a form and with multiple choice items within the form. For example:
Miscellaneous text, etc. plain vanilla chocolate fudge

Your quality of service:

The COST associated with the form is a base price to which any multiple choice item costs are added. The form level COST may be omitted and COSTs can still appear with multiple choice items. The COST associated with a "select" is a default which applies only if no item is selected. When an item is selected, it over-rides the selection level cost and become the price component added into the total form price for that selection. The normally required payment system string can be omitted from some of the form COST parameters, in which case any prices add to the amount for all payment systems. But one or more payment systems and their payment system specific parameters must be determinable if any payment is to be sent. The payment system specific information associated with the last encountered instance of a payment system field in processing the form is used. If no payment system field is encountered, then no payment will be sent with the request even though "COST=" parameters are present. As with anchor costs, it is desirable to indicate the cost of multiple choice items by color coding and the cost of activating the form by color coding the submit button. Note that the submit button could change from free to toll or the like as choices are made in the form. 4.5 Web Payments and Receipts Any HTTP request can be accompanied with payment by including a payment line in the message header. This consists of the "ChargeTo:" Eastlake [Page 13] INTERNET-DRAFT Internet Payment Framework 7 July 1995 header label followed by a payment system string; however, "ChargeTo:" can appear with one or more bare payment system names for the purpose of indicating that the browser understands those systems without conveying any actual payment. Examples: ChargeTo: cybercash:A8jne8W2/sw== ChargeTo: foocash barsys The first example is in the form of a payment via cybercash. The second example is an indication by the sender that it understands the foocash and barsys payment systems. The browser should keep track of such actual payments it has sent and re-send the identical payment if the request needs to be retried with access authorization information or due to a transient error, rather than sending additional funds. The collection of payment or the specifics of the failure of a tendered payment are indicated back to the customer by a receipt line in the response header. This consists of the "receipt:" header label followed by a payment system string. For example: Receipt: cybercash:A8jne8W2/sw== There cases where a larger payment is collected initially and the unused portion refunded or where adjustments are required after a purchase. Because of this, ChargeTo and Receipt headers are both allowed in both HTTP requests and responses. 4.6 Payment Required Error If an HTTP request arrives without sufficient payment (or with none at all) and payment is required by the server, the server can simply provide a web page with limited or no actual information and possibly one or more links with COST parameters embedded in them. Alternatively, a "402 payment required" error is returned in which case there must be a "www-cost:" response header field analogous to the "www-authenticate:" header field for a "401 unauthorized"" response. The value of the www-cost field is the same as for the COST parameter described above. This is similar to an access restriction error in that the browser can just try again with payment included the way they can try again with access information. It may be possible to combine these by returning a 402 error with the HTML accompanying the error having a link with a COST parameter pointing to the originally sought item. This would combine automatic charging for browsers that have 402 Eastlake [Page 14] INTERNET-DRAFT Internet Payment Framework 7 July 1995 error processing implemented with a convenient way for the user to re-request with payment for browsers that understand anchor COST parameters but do not automatically handle 402 errors. 4.7 Web Proxies When information that has an owner and price is being cached and served to multiple different users by a proxy, the payments should be requested by the proxy. The safest thing for the proxy to do is to send payment to the entity it retrieved the data from using an HTTP request with a ChargeTo header and the PAYMENT method. If the proxy understands the payment system well enough and there are no firewall problems, the proxy may be able to collect the payment and directly transfer funds to the information owner. It is not expected that proxy payment collection will be perfect. There will initially be many dumb proxies that don't understand payment and there may be proxies that deliberately avoid collecting and forwarding payment. But any large scale avoidance of payment will be noticed. In any case, if the proxy can cache a copy, so could the user, who could then give copies to all his friends. The ease of automatically making small payments for information through this syntax is hoped to produce a net reduction in unauthorized information copying. Eastlake [Page 15] INTERNET-DRAFT Internet Payment Framework 7 July 1995 5. Use in File Transfer Protocol An FTP server may wish to charge for a file transfer (either way) or for an FTP session. It may do so by requesting that an ACCT command be sent via the 332 or 532 reply codes. 