Misc AADS & X9.59 Discussions

AADS Postings and Posting Index, next, previous - home



Basic credit-card payment question
Basic credit-card payment question
Basic credit-card payment question
Basic credit-card payment question
Digital signatures as proof
Meaning of Non-repudiation
Meaning of Non-repudiation
Meaning of Non-repudiation
Meaning of Non-repudiation
Meaning of Non-repudiation
Federated Identity Management: Sorting out the possibilities
Meaning of Non-repudiation
Meaning of Non-repudiation
Words, Books, and Key Usage
Meaning of Non-repudiation
Meaning of Non-repudiation
international financial standards (ISO TC68)
Alternative to Microsoft Passport: Sunshine vs Hai
IBM alternative to PKI?
IBM alternative to PKI?
IBM alternative to PKI?
IBM alternative to PKI?
IBM alternative to PKI?
Proxy PKI. Was: IBM alternative to PKI?
Proxy PKI. Was: IBM alternative to PKI?
Proxy PKI. Was: IBM alternative to PKI?
Proxy PKI
Proxy PKI
Proposal: A replacement for 3D Secure
Proposal: A replacement for 3D Secure
Proposal: A replacement for 3D Secure
Proposal: A replacement for 3D Secure
ALARMED ... Only Mostly Dead ... RIP PKI
ALARMED ... Only Mostly Dead ... RIP PKI
ALARMED ... Only Mostly Dead ... RIP PKI
ALARMED ... Only Mostly Dead ... RIP PKI .. addenda
ALARMED ... Only Mostly Dead ... RIP PKI .. addenda II
ALARMED ... Only Mostly Dead ... RIP PKI
ALARMED ... Only Mostly Dead ... RIP PKI ... part II
ALARMED ... Only Mostly Dead ... RIP PKI .. addenda
ALARMED ... Only Mostly Dead ... RIP PKI ... part II
Royal Mail pulls plug on ViaCode digital certificate
ALARMED ... Only Mostly Dead ... RIP PKI ... part III
PKI: Only Mostly Dead
Web site exposes credit card fraud
Web site exposes credit card fraud
Giuliani: ID cards won't curb freedoms
maximize best case, worst case, or average case? (TCPA)


Basic credit-card payment question

From: Lynn Wheeler
Date: 04/19/2002 11:09 AM
To: Anders Rundgren <anders.rundgren@xxxxxxxx>
cc: internet-payments <internet-payments@xxxxxxxx>
Subject: Re: Basic credit-card payment question
it is a lot more complicated than that.

the merchant has a relationship with a mechant bank/acquier. the merchant sends transactions to the merchant bank/acquier. The merchant bank/acquier sends the transactions to the issuering bank. The merchant bank approves/disapproves the transaction and sends the response.

Usually sometime that night, the merchant bank/acquier does a batch financial transaction to each issueing bank that has any money to be sent to any merchant at that merchant bank. This is via one (or more) of the wholesale settlement systems. Basically there can be a single wholesale funds transfer from each issuing bank to that specific acquiring bank. The acquiring bank then aggregates all incoming funds and then credits the appropriate amounts to each merchant account.

The wholesale settlement networks are used to do bank-to-bank financial transfers for a number of reasons, ACH (aka check), ATM (debit), credit card, wire transfer, etc. In the case of credit card transfers, the acquiring bank has the mapping between the merchant credit identifier and the merchant bank account number. If a customer knew the merchant bank account number ... they could execute a ACH push ... either via a check or other means to that account. The problem is how to reconcile a financial deposit in the merchant account with a specific customer/merchant transaction. Various of the EDI 8xx transactions have addressed some of this ... merchant gets a bulk financial transaction with an ACH addenda record listing the itemized details of individual accounts. An example might be electronic payments to utility company. Given that the utility company had setup the appropriate EDI mechanism ... they could get a single bulk transfer for the amount transferred for that day ... with EDI ACH addenda record giving the individual account numbers and amounts involved in the transaction.

Some of the supply-chain management infrastructures associated with purchase cards send detailed "credit card statement" to the corporation electronically in EDI addenda record format so that the corporation can reconcile billing with transactions.

When the merchant is initiating essentially a "pull" transaction ... it is easier for the merchant to correlate the payment with specific customer transaction. For a customer initiated "push" transaction (say via ACH push) .... the conventions aren't all in place so that it is easy for a merchant to correlate a received payment with a specific transaction (the various EDI applications that support pushed ACH payments aren't really a consumer product). Just doing the money transfer isn't sufficient ... the merchant has to additioinally have the information that correlates specific payment amount with specific transaction, purchase order, etc

Customers push monthly payments to all sorts of merchants ... but they typically have a pre-established account at that merchant ... and the incoming check includes the customer's account number statement for that merchant. There is big business in payment drop-boxes where a 3rd party outsourcer handles all the incoming checks and statements. They may generate a single large EDI ACH addenda record with line item detail for the account numbers, possibly names, and amounts for each individual account ... so the merchant can reconcile accounts receivable information. In some cases, some of the banking electronic payments system are banks directly offering the service of some of the drop-box operators.

on 4/19/2002 4:28 am wrote:
The common way to perform credit-card transactions is that the merchant initiates the payment with the help of the customers' card (data).

Do credit-card enabled merchants have any possible way to receive money through the credit-card networks through an off-line operation performed by the customer though his/her issuer? I.e. an account-to-account transaction run in the other direction? That would be terrific!

Anders


Basic credit-card payment question

Refed: **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 04/19/2002 11:23 AM
To: Sean Owens <sowens@xxxxxxxx>
cc: Anders Rundgren <anders.rundgren@xxxxxxxx>,
    internet-payments <internet-payments@xxxxxxxx>,
    "'kjvella@xxxxxxxx'" <kjvella@xxxxxxxx>,
    "'razorbacksean@xxxxxxxx'" <razorbacksean@xxxxxxxx>
Subject: RE: Basic credit-card payment question
current internet infrastructure:
http://www.garlic.com/~lynn/aadsm5.htm#asrn2 Assurance, e-commerce, and some x9.59 ... fyi
http://www.garlic.com/~lynn/aadsm5.htm#asrn3 Assurance, e-commerce, and some x9.59 ... fyi

financial standards body standard for all electronic payment types (credit, debit ach, atm, e-check, etc) in all environments (internet, point-of-sale, non-internet, etc):
http://www.garlic.com/~lynn/x959.html#x959

the requirement given the x9a10 working group for x9.59 standard was to preserve the integrity of the financial infrastructure for all electronic retail payments without encryption (i.e. can have digital signatures or other method for data integrity, but not necessary to encrypt the data).

realted is the NACHA atm trials
http://www.garlic.com/~lynn/x959.html#aads

the aads chip strawman (at the above reference) has a booth at cardtech/securetech next week in New Orleans.

on 4/19/2002 6:26 am wrote:
Kevin,

Interesting Scheme. Actually, from a transaction perspective, and Cardholder Authentication, wouldn't it make sense to utilize a cardholder verification scheme, comparable to Card swipe, that effectively proves the card was on premise, and utilized by the cardholder. Effectively, we're looking for a way to decrease processing fee's.

The card orgs up the Discount % a merchant pays for MOTO transactions, since eCommerce trans, are acquired as MOTO (more oft than not), then fee's are the issue. Other than CVV2/CVC2 verification on MOTO trans, what other form of Cardholder authentication can be utilized to verify validity. The eWallet idea, is comparable to Virtual Cards. What if, with your issuer from an online standpoint, payment could be initiated electronically, via an Internet Banking scheme, for Authorization. Effectively, when I want to perform and eCommerce tran, I log onto my Bank, and initiate the payment, immediately from the Issuer to the Acquirer. All Cardholder authentication is handled by the ISSUER, verification is not needed from the Acquirer. This could follow an eWallet structure for authentication and card selection and tran processing.

Sean Owens
Euronet Worldwide
1027 Budapest
14-24 Horvat Utca
HUNGARY
Phone +36 1 224 1626
Mobile: +36 20 932 1635
on 4/19/2002 1:42 pm wrote:
>
>not really terrific from a convenience point of view as that would
>mean the end of our forum!!!!
>
>An idea can include minimising the number of merchants - in other
>words the credit cards of this world would have an ewallet company in
>each and every country and the card holder pays the ewallet company
>and uses his ewallet to buy on-line.  The online shops need not be
>online with the card schemes just with the ewallet company to check
>funds.  this might probably be better from a chargeback and security
>point of view.....  kev
>
>Kevin J. Vella
>International Business Development
>e-shore - Terranet Limited
>http://www.e-shore.net <http://www.e-shore.net>
>e. kjvella@xxxxxxxx <mailto:kjvella@xxxxxxxx>
>t. +356 21 489400
>m. +356 7985 0000
>f. +356 21 489600
>
>Other Company Websites:
>http://www.e-shore.com.mt <http://www.e-shore.com.mt/>
>http://www.di-ve.com <http://www.di-ve.com/>
>http://www.terranet.com.mt <http://www.terranet.com.mt/>
>http://www.maltanet.net
>
>Postal Address:
>Dolphin Centre, Main Street,
>Balzan, Malta (GC) BZN 08
>
>on 4/19/2002 12:28 am wrote:
>>
>>The common way to perform credit-card transactions is that the
>>merchant initiates the payment with the help of the customers' card
>>(data).
>>
>>Do credit-card enabled merchants have any possible way to receive
>>money through the credit-card networks through an off-line operation
>>performed by the customer though his/her issuer?  I.e. an
>>account-to-account transaction run in the other direction?  That would
>>be terrific!
>>
>>Anders


Basic credit-card payment question

From: Lynn Wheeler
Date: 04/19/2002 11:53 AM
To: Anders Rundgren <anders.rundgren@xxxxxxxx>
cc: internet-payments <internet-payments@xxxxxxxx>, kjvella@xxxxxxxx
Subject: Re: Basic credit-card payment question
some of the b-to-b purchase cards do some of this type of stuff ... including translating the statements (both back to the paid to merchant and the paid from customer) into EDI 8xx format.

on 4/19/2002 7:31 am wrote:
Kevin,
My need for this is to be able to pay invoices using credit cards. I.e. a merchant can in a B2B-scenario send an invoice requiring either pre-payment to perform shipping, or this is just another way to pay international low-dollar invoices. I.e. an alternative to Swift or checks.