332 is used to indicate that a received command is being held in abeyance pending receipt of an ACCT while 532 indicates that a received command has been abandoned due to lack of payment and an ACCT command needs to be sent before attempting the command again. Price tags are indicated in the 332 or 532 text by a string at the beginning of the form in the 332 or 532 text, i.e., a literal "". The word "cost" is case insensitive. Arbitrary additional text may be included after the price tag. A payment can be send by simply including a payment string, as defined in section 3, after the ACCT command. A successful receipt is rendered by returning a 233 reply with a receipt payment system string as the beginning of its text. A payment failure receipt is rendered by returning a 433 or 533 reply depending on whether the failure is transient or permanent. In either case, the receipt string can be terminated by white space and additional text human readable text placed after the receipt string in the reply. (See RFC 959.) SUMMARY Price Tags - in existing 332 and 532 replies. Payments - in existing ACCT command. Receipts - in new 233, 433, and 533, replies. Eastlake [Page 16] INTERNET-DRAFT Internet Payment Framework 7 July 1995 6. Use in Telnet A host may wish to charge for a Telnet session. Telnet option code [TBD] is used to initially negotiate agreement of the two parties to speak about payment. As with other Telnet options, either side can sent IAC WILL xxx, in response to which an IAC DO xxx indicates agreement and an IAC DON'T xxx indicate refusal. Or a party can send IAC DO xxx to which IAC WILL xxx indicates agreement and an IAC WON'T xxx indicates refusal. After agreement to speak about payment has been reached, Telnet subnegotiation strings can be exchanged, bracketed with IAC SB and IAC SE. The initial subnegotiation byte indicates the type of payment message following in the rest of the subnegotiation byte string as follows: Byte Meaning ---- ------- 01 Price-tag 02 Payment 03 Receipt If desired, arbitrary binary representations may be used for payment system specific information after the colon terminating the system name in payments and receipts (as long as bytes with value 255 are doubled as per the Telnet standard). Termination can be unambiguously determined by the IAC SE sequence. However, price tags must stick with the ASCII syntax given herein as they must be parsed by systems that may not understand any particular payment system. (See RFCs 854, 855.) Eastlake [Page 17] INTERNET-DRAFT Internet Payment Framework 7 July 1995 7. Use in Simple Message Transfer Protocol A host or user may wish to charge for the receipt of mail. This is accomplished via the new 332 reply code. This is an interim success code that indicates that further information is required to complete a pending command. Note that use of 332 after the SMTP RCPT command would be a simple way to implement any particular user requiring payment for mail to be delivered to them and its use after the MAIL command would be a simple way to implement a system requiring payment for mail from all or certain sources (although this information is easy to forge). Payment is indicated by the new ACCT command. This is followed by a payment string as defined in section 3 above. Charging for mail may cut a host or user off from the normal flow of mail. It seems unlikely that most individuals or mailing lists would be willing to pay to send mail to an address. However, it is easy to envision cases where a service for which it would be reasonable to charge is requested via email. Or there may be individuals who do want to substantially cut themselves off from most mail or mail from certain senders. SMTP servers that speak ESMTP (see RFC 1651) may optionally give the new EHLO keyword ACCT. However, ESMTP is designed for servers to list features to be optionally invoked by clients. It is not really appropriate as a means for servers to indicate features that they will *require* of clients. In any case, it is believed that no negotiation is necessary for an SMTP server to use the new 332 reply code. RFC 821 is clear that the receipt of any 3xx reply code after a MAIL, RCPT, etc. command is to be considered an error. This is the appropriate understanding for an SMTP client that does not understand payment when an SMTP server