Anders


Basic credit-card payment question

From: Lynn Wheeler
Date: 04/20/2002 01:19 PM
To: "R. A. Hettinga" <rah@xxxxxxxx>
cc: internet-payments <internet-payments@xxxxxxxx>
Subject: Re: Basic credit-card payment question
note that the comment was taken somewhat out of context since the original question was directed at a b-to-b scenario. misc. refs (especially #2 below):
http://www.garlic.com/~lynn/aadsm11.htm#0 Basic credit card payment question
http://www.garlic.com/~lynn/aadsm11.htm#1 Basic credit card payment question
http://www.garlic.com/~lynn/aadsm11.htm#2 Basic credit card payment question

the problem has been the merchant matching the pushed payment with invoice or transaction. it is somewhat easier when the merchant presents the payment requests and gets back an answer. having the payee push the payment .... requires some like EDI 8xx ACH addenda record containing enuf infomration so that the merchant can correlate with stuff like their accounts receivable system. Even payment cards are a merchant presented business process. Some of the electronic bill payment systems do this and have the ACH addenda record (that is for merchants where the electronic bill payment processor doesn't have to resort to printing & mailing a paper check because the merchant doesn't have the facilities for processing a combined electronic payment plus electronic statement/invoice detail information).

on 4/20/2002 5:59 am wrote:
--- begin forwarded text from Digital Bearer Settlement List

"David G.W. Birch" <dgw-lists@xxxxxxxx> wrote:
lynn.wheeler@xxxxxxxx e-said:
> If a customer knew the merchant bank
> account number ... they could execute a ACH push

Surely they would never do this, because of the guarantees and legal
protections (and frequent flier miles) associated with their credit card.

Try charging back a wire transfer.

Regards,
Dave Birch.

...      My own opinion (I think!) given solely in my capacity as      ...
...             an interested member of the general public             ...
...                                                                    ...
...  mail(at)birches.org      .........      http://www.davebirch.org/ ...

--- end forwarded text

--
-----------------
R. A. Hettinga <mailto: rah@xxxxxxxx>
The Internet Bearer Underwriting Corporation <http://www.ibuc.com/>
44 Farquhar Street, Boston, MA 02131 USA
"... however it may deserve respect for its usefulness and antiquity,
[predicting the end of the world] has not been found agreeable to
experience." -- Edward Gibbon, 'Decline and Fall of the Roman Empire'


AW: Digital signatures as proof

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 04/26/2002 12:36 PM
To: "Weber, Arnd" <Arnd.Weber@xxxxxxxx>
cc: 'Anders Rundgren ' <anders.rundgren@xxxxxxxx>,
    'Clara Centeno ' <Clara.Centeno@xxxxxxxx>,
    "'Gary W. Fresen '" <gfresen@xxxxxxxx>,
    "'heikki.sundquist@xxxxxxxx '" <heikki.sundquist@xxxxxxxx>,
    '''internet-payments ' ' ' <internet-payments@xxxxxxxx>
Subject: Re: AW: Digital signatures as proof
part of the EU finread standard .... seems to be chipcard readers that have certified security modules comparable to POS terminals .... aka class-4 terminals that include secure pinpads and secure displays.

there is the concept of hardware token holding a private key that is never divulged and therefor the token can uniquely authenticate itself (something you have). what finread doesn't quite get around to specifying is a compareable mechanism in the terminal/reader to uniquely authenticate itself .... with the transition to open network ... rather than VANs & other forms of private networks ... it is a lot more difficult to accurately assure the integrity of the end-point (aka is it really a class-4 terminal/reader).

x9.59 payment protocol for all types of payments in all types of environments (aka not just credit on the internet) .... included the provision that the end-point terminal also may be required to authenticate itself as well as any hardware token.

we demo'ed such a hardware token & terminal combination this week at securetech/cardtech (aka both the hardware token and the secure terminal signing different kinds of transactions ... payment, access, authorization, etc).

misc. recent security & finread threads
http://www.garlic.com/~lynn/2002c.html#10 Opinion on smartcard security requested
http://www.garlic.com/~lynn/2002c.html#21 Opinion on smartcard security requested
http://www.garlic.com/~lynn/2002f.html#46 Security Issues of using Internet Banking
http://www.garlic.com/~lynn/2002f.html#55 Security Issues of using Internet Banking

arnd.weber@xxxxxxxx on 4/24/2002 1:55 pm wrote:
I'm not sure. At least here in Europe we had "phantom transactions" at ATMs. Reasons I remember have been eavesdropping of PINs, maintenance errors, fake ATMs, etc. Today, it is hard to tell between somebody whose ATM-PIN was attacked, and somebody who only claims that his PIN was attacked. I think we have to anticipate that log-in procedures into signature systems may also be attacked. Actually the difference between using a local signature implementation in a networked office-PC and using a server-based one may be small - the user doesn't really control either system. But on the server-based system, by definition other people have control of the password. And not all transactions can easily be reversed, in particular not money transations.


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/02/2002 02:53 PM
To: "Trevor Freeman" <trevorf@xxxxxxxx>
cc: "Denis Pinkas" <Denis.Pinkas@xxxxxxxx>,
     "David P. Kemp" <dpkemp@xxxxxxxx>, ietf-pkix@xxxxxxxx
Subject: RE: Meaning of Non-repudiation
the NR bit in a certificate seems to be almost totally unrelated to whether a person really intended for a signature to occur. within the chipcard domain (on the way to showing non-repudiation need to first establish intention &/or approval) ... there has been two different kinds of ... lets say personalities;

1) access card personality .... the card powers on, a PIN is entered, and the card signs an unlimited number of messages ... as an authentication indication; from 3-factor authentication: two factors, a) something you have and b) something you know.

2) financial card personality ... the card requires that a PIN be entered for every signature; this has an implicit "intention" or "approval" in addition to authentication; the card has the aspect of an access card personality (2-factor authentication) as well as the additional implicit aspect of "intention" or "approval" because a human interaction is required for every signature ... aka the re-entry of the PIN for every signature implying intention. The finread reader standard goes further ... in that there needs to be a "secure" card acceptor device with secure display & secure pin-entry (further increasing the probability that the specific human was involved in the re-entry of the PIN ... and it was done specifically in conjunction with a specific transaction being displayed).

Having a "NR" flag in a certificate seems to have little or no bearing on whether there is an implicit intention of a person that a specific signature be performed (in the sense of a chipcard with a "financial" personality that carries with it some action implying approval or intention).

misc. past "intention" post (aka before getting to non-repudiation ... need to first establish something like intention &/or approval before there is a basis for repudiation or non-repudiation):
http://www.garlic.com/~lynn/aadsmore.htm#schneier Schneier: Why Digital Signatures are not Signatures (was Re :CRYPTO-GRAM, November 15, 2000)
http://www.garlic.com/~lynn/2001g.html#60 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001g.html#62 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001j.html#7 No Trusted Viewer possible?

misc. past finread posts:
http://www.garlic.com/~lynn/aadsm9.htm#carnivore Shades of FV's Nathaniel Borenstein: Carnivore's "Magic Lantern"
http://www.garlic.com/~lynn/aepay7.htm#3dsecure 3D Secure Vulnerabilities?Photo ID's and Payment Infrastructure
http://www.garlic.com/~lynn/aadsm10.htm#keygen2 Welome to the Internet, here's your private key
http://www.garlic.com/~lynn/aadsm11.htm#4 AW: Digital signatures as proof
http://www.garlic.com/~lynn/2001g.html#57 Q: Internet banking
http://www.garlic.com/~lynn/2001g.html#60 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001g.html#61 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001g.html#62 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001g.html#64 PKI/Digital signature doesn't work
http://www.garlic.com/~lynn/2001i.html#25 Net banking, is it safe???
http://www.garlic.com/~lynn/2001i.html#26 No Trusted Viewer possible?
http://www.garlic.com/~lynn/2001k.html#0 Are client certificates really secure?
http://www.garlic.com/~lynn/2001m.html#6 Smart Card vs. Magnetic Strip Market
http://www.garlic.com/~lynn/2001m.html#9 Smart Card vs. Magnetic Strip Market
http://www.garlic.com/~lynn/2002c.html#10 Opinion on smartcard security requested
http://www.garlic.com/~lynn/2002c.html#21 Opinion on smartcard security requested
http://www.garlic.com/~lynn/2002f.html#46 Security Issues of using Internet Banking
http://www.garlic.com/~lynn/2002f.html#55 Security Issues of using Internet Banking

trevor freeman on 5/2/2002 12:52 pm wrote:
I am breaking one of my New Year resolutions by mailing on this topic, but here goes...

Signing anything is always a challenge to prove position of a private key to authenticate whether it's in the context of a protocol like SSL or over a SMIME message. Technically all we can say is the signature occurred. The intent behind why the signature occurred is beyond the scope of this discussion. Use of certificates with the NR bit asserted is ultimately a matter of local policy on what constitutes usage as part of a non-repudiation service since two organisations could have two separate non-repudiation service where one requires a NR signature as part of an SSL session, and the other only wants NR signatures over SMIME messages.

Never in the course of IETF has history so much been written over something so small.

Trevor


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/03/2002 11:48 AM
To: ietf-pkix@xxxxxxxx
cc: epay@xxxxxxxx
Subject: RE: Meaning of Non-repudiation
in the X9A10 working group (for financial payment standards) provisions were made for a token acceptor device to also be able to sign a transaction. the operational characteristic of the token acceptor device (like finread standard) would be registered (RA; with or w/o a certificate necessary) which would indicate whether it operated in such a manner as to require a human interaction implying intention.

a simple hardware token digital signature by itself wasn't sufficient for implying intention (a necessary pre-requisite before even talking about NR) ... but the process around the execution of the token digital signature was also necessary for implying intention. A finread complient reader provides basis for inferring intention ... but then you still need proof that such a reader was used for the specific transaction (either because of some physical boundary, like ATM machines on private networks ... or because the token acceptance device is certified and also signs the transaction).

whether or not the digital signing environment was operated according to acceptable mechanism for implying intention ... is pretty much independent of whether there is a NR bit flag in a certificate. The additional signature by a complient/registered token acceptor device attests to the fact that specific operations were actually observed in conjunction with the specific signing operation that might then be used to imply intention on part of a person (as a necessary prerequisite for any attempt at establishing non-repudiation).

Nominally a person would have to make some indication as to their intention ... that could be done up front to a device ... which would then know to also only select a public key that had a certificate with a NR bit set. Note however, it isn't whether a certificate with a NR flag that is important ... it is important that the process of performing the digital signature followed precedures necessary for infering intention (and some registered device can attest to the intention infering process was followed).

__________________

previous ref:
2) financial card personality ... the card requires that a PIN be entered for every signature; this has an implicit "intention" or "approval" in addition to authentication; the card has the aspect of an access card personality (2-factor authentication) as well as the additional implicit aspect of "intention" or "approval" because a human interaction is required for every signature ... aka the re-entry of the PIN for every signature implying intention. The finread reader standard goes further ... in that there needs to be a "secure" card acceptor device with secure display & secure pin-entry (further increasing the probability that the specific human was involved in the re-entry of the PIN ... and it was done specifically in conjunction with a specific transaction being displayed).

http://www.garlic.com/~lynn/aadsm11.htm#5 Meaning of non-repudiation

Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/04/2002 10:00 AM
To: "Santosh Chokhani" <chokhani@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx, "'Trevor Freeman'" <trevorf@xxxxxxxx>,
    epay@xxxxxxxx
Subject: RE: Meaning of Non-repudiation
i still think the semantics are confusing .... rather than random challenges ... lets say authentication vis-a-vis approval/agreement/intention ... is much more sensible semantics from business and/or process orientation.

an authentication could be a radius implementation sending

time-value LOGIN:

the respondent creates a PPP radius response that includes the time-value, the entities userid and a FIPS-186 digital signature. FIPS186 digital signature contains a random component by definition. simple authentication, no sense of approval/agreement/intention.

this is as opposed to something like a financial transaction or other physical signature analog that carries with it the sense of human approval/agreement/intention. non-repudiation is something of a legal issue that may use something about human approval or intention as part of proving something. An electronic device that performs digital signatures automagically doesn't carry with it the same implicit connotation that a physical signature does ... which requires an explicit human interaction. The human interaction of executing a physical signature carries with it some concept that the person, in fact intended to write the signature that they were writing. An automagic electronic device creating digital signatures doesn't carry with it the same connotation that the person was explicitly involved in the execution of that specific digital signature.

From a process standpoint involving automagic electronic digital signing devices ... there is no difference between the signing of random bits and the signing of data. There may be a business reason to distinquish between signature operations on challenge information vis-a-vis signature operations on non-random data. However, in that case, the signer should get to contribute a field to the signed message indicating what it believes it is signing. Since an appended certificate isn't part of the signed message ... it wouldn't directly establish what the person believed (if they even knew) that they were signing.

Trying to load a static certificate with semantic meaning as to what a person believed to be signing ... for each signing operation ... leads down this garden path that there needs to be a unique certificate for each possible signing business case ... and since a static appended certificate can't directly indicate what a person believe they were signing ... then there would also have to be a unique public/private key pair for each possible certificate for each possible business case. Given that you can proove that a person has only registered a public key for one and only one type of business process, and there exists one and only unique certificate for that public key ... specificying one and only one unique signing business process ... then you may be able to infer that the use of a corresponding private key with the corresponding appended certificate indicates what type of data the person believed they were signing.

Moving the type of (business/process) data indication (that the person believed they were signing) out of the certificate and into the content of the signed message .... eliminates having to infer it from a static appended certificate (along with the corresponding requirement of being able to proove that a specific public/private key pair has only been registered for a certificate with a unique business type indication). Having a public/private key registered with multiple different certificate business types .... or the same certificate indicating multiple possible business uses (types of data being signed) .... creates ambiquity as to what type of data the person believed it was signing (i.e. no longer able to infer unique type of data the person believed it was signing based on the business use indications in the certificates(s)).

The issue of type of data the person believed it was signing (by inferring it from their choice of a unique signing key that is uniquely associated with each specific type of business data) is orthogonal to the business process issue that may require some physical act on part of a person in conjunction with a signing operation which can be used to infer approval/agreement/intention for the execution of that specific digital signature.

santosh chokhani on 5/4/2002 7:15 am wrote:
Let us call NR bit bit A and DS bit bit B.

You are welcome to show how you interpret the bits. What I am proposing is as follows:

A = 0, B = 0 => The public key may not be used to verify signature on data on random challenges.

A = 1, B = 0 => The public key may be used to verify signature on data, but not on random challenges.

A = 0, B = 1 => The public key may not be used to verify signature on data, but can be used to verify signature on random challenges.

A = 1, B = 1 => The public key may be used to verify signature on data and can be used to verify signature on random challenges.

I suspect of these, case "c" may be the most controversial, if any. How many implementations use case c and use the public key to verify the signature on data?


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/06/2002 09:58 AM
To: ietf-pkix@xxxxxxxx, epay@xxxxxxxx
Subject: RE: Meaning of Non-repudiation
somewhat as aside, in theory the whole point of certificates & the "certified" information is for use/reliance by relying parties.

having a

1) bit that says the sender believes that they are signing something with no meaning (random data) and

2) diffferent bit that says the sender believes that they are signing something with meaning (not random bits) containing some semantic meaning with possible implication that the signing operation indicates the sender is in some agreement with the contents of the signed message.

backing those bits out of the actual signed message and into an appended certificate could work if there was an exact one-to-one correspondance between the key doing the signing and the appended certificate. However, if it is a certificate with both of the bits on ... then it is somewhat defeating the purpose of having the bits .... because the relying party no longer can rely on specific deterministic meaning (aka the sender believes that they are signing something with no meaning ... or signing something with meaning ... but relying party isn't able to tell which because both bits are on, defeating the purpose of having the bits at all).

The other problem is being able to proove that sender has never acquired multiple certificates for the same key pair with different bits set. Since the appended certificates aren't part of the signed message ... the sender could claim that they had appended the certificate with the "no meaning message" ... and somebody substituted a certificate the "no meaning" bit off and the "meaning" bit on. That problem then suggests two extremes:

1) never allow people to generate their own keys and supply them to 3rd parties for certification. CAs can guarentee that key pairs are never submitted for multiple certificates if they only certify random keys that the CA generate themselves .. and that creating a new certificate always requires generating a new key pair (unless there is a global repository used by all CAs checking to see if a specific key-pair had ever previously been certified).

2) moving the certificate inside the signed message. the reason to do this is the original suggestion involving the flag bits indicating the senders belief as to the flag indicating message type/characteristic needs to be part of the signed message. moving the whole certificate inside the signed message is somewhat overkill just to get a couple bit indicators moved inside. this solution requires a convention that signers always include flags indicating their belief as to the type of message (people doing brain-dead signing of messages w/o looking at the contents might get themselves into situation similar to people signing blank checks or blank pieces of paper).

previous pieces of thread:
http://www.garlic.com/~lynn/aadsm11.htm#5 Meaning of Non-repudiation
http://www.garlic.com/~lynn/aadsm11.htm#6 Meaning of Non-repudiation
http://www.garlic.com/~lynn/aadsm11.htm#7 Meaning of Non-repudiation

Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/06/2002 11:59 AM
To: Denis Pinkas <Denis.Pinkas@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx
Subject: Re: Meaning of Non-repudiation
a "problem" is that certificates are suppose to be something that a relying party ... can rely on ... because relying parties supposedly can't directly trust the signer/sender.

if you are going to be able to always trust the signer/sender in all things ... then why do you need certificates ... which certify information by independent parties ... just trust the signer/sender and completely do away with certificates.

says that the relying party can't trust the signer ... but can trust the CA ... except it has to trust the signer to not have two certificates with different meanings and the same key ... so that if there is a dispute ... the signer can repudiate the transaction by saying that the sender had sent the certificate with the "random data" bit set ... but somebody substituted another certificate w/o the random data bit set .... unless somebody gets a law passed saying a signer can't repudiate a transaction if they ever had two different certificates issued for the same key. Even at that, certificates don't include the signature on the data in a certificate by the sender ... only by the CA ... what is to prevent the claim that some CA generated an arbritrary certificate containing my public key ... w/o adequately prooving the corresponding private key?

so the movement of all flags/indications ... especially related as to the intention/belief of the sender associated with a specific message ... into the signed body of the message. In effect, if there is some trust issue as to what I believe or intend ... then I need to have explicitly signed it. If there is some dependency on some state with respect to some specific signed message/content ... then the signer can repudiate it if it isn't covered by the signature. That is somewhat independent of whether a signer can also repudiate stuff they've signed. This is specifically that signers can repudiate a "not random bit" in an appended certificate that isn't actually part of the signed message.

denis pinkas on 5/6/2002 10:42 am wrote:
It is the problem of the signer (not of the CA) to make sure that when it uses these two different certificates, two different keys are being used.