requires payment. The rules and state diagrams in RFC-821 are hereby amended and the state diagram for MAIL, RCPT, SEND, SOML, and SAML is modified to the following: Eastlake [Page 18] INTERNET-DRAFT Internet Payment Framework 7 July 1995 1 +---+ 1,3 FLOW FOR +----------->| E |<-----+ PAYMENT AWARE | +---+ | SMTP SERVER | | | 2 +---+ 2 | +----------->| S |<-----+ | +---+ | | | | | +---+ cmd +---+ 3 +---+ ACCT +---+ | B |--------->| W |----->| |------->| W | +---+ +---+ +---+ +---+ | | | 4,5 +---+ 4,5 | +----------->| F |<-----+ +---+ A successful receipt is rendered by returning the new 233 reply with a receipt payment system string as the beginning of its text. A payment failure receipt is rendered by returning the new 433 or 533 replies depending on whether the failure is transient or permanent. In either case, the receipt string can be terminated by white space and additional text human readable text placed after the receipt string in the reply. The middle digit 3 in SMTP reply codes is reserved for accounting, corresponding to its existing use in FTP. (See RFCs 821, 1651.) SUMMARY Price Tags - in new 332 reply. Payments - in new ACCT command. Receipts - in new 233, 433, and 533 replies. Eastlake [Page 19] INTERNET-DRAFT Internet Payment Framework 7 July 1995 8. Protocols to Which Payment is not Applicable Some protocols are sufficiently basic to the operation of the network or provide sufficiently light-weight access to public information that attempts to impose application level payment would be inappropriate. Some of these protocols, listed below, SHOULD NOT make use of this syntax or impose prices or payments. 8.1 The Domain Name System Part of the philosophy of the Domain Name System (DNS) is that it contains public information and generally gives the same answers to all inquirers. It is used for such fundamental purposes as translating domain names (such as tam.cybercash.com) into IP addresses or specifying SMTP mail backup and routing servers. As such, charges SHOULD NOT be imposed for DNS queries. (See RFCs 1034, 1035.) 8.2 The Finger Service Finger is an optional information service intended to permit remote users to learn a limited amount of information about a user or users on an Internet host. Information such as the time they last logged in or contents of their ".plan" file. There are serious security considerations involved in allowing finger access to a host and hosts are free to decide how much such access, if any, they will provide. In some cases, finger servers have been set up to act as information retrieval or reporting mechanisms, but this was not the designed purpose of finger and, in most cases, there are better mechanisms to provide such access. If finger access is provided because a site wishes to be open, charges SHOULD NOT be imposed. (See RFC 1288.) 8.3 The Auth Service This service, when implemented, allows a remote host to determine the user associated with a TCP connection. It is intended as a security and auditing tool although it is weak in the face of anyone with direct access to the TPC or IP level who was attempting to mislead Eastlake [Page 20] INTERNET-DRAFT Internet Payment Framework 7 July 1995 it. Implementation is optional. Those who chose to provide this service are doing so to cooperate in such security or auditing at some sacrifice in the privacy of their users. Charging for this service makes little sense in this context. (See RFC 931.) 8.4 The ECHO, DISCARD, and CHARGEN Services These are light weight services intended for network maintenance. ECHO echoes the packet sent to it (see RFC 862), DISCARD throws away packets sent to it but maintains the connection (see RFC 863), and CHARGEN generates an infinite number of random characters and sends them until the calling party disconnects (see RFC 864). Hosts are free to decide which, if any, of these three services they wish to provide (although ECHO is Recommended), but SHOULD NOT impose any charges for them. Eastlake [Page 21] INTERNET-DRAFT Internet Payment Framework 7 July 1995 9. Security Considerations Getting authorization to construct payments may, depending on the payment system, require the user to enter a passphrase. For example, a passphrase might be required at the beginning of their session to unlock a private key. Thus the user could be vulnerable to Trojan horse web browsers, ftp clients, telnet clients, etc., as they are to many other types of Trojan horse applications. Use of "secure" application distribution with signed executables, checksums, virus detection, etc., should be encouraged. An adversary may be able to observe or modify traffic to and from an application. Payment systems should be designed so that such observation results in minimal loss of privacy and such observation or modification can not result in hijacking a payment. Note that an adversary that has complete control over application communications can pretend to be a merchant just as it could by controlling an end node. However, such impersonation from an end node may be easier to trace and control than impersonation at an unknown point along the communications path. Message (MOSS, SHTTP, etc.) and connection (IPSEC, IPv6, SSL, etc.) security protocols are available to help protect the communications path. On receipt of an advance payment, a server is capable of charging the user regardless of whether the server actually provides the data or services being charged for. A server could even send back an error message but keep and use the payment. Some means of automatically logging payments that result in a software or human detectable failure to deliver should be implemented so these can be examined for patterns or cross checked with payment system statements of account. A merchant can withhold and fail to send back to the user a receipt. Applications should assume any payment sent will be collected regardless of whether they get a receipt back. With payment systems, a monetary cost can sometimes be associated with downloaded data. Caching algorithms may wish to take this into account and cache costly data in preference to free data. Servers should accept the identical data request from the same net entity for a reasonable amount of time even if the payment being presented appears to be a duplicate. Transient errors may have prevented use of the data previously downloaded for that request. A bad client application could generate payments exceeding the funds or authorization available to it. Servers should verify payments promptly and be cautious of extending services or goods unless they can confirms that payment is good. Applications and payment systems should be designed to limit the amount of funds a rogue application could transfer. Eastlake [Page 22] INTERNET-DRAFT Internet Payment Framework 7 July 1995 References [ISO 4217] - Codes for the representation of currencies and funds [RFC 821] - J. Postel, "Simple Mail Transfer Protocol", 08/01/1982. [RFC 854] - J. Postel, J. Reynolds, "Telnet option specifications", 05/01/1983. [RFC 855] - J. Postel, J. Reynolds, "Telnet Protocol specification", 05/01/1983. [RFC 959] - J. Postel, J. Reynolds, "File Transfer Protocol", 10/01/1985. Author's Address Donald E. Eastlake, 3rd CyberCash, Inc. 318 Acton Street Carlisle, MA 01741 USA Telephone: +1 508 287 4877 +1 703-620-4200 (main office, Reston, Virginia, USA) email: dee@cybercash.com Expiration and File Name This draft expires 28 December 1995. Its file name is draft-eastlake-internet-payment-00.txt. Eastlake [Page 23] INTERNET-DRAFT Internet Payment Framework 7 July 1995 Appendix A: Initial Payment System Names This is the initial alphabetic list of the initial registered payment system names that intend to be usable via this syntax. [send email to author, dee@cyercash.com, if you would like to be added] Company Name Email Contact Home Page ------------ ------------- --------- Payment System Name - (brief description) ------------------- CyberCash, Inc. info@cybercash.com http://www.cybercash.com cybercash - (credit card) cash - (cash) Eastlake [Page 24] INTERNET-DRAFT Internet Payment Framework 7 July 1995 Appendix B: Simplified BNF This is a BNF-like description of the Payment Protocol syntax syntax, using the conventions of RFC822, except that "|" is used to designate alternatives, and brackets [] are used around optional or repeated elements. Briefly, literals are quoted with "", optional elements are enclosed in [brackets], and elements may be preceded with * to designate n or more repetitions of the following element; n defaults to 0. ;prices isocurrency = alpha alpha alpha ietfcurrency = 4*alpha privatecurrency = "x-" 4*alpha currency = isocurrency | ietfcurrency | privatecurrency digits = 1*digit decimalpoint = "." | "," number = digits | digits decimalpoint *digit cost = number currency ;payment system strings ianapaysys = 4*alpha privatepaysys = "x-" 4*alpha paysysname = ianapaysys | privatepaysys paysys = paysysname ":" *uchar ;price tag pricetag = *sp paysys *[ 1*sp cost | 1*sp paysys ] *sp | *sp *[ cost 1*sp | paysys 1*sp ] paysys *sp ;miscellaneous definitions lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" | "j" | "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" | "s" | "t" | "u" | "v" | "w" | "x" | "y" | "z" hialpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" | "J" | "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" | "S" | "T" | "U" | "V" | "W" | "X" | "Y" | "Z" alpha = lowalpha | hialpha digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | Eastlake [Page 25] INTERNET-DRAFT Internet Payment Framework 7 July 1995 "8" | "9" other = "$" | "-" | "_" | "." | "+" | "/" | "=" | "@" hex = digit | "A" | "B" | "C" | "D" | "E" | "F" | "a" | "b" | "c" | "d" | "e" | "f" escape = "%" hex hex sp = " " uchar = alpha | digit | other | extra | escape Eastlake [Page 26]