Federated Identity Management: Sorting out the possibilities

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/06/2002 09:49 PM
To: "Arnold G. Reinhold" <reinhold@xxxxxxxx>
cc: cryptography@xxxxxxxx, dcsb@xxxxxxxx
   mac-crypto@xxxxxxxx, "R. A. Hettinga" <rah@xxxxxxxx>
Subject: Re: Federated Identity Management: Sorting out the possibilities
I believe RBAC is supposed to be more like at lattice or complex graph.

basically RBAC has been a KISS solution for security administration ... some number of fine grain entitlements have been worked out for groupings that match stylized job descriptions or roles. In theory, the issues of things like insider fraud have all been worked out for the idealized RBAC cases. It is possible to enumerate the fine grain entitlements with a large matrix ... however the challenge for RBAC systems is working out all the interdependencies and security/integrity failure modes ... which is more of a complex graph.

Given that the insider fraud and security/integrity failure modes have been worked out for the idealized single role cases ... then where are the points of failure when there is possible collusion ... aka when failure security modes involve the coorperation of two or more individuals ... actually two or more roles .... either involving collusion between individuals or situations were security administrators have to assign two or more roles to the same individual because the idealized scenarios don't actually handle all possible real world cases.

fine grain entitlements grouped for RBAC roles then become sub-graphs of a complex aggregate graph, aka real issues of RBAC-like operations isn't representing the fine grain entitlements as either matrix or sparse matrix but the complex interrelationships and implications of the entitlement aggregations (aka roles). the interrelationships of entitlement aggregations is the things that lead towards complex graphis ... and then towards multiple subgraphs when dealing with multiple role aggregates. strict object encapsulation can fail to recognize the interaction of specific nodes (using graph sematnics) of individual entitlements; aka RBAC roles (entitlement aggregations) are paradigm convinience for security administrators ... however the fraud and integrity/security failures involve interactions of the individual entitlements (aka KISS RBAC encapsulation for security administrators doesn't obsolve security tools for dealing with fine-grain entitlement interactions).

rows & columns show up in relational databases and spreadsheets .... paradigms oriented towards addressing relatively straighforward accounting problems. real-world problems frequently have problems being force fitted to relational or spreadsheet models. Even RDBMS implementations with operations encomposing multiple databases ... with each database potentially have large number of tables ... normally can't represent the federated data dictionary for all the tables in all the databases using straight-forward relational semantics.

adding/deleting a single entitlement in a sparse matrix shouldn't be the objective. adding/deleting an entitlement with all its complex interactions with other entitlements gets to be an issue. start is simple pair-wise issues (somewhat like simple drug interactions) ... and then proceed to more complex integrity/security failure modes involving complex interactions of three or more entitlements.

other paradigms for cateloging and maintain complex information may be required .... like something like we do in attempting to mirror the IETF process http://www.garlic.com/~lynn/rfcietff.htm

or the complexities of taxonomies (& glossaries): http://www.garlic.com/~lynn/index.html#glossary

btw, the merged security glossary in the above was updated with the nsa intrusion glossary this weekend.

on 5/6/2002 1:19 pm wrote:
At 12:46 PM -0400 5/6/02, R. A. Hettinga wrote:


>http://www.simc-inc.org/archive0002/February02/Speakers/geer-keynote.htm
>
>
>Logging in to Wall Street
>Federated Identity Management: Sorting out the possibilities
>SIMC's Third Annual Security Meeting
>February 26, 2002
>
>Keynote Address
>
>Dr. Daniel Geer
>
>Chief Technology Officer
>@stake
>

This is an interesting talk, but I'd question one assumption. Dan writes:

...
>If you look at the access control problem, it is a matrix. The rows are
>requestors and the columns are objects of their desire. Linear growth in
>either or both means geometric growth in the number of table entries in
>that matrix. Oh, sure, you can cut down the number of rows through
>role-based collective nouns and you can cut down the number of columns
>through the object-oriented sense of the word inheritance, but the natural
>path is that of geometric growth in the matrix even where interrupted
>occasionally by a tactical fall back like RBAC.

Seems to me that this is a sparse matrix. Most requesters have a default status with regard to most objects. So the storage size of this matrix is the number of requesters times the number of managed objects per requester. That doesn't seem nearly as unmanageable. And presumably the objects being managed have some value. The cost of adding and maintaining one more cell to store an individual's status regarding a particular object should simply be added to the cost of her gaining access to that object.

Arnold Reinhold


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/07/2002 11:28 AM
To: Sharon Boeyen <sharon.boeyen@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx,
    <osidirectory@xxxxxxxx>,
    "Bill Burr (E-mail)" <william.burr@xxxxxxxx>
Subject: Re: Meaning of Non-repudiation
so the bit in the certificate just means that the CA dsavows all knowledge of the private key ... aka "private key not divulged".

then possibly if it can be prooven that the private key was also not divulged to anybody else ... then some conditions for meeting non-repudiation requirements might be met (i.e. a CA not having knowledge of the private key is several steps removed from non-repudiation).

as aside ... definition for non-repudiation (from nsa intrustion glossary) and non-repudiation sevice (from rfc2828) are in merged security glossary:
http://www.garlic.com/~lynn/secure.htm

... quick path is to click on "PAIN" in Acronym fastpath ... and then click on non-repudiation.

on 5/7/2002 7:28 am wrote:
I'd like to introduce another perspective on this topic, one that I really think helps add some fundamental clarity to the topic. During the US FPKI TWG meeting last week, this topic was raised and a short discussion was held. Bill Burr commented that he understood the meaning of the non-repudiation bit being set to be a statement by the certificate issuer (CA) that they at no point had any access to the private key corresponding to the public key being certified. I really like this because it states what role this bit plays in a non-repudiation context. It is simply that and nothing more. The remaining mechanisms for supporting a non-repudiation service are outside the scope of the definition of this bit. Legal and policy schemes can determine the behaviour of relying parties and users when this bit is set. The bit itself cannot do that. If an assertion that a user who signed something "knew and intended to sign whatever", then that assertion should be something that accompanies a specific signature. This bit cannot be expected to act as that assertion.

I also agree with Stefan that we need describe the digital signature bit separate from the description of the non-repudiation bit in 509. Then it is left to profiling groups, as it shoud be, what combinations of bits they want set in their own environments.

Re the debate on changing the meaning - The reason we are having this discussion (at least one of the reasons) is because the meaning of these bits is not clear to many readers. Both the process of defect reports as well as the enhancement process allows us to clarify text in the standard, as well as to fix bugs etc). Part of the problem is that there isn't agreement on what the original text meant, hence the clarification. Once there is some sort of agreement on what the "right thing to do" is, then we'll determine what the process is to achieve the intended result. That's part of the reason why we are having the discussion before writing up a DR or an enhancement request.

Sharon

Sharon Boeyen
Principal, Advanced Security
Tel: 613 270 3181
www.entrust.com


Entrust, Inc. Securing the Internet


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/07/2002 02:21 PM
To: ietf-pkix@xxxxxxxx,
    "500 list (E-mail)" <osidirectory@xxxxxxxx>,
    Sharon Boeyen <sharon.boeyen@xxxxxxxx>,
    "Bill Burr (E-mail)" <william.burr@xxxxxxxx>
Subject: Re: Meaning of Non-repudiation
sorry, finger fumble with the keys ... or the fingers moving faster than the brain.
http://www.garlic.com/~lynn/secure.htm

is the merged security glossary.

for more info about merged glossaries and sources see:
http://www.garlic.com/~lynn/index.html#glossary

lynn.wheeler@xxxxxxxx on 5/7/2002 11:28 am wrote:
as aside ... definition for non-repudiation (from nsa intrustion glossary) and non-repudiation sevice (from rfc2828) are in merged security glossary:
http://www.garlic.com/~lynn/secure.htm

... quick path is to click on "PAIN" in Acronym fastpath ... and then click on non-repudiation.


Words, Books, and Key Usage

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/07/2002 05:21 PM
To: "Trevor Freeman" <trevorf@xxxxxxxx>
cc: "David P. Kemp" <dpkemp@xxxxxxxx>,
     ietf-pkix@xxxxxxxx
Subject: RE: Words, Books, and Key Usage
however, it is possible that a label for a bit has some semantic meaning that is no way related to the feature/function implemented for the bit.

If the bit means that the CA has never seen the originator's private key ... it is possible for the CA to certify that by turning the bit on in a certificate, signing the certificate and attesting to the fact that the CA has never seen the originator's private key.

If the originator wants to have a certificate that has a bit turned on indicating the originator's intention not to have ever shared the private key with anybody ... then possibly, the originator can sign a message sent to the CA with that assertion; the CA just copies the assertion to its certificate ... not actually certifying anything other than it has copied some assertion by the originator into the certificate.

The NSA Intrustion glossary
non-repudiation
Method by which the sender of data is provided with proof of delivery and the recipient is assured of the sender's identity, so that neither can later deny having processed the data.

obviously, the CA turning on a bit in a certificate doesn't meet that definition. somewhere lurking behind the scenes is the stuff about being able to disproving non-repudiation by showing that entities other than the sender may have had access to the private key. Having the CA certify that they have no knowledge of the private key (at least at the time of the creation of the certificate) ... isn't an exhaustive proof that there is nobody else that might have knowledge of the private key (or that the CA might acquire knowledge of the private key in the future).

earlier parts of this thread also brought up non-repudiation in the context of "processing" signed data ... which might imply some agreement or intention regarding any possible semantic meaning of the data processed. The simplest is being able to demonstrate that the sender sent the data ... w/o implying anything about any knowledge, intention, and/or agreement that the sender might have with regard to the semantic meaning of the data transmitted. Being able to show that somebody else could have sent the message ... would obviously negate assertions about non-repudiation. However, not being to find anybody else with knowledge of the private key ... isn't necessary proof that there isn't somebody else with knowledge of the private key. Also, even if it is possible to proove that nobody else has knowledge or access of the private key ... doesn't proove that the sender did anything else but transmit the data .... it doesn't proove any knowledge or processing of the data ... other than the transmission.

in any case, the bit seems to represent a meaning that isn't consistent with the meaning of the label applied to the bit (aka non-repudiation).

somewhat total aside did have a booth a couple weeks ago in new orleans at cardtech/securetech with a hardware token where there is public/private key-pair generated in the chip ... and the private key is never divulged (leaves the chip) ... basically aads chip strawman
http://www.garlic.com/~lynn/x959.html#aads

with couple twists ... like PIN-mode (and shortly biometric-scoring-mode) operation ... and able to operate in both access & financial token personality mode (aka earlier post in this thread):
http://www.garlic.com/~lynn/aadsm11.htm#5

<trevorf@xxxxxxxx on 5/7/2002 2:55 pm wrote:
Dave,

Semantics
1) The branch of linguistics and logic concerned with meaning
2) The meaning of a word, phrase, sentence or text. Concise Oxford English Dictionary - Queens English edition

If you are changing the meaning of bits when they are asserted in an extension, then its semantics not structure we are dealing with. I don't dispute that the semantics behind non-repudiation are a little fuzzy to say the least, but there are some basic ground rules in there which have been agreed and included in RFCs. If you want to define new semantics with cleaner definitions, feel free to do so, but you cannot do so within the confines of the existing extension as identified by the existing OID. I don't have a problem with what you are doing, just don't do it in such a way as you break existing standards.

Trevor


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/11/2002 11:44 PM
To: Sharon Boeyen <sharon.boeyen@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx, "500 list (E-mail)" <osidirectory@xxxxxxxx>,
    epay@xxxxxxxx, "Bill Burr (E-mail)" <william.burr@xxxxxxxx>
Subject: Re: Meaning of Non-repudiation
while i was at it, i thot i would go ahead and update the merged security glossary with the latest from ISO SC27 http://www.garlic.com/~lynn/secure.htm

some of the new defintions (from merged security taxonomy/glossary; i'm still working on taxonomy for many of the new terms in the latest sc27 glossary).


non-repudiation

A security service by which evidence is maintained so that the sender of data and recipient of data cannot deny having participated in the communication. [IATF] Method by which the sender of data is provided with proof of delivery and the recipient is assured of the sender's identity, so that neither can later deny having processed the data. [NSAINT] The ability to prove an action or event has taken place, so that this event or action cannot be repudiated later. [ISO/IEC PDTR 13335-1 (11/2001)] [sc27] The reasonable assurance that a principal cannot deny being the originator of a message after sending it. Non-repudiation is achieved by encrypting the message digest using a principal's private key. The public key of the principal must be certified by a trusted certification authority. [misc] (see also Generic Security Services API, IT security, NRD token, NRO token, NRS token, NRT token, assurance, defense-wide information assurance program, digital signature, distinguishing identifier, information assurance, invalidity date, non-repudiation exchange, non-repudiation information, non-repudiation of creation, non-repudiation of delivery, non-repudiation of knowledge, non-repudiation of origin, non-repudiation of receipt, non-repudiation of sending, non-repudiation of submission, non-repudiation of transport, non-repudiation policy, non-repudiation token, notarization token, originator, proof, recipient, sandboxed environment, secure single sign-on, certification authority, public-key infrastructure, quality of protection) (includes non-repudiation service, privacy, authentication, identification, integrity, non-repudiation, privacy, authentication, identification, non-repudiation)

non-repudiation exchange

A sequence of one or more transfers of non-repudiation information (NRI) for the purpose of non-repudiation. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation information

A set of information that may consist of the information about an event or action for which evidence is to be generated and validated, the evidence itself, and the non-repudiation policy in effect. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of creation

This service is intended to protect against an entity's false denial of having created the content of a message (i.e. being responsible for the content of a message). [ISO/IEC WD 13888-1 (11/2001)] Protection against an entity's false denial of having created the content of a message (i.e., being responsible for the content of a message). [ISO/IEC 15945: 2002] [sc27] (see also non-repudiation)

non-repudiation of delivery

This service is intended to protect against a recipient's false denial of having received the message and recognised the content of a message. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of knowledge

This service is intended to protect against a recipient's false denial of having taken notice of the content of a received message. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of origin

This service is intended to protect against the originator's false denial of having approved the content of a message and of having sent a message. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of receipt

This service is intended to protect against a recipient's false denial of having received a message. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of sending

This service is intended to protect against the sender's false denial of having sent a message. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of submission

This service is intended to provide evidence that a delivery authority has accepted the message for transmission. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)

non-repudiation of transport

This service is intended to provide evidence for the message originator that a delivery authority has delivered the message to the intended recipient. [ISO/IEC WD 13888-1 (11/2001)] [sc27] (see also non-repudiation)


Meaning of Non-repudiation

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/12/2002 03:24 PM
To: epay@xxxxxxxx, ietf-pkix@xxxxxxxx,
    "500 list (E-mail)" <osidirectory@xxxxxxxx>,
    Sharon Boeyen <sharon.boeyen@xxxxxxxx>,
    "Bill Burr (E-mail)" <william.burr@xxxxxxxx>
Subject: Re: Meaning of Non-repudiation
I've finished some more of the taxonomy .... SC27 is big on evidence, proof, audit trails and time-stamping.

try clicking on "evidence" (new to the concept section at the start of the glossary frame).

lynn.wheeler@xxxxxxxx on 5/11/2002 11:44 pm wrote:
while i was at it, i thot i would go ahead and update the merged security glossary with the latest from ISO SC27 http://www.garlic.com/~lynn/secure.htm

some of the new defintions (from merged security taxonomy/glossary; i'm still working on taxonomy for many of the new terms in the latest sc27 glossary).


international financial standards (ISO TC68)

From: Lynn Wheeler
Date: 05/13/2002 09:23 AM
To: internet-payments@xxxxxxxx
Subject: international financial standards (ISO TC68)
somewhat ABA, ANSI, & ISO fluff

===================================================

For Immediate Release

Contacts:
Cindy Fuller
Accredited Standards Committee,
X9, Incorporated
c/o American Bankers Association
(202) 663-5284
cfuller@xxxxxxxx


Staci Busby
First Data
(303) 967-7188
staci.busby@xxxxxxxx

Gene Kathol Named Leader of International Standards Group

Washington, 2002? Gene Kathol, vice president of global electronic commerce and payment services leader First Data Corp., has been named chair of Technical Committee 68 (TC 68) Financial Services, an official committee of the International Standards Organization (ISO) headquartered in Geneva, Switzerland. Mr. Kathol assumed the ISO post by unanimous acclamation of the TC 68 membership during their recent annual meeting held in Delft, Netherlands. The national standards organization, ASC X9, Inc. proposed Mr. Kathol from among its USA advisory committee to serve as chair for TC68.

The work of TC 68?to formulate banking, securities and other financial services standards?is important to assisting worldwide commerce. The global committee includes several working groups staffed by experts representing 56 countries who prepare standards made up of written processes and procedures for the banking and financial services industry.

Standards are documented agreements containing technical specifications and other precise criteria to be used consistently as guidelines, or definitions of characteristics, to ensure that products and services are fit for their purpose. To date TC 68 has prepared 72 global standards. The standards assist the international industry in ensuring that financial transactions are efficient, accurate, secure and in the best interest of the general public.

A fundamental example of an international standard is the credit card. The standard format specifies shape and thickness as well as communications elements for card processing. This assures that cards adhering to ISO standards can be honored worldwide.

(more)

Gene Kathol Named Leader of International Standards Group, page 2

Mr. Kathol's involvement in financial services standards began in 1985 when he became active in his first US standards group?Accredited Standards Committee X9, Incorporated (ASC X9?Financial Services) ? a group accredited by the American National Standards Institute (ANSI).

Over the past eighteen years, Mr. Kathol and his company, First Data Corp., have been very active in helping to establish financial industry standards. Mr. Kathol has participated in or chaired several domestic work groups, and held various management positions, including serving as chair of a national standards group ? the Retail Electronic Financial Transactions sub-committee from 1996 to 2001.

Committees TC 68 and ASC X9, Incorporated are the only industry forums that bring together bankers, securities professionals, manufacturers, regulators, associations, consultants and others from the financial services arena to address technical issues, find the best solutions and codify them as nationally and internationally accepted standards. The American National Standards Institute (ANSI) accredited the current secretariat in 1984, and the International Standards Organization (IS0) accredited the TC 68 secretariat in 1986.

For information on financial industry standards, please contact Cindy Fuller, via the Web site at http://www.iso.org/tc68 (international) or http://www.X9.org (domestic).


Alternative to Microsoft Passport: Sunshine vs Hai

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/13/2002 10:17 AM
To: arti@xxxxxxxx
cc: internet-payments@xxxxxxxx
Subject: Re: Alternative to Microsoft Passport: Sunshine vs Hai
one of the vulnerabilities identified by the X9A10 working group .... was that the account number was an unauthenticated shared-secret at all .... aka evesdroping at anypoint in the infrastructure could lead to fraudulent transactions. the most secure part of the current infrastructure was the SSL transport of the data-in-flight .... the least secure was all the data-at-rest copies of the account number thru-out the infrastructure (not just at the consumers private computer but at several other points in the infrastructure as well).

the approach taking by the working group in the x9.59 standards work was to make the financial transaction authenticated (with fips186-2/ecdsa) and only the account number was carried in the transaction .... and that account number could only be used in authenticated transaction. the result was that the account number no longer was a shared-secret ... and there was no longer any fraud concern regarding all the points in the infrastructure where that account number might appear as data-at-rest in the clear (not just the consumer's private PC).

random refs to various fraud/exploits:
http://www.garlic.com/~lynn/subtopic.html#fraud

a specific discussion/thread
http://www.garlic.com/~lynn/2001j.html#5 E-commerce security??

arti@xxxxxxxx on 5/13/2002 7:01 am wrote:
Brent,

All this is interesting (from a theoretical point of view), but forgive me if I think that the entire concept of your ActiveCheckout applet misses the boat - in the real world. It is a well-established fact that the least secure part of the "Internet world" is the majority of PCs from which average users access cyberspace; most users have neither the skill NOR THE INTEREST to properly protect their systems, and the advent of DSL and cable modem connection (with their permanent I.P. addresses) has obviously made the situation even far worse!

As a result, the worst possible tack to take, IMHO, is to maintain users' personal information (credit card numbers, etc.) on their own machines - you might as well tattoo that information on their foreheads! Frankly, I believe that your company has made exactly the same mistake as Microsoft, which caused them to become so vilified - you have emphasized convenience over security. Try to imagine that it is not necessarily a "good thing" for users to have that level of convenience when purchasing something online - and before you jump into the fray and start arguing this point with me from a theoretical basis, take note of the fact that, in a recent and very comprehensive study carried out by U.C.L.A., it was found that 91% of online users rated themselves as being somewhat or totally uncomfortable with the idea of passing their financial information online at all! That is NOT a minor blip which can be overcome by convincing all those misguided people that they're wrong to be afraid, because of this "great new technology" or that ...

Ask me sometime if I believe that there will EVER be a secure way to store or transmit personal information safely online ...

Cheers,

John Vinokur
President
Payment Central Inc.
mailto:john@xxxxxxxx
---Original message---
BR>As you might agree, the business of authenticating internet
BR>payments needs to stay within a trusted domain such as financial
BR>institutions rather than a technology company such as Microsoft.

BR>However, most banks do not want to develop, manage or support
BR>additional technology for authenticating consumers over Internet
BR>channels.

BR>To this end we have developed a FREE applet, ActiveCheckout, which
BR>gives online consumers control of their identity including their
BR>credit card information. This allows the consumer to store their
BR>information on their local machine, rather than on Microsoft's
BR>Passport servers, and assists in transferring this information to
BR>websites during registrations or online purchases.

BR>The applet has the ability to communicate with issuing bank(s) for
BR>realtime authentication during online purchases. It currently
BR>supports MasterCard's SPA standard and we are in discussion with
BR>Visa regarding support for Verified by Visa (3-D secure). This
BR>could improve the security of Verified by Visa transactions by
BR>reducing the possibility of "man-in-the-middle" attacks.

BR>ActiveCheckout can be installed from http://checkout.gpayments.com/ and
BR>further information can be found at
BR>http://checkout.gpayments.com/faq.htm

BR>Our ActiveCheckout project was codenamed "Sunshine" as a direct response to
BR>Microsoft's Hailstorm initiative and you can even download a Sunshine character animation for the applet.

BR>We are always impressed with the quality of contributions made
BR>through this forum and consider the members to be among the leading
BR>thinkers in the ePayments industry. We would greatly appreciate and
BR>seriously consider any feedback that you may have regarding this
BR>applet approach to identity and authentication management for
BR>Internet payments.

BR>Regards,

BR>Brent Clark
BR>GPayments


IBM alternative to PKI?

From: Lynn Wheeler
Date: 05/15/2002 01:50 PM
To: <Liaquat.Khan@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx
Subject: RE: IBM alternative to PKI?
there have been a number of things done for relying party operations. for various of the relying-party-only certification authority pilots done in europe and other countries .... it was possible to show that the public key could be registered ... and it wasn't even necessary to send the certificate back to the key owner .... just put it in the relying partie's account record (i.e. they were sending the certificate back to the key owner ... because there was some COTS software that could be used ... but other than that it served no useful business function).

is this in anyway related to the previous work on relying-party-only pilots that have been done in europe?

on 5/15/2002 1:10 pm wrote:
This model came out of the work down by IBM and others as a part of the TIE (Trust Infrastructure for Europe) project. It is based on public key cryptography, but looking at things from a slight different angle, i.e. from the viewpoint of the RP. I need to be careful on how much I can say. Although there maybe publicly available information out there somewhere.

Regards,
Liaquat Khan


IBM alternative to PKI?

Refed: **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/16/2002 08:48 AM
To: "Anders Rundgren" <anders.rundgren@xxxxxxxx>
cc: "Dean Adams" <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: Re: IBM alternative to PKI?
my impression is that server-side signings are typically the desire for putting in a "real" PKI infrastructure .... but single, large roll-out of all the technology is an issue ... so the backends implement real PKI, and there is a middle-man which interacts with the clients in a more traditional (password) paradigm and the middle-man acts as a proxy doing the PKI stuff on behalf of the client. At any point, individual clients might be able to deploy their own PKI operation and eliminate using the proxy (theory is that it facilitates the transition from password-based paradigm to PKI-based paradigm by not requiring all backends and all clients to make the paradigm transition in a single, synchronous event).

Many of the EU and other relying-party-only PKI deployments were addressing a different issue. They came to realize that the traditional PKI identity certificates didn't address any of their business needs ... the client was doing digital signature operations and appending things that bore a little resemblence to identity certificates ... enuf so that traditional PKI sottware might work. However, other than the facilitating of traditional PKI software ... the certificates weren't not serving any actual business function.

The first case of server-side PKI isn't so much a PKI alternative issue ... it is a traditional PKI roll-out/transition facilitor. I believe the second case of experience from relying-party-only certificates ... could be considered more of a PKI alternative issue (i.e. discovery that certificates weren't serving any valid business function).

anders.rundgren@xxxxxxxx on 5/16/2002 6:53 am wrote:
Liaquat
>This model came out of the work down by IBM and others as a part of the TIE
>(Trust Infrastructure for Europe) project.  It is based on public key
>cryptography, but looking at things from a slight different angle, i.e. from
>the viewpoint of the RP. I need to be careful on how much I can say.
You just triggered my curiosity!

Is it fundamentally different to 3D Secure and SAML? I.e. schemes where the client authenticates to a server-based PKI-thing that does the actual work (signs resp. creates auth).

Anders


IBM alternative to PKI?

Refed: **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/16/2002 10:59 AM
To: "Anders Rundgren" <anders.rundgren@xxxxxxxx>
cc: "Dean Adams" <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
     Liaquat.Khan@xxxxxxxx
Subject: Re: IBM alternative to PKI?
i.e. middle-man tends to represent transition/roll-out solutions .... not security infrastructure solutions.

from traditional security

end-to-end

middle-man approaches tend to always negate any end-to-end attribute

additional steps represent additional vulnerabilities

middle-man approaches increase the number of steps/processes .... each of which represent an additional vulnerability, each additional vulnerability represents additional points of failure/fraud

only as strong as the weakest link

shared-secret password paradigm in series with digital signature isn't additive.

2-factor authentication

secret password (as opposed to shared-secret password) in conjunction with hardware token (i.e. processing in parallel instead of in series) represents a combination link (both hardware token and secret password has to be compromised). a shared-secret password process in series with a digital signature process can fail with just the shared-secret password.

IBM alternative to PKI?

Refed: **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/16/2002 02:17 PM
To: Trevor Perrin <Tperrin@xxxxxxxx>
cc: Anders Rundgren <anders.rundgren@xxxxxxxx>,
    Dean Adams <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: RE: IBM alternative to PKI?
but purely monitor/audit isn't usually a transition scenario also.

if you are looking at the financial "middle-man" given in the example by anders ... the financial end-point already contains extensive audit and dynamic risk management (i think there was an article in NYTimes this week or last week about the neural net stuff that looks for complex fraud patterns). Given end-to-end integrity and no intermediate stand-in .... I would contend that the end-point risk management can do a much better job of deciding whether or not to approve a transaction ... based on more comprehensive understanding of the situation ... that wouldn't be available to middleman processing.

the issue is if you have a no-security infrastructure ... then adding a monitor for auditing purposes can be a risk management benefit. we defined a bunch of that stuff back in the '80s for something we were calling "middle-layer" ... sort of the precursor to 3-layer architecture.

Note however, this doesn't have to be a middle-man in the transaction processing sense ... in the current internet ... a packet can flow thru a large number of nodes ... all of which can be doing monitoring and auditing (do a traceroute sometime) ... but wouldn't be considered a transaction processing middle-man ... in the sense that it takes in a transation ... processes the semantics of that transaction and then generates a different transaction.

The scenario example is that a client does a password transaction with a middle-man ... the middle-man processes the password transaction and then generates a totally different digitally signed transaction ... possibly even emulating that the transaction originated directly from the client.

A purely intermediate monitor/audit environment with no "middle-man" processing and end-to-end security would have the client directly generating the digitally signed transaction and sending it all the way thru to the processing end-point. It isn't the monitoring/audit that descreases security, it is any middle-man processing interferring with end-to-end integrity.

Another scenario was "SET". SET had relying-party-only certificates with digitally signed transaction. A "SET" gateway .... accepted incoming SET financial transactions, verified the certificate, verified the signature, stripped everything out ... and generated a standard ISO 8583 transactions .... with an additional flag indicating whether or not the "SET" gateway believe the certificate and signature to be correct. Then a consumer's financial institution was dependent on the "flag" in deciding whether or not it was a valid transaction (aka there wasn't any end-to-end integrity). Furthmore, I don't know if any of the "SET" gateways that had any financial liability associated with fraudulent transaction (i.e. what happens if they were to lie). At one point one of the associations gave a talk at an ISO meeting giving the number of 8583 transactions that had the "SET" valid signature flag set .... and they could proove that there was no cryptographic capability involved at all (i.e. it wasn't even a SET gateway that might not be telling the truth ... but there were financial reasons why others might want the flag set also).

now, I would claim that an end-point business operation might compartmentilize some of its functions ... so that any compromize in any single component doesn't compromise all components. i would contend that a compartmentilzed end-point security solution is different than several different business operations all implementing various kinds of intermediate processing (middle-man) preventing any kind of end-to-end integrity.

random old middleware/middle layer refs:
http://www.garlic.com/~lynn/96.html#16 middle layer
http://www.garlic.com/~lynn/96.html#17 middle layer
http://www.garlic.com/~lynn/98.html#50 Edsger Dijkstra: the blackest week of his professional life
http://www.garlic.com/~lynn/99.html#123 Speaking of USB ( was Re: ASR 33 Typing Element)
http://www.garlic.com/~lynn/99.html#124 Speaking of USB ( was Re: ASR 33 Typing Element)
http://www.garlic.com/~lynn/99.html#201 Middleware - where did that come from?
http://www.garlic.com/~lynn/99.html#202 Middleware - where did that come from?
http://www.garlic.com/~lynn/2000b.html#59 7 layers to a program
http://www.garlic.com/~lynn/2000e.html#45 IBM's Workplace OS (Was: .. Pink)
http://www.garlic.com/~lynn/2001d.html#69 Block oriented I/O over IP
http://www.garlic.com/~lynn/2001j.html#4 I hate Compaq
http://www.garlic.com/~lynn/2001j.html#20 OT - Internet Explorer V6.0
http://www.garlic.com/~lynn/2001k.html#18 HP-UX will not be ported to Alpha (no surprise)exit
http://www.garlic.com/~lynn/2001m.html#15 departmental servers
http://www.garlic.com/~lynn/2001n.html#23 Alpha vs. Itanic: facts vs. FUD
http://www.garlic.com/~lynn/2001n.html#34 Hercules etc. IBM not just missing a great opportunity...
http://www.garlic.com/~lynn/2001n.html#55 9-track tapes (by the armful)
http://www.garlic.com/~lynn/2002.html#2 The demise of compaq
http://www.garlic.com/~lynn/2002.html#7 The demise of compaq
http://www.garlic.com/~lynn/2002.html#11 The demise of compaq
http://www.garlic.com/~lynn/2002b.html#4 Microcode? (& index searching)
http://www.garlic.com/~lynn/2002b.html#37 Poor Man's clustering idea
http://www.garlic.com/~lynn/2002d.html#4 IBM Mainframe at home
http://www.garlic.com/~lynn/2002d.html#14 Mainframers: Take back the light (spotlight, that is)
http://www.garlic.com/~lynn/2002e.html#2 IBM's "old" boss speaks (was "new")



tprerin@xxxxxxxx on 5/16/2002 1:33 pm wrote:
Hi Lynn (and greetings to PKIX, 1st-time poster here),

A counter-argument: while adding a middleman adds a vulnerability, it also adds auditing: the middleman can monitor all password attempts, so as to:
- lockout guessers
- detect anomalous patterns of use which may indicate a compromised password is being exploited
- allow users to review audit trails to detect compromises themselves
- allow users to review audit trails to determine the extent of a compromise

Without this auditing, it may be much more difficult to detect a compromise of a private key, and to determine precisely what has been compromised.

So there are security benefits as well as disadvantages to middlemen, I suppose a comparison depends on the exact situation and how you weight the factors..

Trevor


IBM alternative to PKI?

From: Lynn Wheeler
Date: 05/16/2002 03:15 PM
To: Trevor Perrin <Tperrin@xxxxxxxx>
cc: Anders Rundgren <anders.rundgren@xxxxxxxx>,
    Dean Adams <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: RE: IBM alternative to PKI?
so lets do the counter argument in the business process sense ... a middle-man that is a different company than the end-points (as opposed to possibly middle-man implementation which is actually a business entity's compartmentalized operation ... which has various security/vulnerability trade-offs).

so in the business middle-man scenario ... the middle-man is your strongest competitor/advisary. they process incoming transactions and then generate something from scratch that is forwarded to you. you have no way of prooving that the incoming transactions are fraudulent or not. furthermore the business middle-man has no financial penalty/liability with regard to fraudulent transactions.

Proxy PKI. Was: IBM alternative to PKI?

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/17/2002 05:02 AM
To: "Anders Rundgren" <anders.rundgren@xxxxxxxx>
cc: "Dean Adams" <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: Re: Proxy PKI. Was: IBM alternative to PKI?
you can do server-based PKI in conjunction with multiple authorization operations .... i.e. somebody uses a password to talk to a server-based PKI ... and the server turns a password operation into a PKI operation at the same time doing other organizational functions. However, it would be a mistake to assume that because two different operations are being performed by the same server ... that one kind of operation (password to PKI translation) also inherits the other kind of operation (say multiple approval level operations).

The counter example is an end-to-end security implementation where the original client signs the transactions ... and if there are multiple approvals required because of business reasons ... that the other agents (human or software) along the way also sign the same transaction ... providing end-to-end security for both the original transaction as well as any intermediate approval steps.

If a organization has both a requirement for multiple level approval and can't deploy real PKI out to the end clients ... that they could implement both business functions in a single data processing unit.

Various scenarios for end-to-end security and multiple levels of approval/signing are

1) in the NR thread in this list, I made reference to the EU FINREAD standard for token acceptor device having secure display and secure pin-entry ... as part of trying to close some of the NR-service gap with respect to human intention .... i.e. all operations requiring some indication of human intention in support of any NR ... required that every signed transaction need to have been done with a hardware token in conjunction with a FINREAD terminal configured such that every transaction signing first required a person to first enter a PIN. Provisions were made in the x9.59 standard that not only would the originating client sign the financial transaction ... but the "environment" that the signing took place in would also sign the operation (i.e. not only could you claim that an EU FINREAD certified device was used for the signing environment ... but you could proove that an EU FINREAD certified device was used for the signing environment .... because the certified device also signed the transaction).

2) various kinds of workflow systems ... may require two or more levels for final approval for purchase orders. The purchase order business protocol can have that external business entities don't have to actually authenticate the lower-level or originating digital signatures .... but just the final approval agent that is responsible for interfacing to external organizations. However, it can be useful to carry the originating and intermediate digital signatures along with the transaction as an internal audit trail.

Now it would be possible to implement such a function where all the internal corporate processes were password based and that only purchase orders (or other types of transactions) that actually left the corporate boundaries carried a digital signature .... in effect the digital signature of the final approval agent. I would claim that any use of a unique signature per originating employee ... rather than the digital signature of the approval function (human or software) is an artificial fabrication. In this particular scenario, the trust is between all the external business processes and the unit performing the signing ... any use of unique signatures per originating employee rather than a single signature of the signing authority is an artificial fabrication. The only purpose of such an artificial fabrication might be because we want to pretend that the trust relationship is directly between the external units and the individual employee (because there is a transition plan that employees will be able to directly sign individual transactions and send them directly to external agencies w/o any individual approval authority). Otherwise the artificial fabrication is merely obfuscation ... internal audit procedures will be able to tie individual password transactions to specific approval transactions; each PO has unique identifier and audit trail of all steps including relating to the originating employee and their password transaction.

===============================================================

Bifurcating the end-to-end trust relationship with some passwords and some PKI ... implies that there is different levels of trust between the password-based trust relationship and the PKI-based trust relationship. The bifurcation is either because 1) we are planning on transitioning to a real end-to-end PKI relationship .... in which case we create the artificial appearance of each originator signing with a unique key as an aid to that transition or 2) there isn't any "real" end-to-end trust relationship ever required, that the trust infrastructure really is partitioned .... that password is sufficient for "internal" operation and only the interface agent is ever going to be signing, in which case only a single key is needed by that interface agent. Since there is never going to be any real end-to-end trust relationship ... the signing agent only needs a single key. Having the server/signing agent sign with a different unique key per originator is an artificial fabrication since there is actually no end-to-end trust (or planned to be).

My original statement is that the server-based PKI is either an a) aid to transition to individual key signing ... in which case there is a point of creating the fabrication that there is some end-to-end trust or b) a permanent solution ... if it is a permanent solution there is no real point in creating the artificial fabrication of an end-to-end trust relationship (with different key per originator) ... the trust relationship is between the signing agent and the external agencies ... for the purposes of trust ... it is only sufficient that the signing agent sign the operation because it would be a total fabrication that there is an end-to-end PKI trust relationship between the external entities and the individual. A properly designed business process designed PKI-trust operation would have the keys belonging to the operational entities being trusted. Any implication that the PKI-trust boundaries are different (i.e. server signing agent using individual associated signing keys) is an artificial fabrication that would need some valid business justification ... like the planned transition to real end-to-end PKI-trust boundaries. If there is never going to be any real end-to-end PKI trust relationship ... and the trust relationship is only between the signing agent and the external entities then I would claim that multiple different keys (one per originating employee) is superfluous.

There is also always the danger when creating artificial fabrications that somebody might incorrectly believe there is a real end-to-end PKI trust relationship that doesn't really exist. Business executives will typically sign-off a risk acceptance when creating artificial fabrication when it is shown that it is a temporary solution pending transition to real implementation.



anders.rundgren@xxxxxxxx on 5/17/2002 3:31 am wrote:
Lynn,

That was a rather prejudiced description of server-based PKI.

There must be a reason why 42M people use Internet-banks just in Europe. These banks serve as giant "proxies" and I don't see how Internet-banks could ever be replaced by client-side PKI solutions! Do you?

I.e. the use of a "middle-man" or "proxy" has other qualities than just enabling PKI-transition/roll-out.

Another example are B2B-systems, where clients must perform actions through the local business system which enforces the authorization and policy rules. In addition to storing in- and out- going business-messages. A structure that is in daily use since at least 20 years back or more, so it is definitely time-proven.

And what's more, employees' privacy is protected by this arrangement.

By using server-PKI, companies and banks can abandon expensive leased-lines and use the Internet. Probably with an increase in security compared to today. Message integrity control is essentially for free using PKI.

Server-PKI actually supports end-to-end security as well, although there are two pair of end-points: client-to-proxy, proxy-to-rp. When/If client-side PKI becomes ubiquitous, it just makes the link between the proxy and client stronger. Including NR support in case the proxy want to sue the client for malpractice. Or the reverse BTW.

I think the real discussion is really what constitutes an "end-point". In B2B it does not have to be an individual as an individual is not a company. Few PKI-lawyers understand this as companies cannot "sign" in the paper-world. But in the "e-world" that's a piece of cake!

In case you want to try a B2B-system implementing server-based PKI, using XML-Signatures for all B2B-communication, you are invited to try out https://buyer.x-obi.com/BuyerASP/buyer Properly written, and protected by firewalls, HW-crypto, and secured physical facilities, such systems can work wonders. As Internet-banks already do, and that on a massive scale.

The only "business" that so far has not fully grasped the usefulness of using server-PKI is the public sector, but I stay confident that they soon will, as running a government authority should be rather similar to running a company. I.e. there are internal and external communication, policies etc. that means that you must have a "proxy" somewhere in your organization to maintain this in a reasonable way. The net effect is the external communication will be performed by the proxy rather than by the employees.

Cheers,
Anders


Proxy PKI. Was: IBM alternative to PKI?

Refed: **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/17/2002 06:17 AM
To: "Anders Rundgren" <anders.rundgren@xxxxxxxx>
cc: "Dean Adams" <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: Re: Proxy PKI. Was: IBM alternative to PKI?
note that i believe all the original ipsec stuff was end-to-end .... lightweight ipsec i believe was introduced with VPN at a router working group at the fall '94 IETF meeting in San Jose (by somebody who had originally built one for their personal use).

VPNs have some of the characteristics you claim for proxy PKI ... but they don't claim to be signing stuff on behalf of other entities ... they have a specific trust model and the PKI mirrors the trust model of the units involved in the trust operations. There is no artificial fabrication that the VPN unit that does the signing as part of a PKI trust operation is some totally other entity pretending to fabricate some totally different trust operation.

VPNs also allow corporations to replace expensive leased lines with internet connections.

bank-to-bank financial transfers don't happen under the pretend signatures of the individual account holders. bank settlement has uthentication between the two bank entities (either implicit, possibly because of leased line, routing code, etc ... or explicit ... the banks actual authentication process). As part of a bank-to-bank transfer ... banks will include adenda records that break out the individual accounting detail.

Proxy PKI. Was: IBM alternative to PKI?

From: Lynn Wheeler
Date: 05/17/2002 01:06 PM
To: Trevor Perrin <Tperrin@xxxxxxxx>
cc: Anders Rundgren <anders.rundgren@xxxxxxxx>,
    Dean Adams <da@xxxxxxxx>, ietf-pkix@xxxxxxxx,
    Liaquat.Khan@xxxxxxxx
Subject: RE: Proxy PKI. Was: IBM alternative to PKI?
1) there isn't end-to-end integrity ... there is a chain of trust .... that has to be stiched together to approx. some end-to-end operation; then we are back to the original point about the security being only as strong as the weakest link.

2) you can either be using password/pki security bifurcation as

a) transition solution not a security solution (as in the original post) or

b) the approving agency signs something that the external agencies trust ... (aka as in VPN) and the external agencies need to have absolutely zero awareness about the internal machinations and internal trust relationships, and/or

c) creating a total fabrication that individuals are really signing stuff when they aren't.

The difference between "b" and "c" ... is that the relying parties are really relying on the signing agency ... in "b" there is a one-to-one relationship between the signature and the trust relationship ... and in "c" the trust relationship is significantly obfuscated with lots of different key pairs that contribute nothing to the intrinsic trust infrastructure.

tperrin@xxxxxxxx on 5/17/2002 12:14 pm wrote:
Hi Lynn,

I agree that using a password<->PKI middleman doesn't provide end-to-end PKI trust. But it could provide end-to-end trust, in the sense that, when the middleman signs, he explicitly mentions the name being signed for, or when something is encrypted to the middleman, it explicitly includes the name the encrypted data is intended for.

Such a middleman would be a replacement for an identity certificate/key pair, but it would not "artifically fabricate" signatures by holding multiple keys, one for each client, but would instead make honest statements such as "I authenticated Alice with a password, and am signing this on her behalf".

The point, I guess, is that you can have end-to-end security (meaning authenticated or confidential communications with end-users) without end-to-end PKI, if you trust the PKI endpoints to make authentic statements from, or deliver confidential statements to, end-users..


Proxy PKI

From: Lynn Wheeler
Date: 05/22/2002 09:41 AM
To: Trevor Perrin <Tperrin@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx, owner-ietf-pkix@xxxxxxxx
Subject: RE: Proxy PKI.
and of course the other case for Proxy PKI is where there is a hierarchy of trust .... like in B-to-B scenario and the proxy may implement some number of business rules in addition to authentication on subordinate trust units (like a corporate purchasing department approval iinfrastructure). From a design standpoint the authentication paradigm (password or digital signature) on inbound requests from subordinate trust units should be transparent to the outbound digital signed requests. top-level trust units (say in b-to-b corporate scenario) shouldn't care about the internal trust mechanism and/or even that their are subordinate trust units. This is sort-of like inverting the standard CA trust hierarchy ... rather than the CA distributing static proxy trust certificates to each individual .... and each individual distributing the static proxy trust certificates with each of their signed messages .... the individuals send each authenticated message directly to the CA-equivalent ... and the CA does real-time rules in addition to the authentication operation before replacing the lower-level trust authentication with their own digital signature and sending it on. In this sort-of inverted trust paradigm, each message becomes a little like a real-time certificate (and the originating subordinate trust unit information doesn't even have to be propagated). The trust paradigm is still with the entity applying the (final) digital signature ... regardless of whether it chooses to use a single key-pair or a multitude of key-pairs creating the facade of multiple entities (simulating each of its sub-ordinate trust units).

Proxy PKI

From: Lynn Wheeler
Date: 05/22/2002 12:03 PM
To: Trevor Perrin <Tperrin@xxxxxxxx>
cc: ietf-pkix@xxxxxxxx, owner-ietf-pkix@xxxxxxxx
Subject: RE: Proxy PKI.
aka from trust standpoint, the described proxy PKI is a certification authority in disguise ... implying possibly it might be in need of things like CPS.

Proposal: A replacement for 3D Secure

Refed: **, - **, - **
From: Lynn Wheeler
Date: 05/27/2002 11:49 AM
To: Sebastian Kübeck <kuebeck@xxxxxxxx>
cc: internet-payments <internet-payments@xxxxxxxx>
Subject: Re: Proposal: A replacement for 3D Secure
the x9a10 working committee (part of x9 financial standards body) was given the charter of preserving the integrity of financial infrastructure for all retail transactions (not just internet, not just point-of-sale, etc.).

basically the result is X9.59 which is a light-weight digital signature authentication that can be mapped into international standard financial industry 8583 networks. basically it was done in such a way that it provides for end-to-end digital signature authentication and security ... being able to be mapped directly into the actual international standard 8583 transactions on an end-to-end basis from customer striaght thru to the customer's financial institution. It isn't specific to internet and it isn't specific to credit .... but is appicable to all retail payments (credit, debit, stored, value, atm, etc) in all environments (POS, internet, atm, etc).

the advantage is that the actual financial transaction is the thing being authenticated on an end-to-end basis. there isn't a truncated authentication that doesn't make it thru to the customer's (issuing) financial institution. There isn't a separate authentication transaction that is totally different from the actual financial transaction. The actual financial transaction carries the authentication of the financial transaction on an end-to-end basis. There isn't any separate need for a different routing/lookup mechanism for the authentication operation that travels totally different from the actual transaction nominally when talking about end-to-end authentication .... the normal reference is to the actual transaction .... some of these proposals creates a totally different transaction ... that flows via a totally different path ... and while it might get from the client/customer to the customer's (issuing) financial intitution ... it isn't an end-to-end secure financial transaction ...... it is a totally different transaction (and doesn't meet the nominal definition of end-to-end security).

random refs:
http://www.garlic.com/~lynn/x959.html#x959

also reference to the nacha/debit/atm pilot
http://www.garlic.com/~lynn/x959.html#aads

kuebeck@xxxxxxxx on 5/27/2002 at 1:43am wrote:
Hi Anders,

About the proposition:

I think the main problem of 3D Secure is that it's only an authentication mechanism and not a payment protocol at all. Compared to SET, the seperation auf authentication mechanism and payment protocol is a huge step back.

About core technology:

it's interresting that in Jannuary, we had a similar idea. However, we didn't think that the customer wants to remember the issuer's domain, so we proposed a lookup mechanism (similar to DNS) mapping the user's eMail adress to the domain of his/her issuing bank.

About the idea:

Great approach!

Yours,

Sebastian Kübeck
____________________________________________________
QENTA paymentsolutions      sebastian kübeck
www.qenta.com               kuebeck@xxxxxxxx
tel: +43 316 81 36 81-0     fax: +43 316 81 36 81-20


Proposal: A replacement for 3D Secure

Refed: **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/27/2002 02:09 PM
To: Anders Rundgren <anders.rundgren@xxxxxxxx>
cc: internet-payments@xxxxxxxx
Subject: Re: Proposal: A replacement for 3D Secure
... following is in response to a private communication with regard to X9.59 and information at the web site:
http://www.ca0.net/ Top Cat ANSI X9.59 Main Page

basically there is a signed payment instruction message ... that the client sends to the merchant (either at POS or internet).

The appendix describes how that message can be mapped into a standard 8583 transaction by the merchant (credit, debit ... or even those stored value things that work in the same POS network). the digital signature then flows with the standard 8583 message.

iso group responsible for 8583 has already approved the standard for the 8583 field to carry the x9.59 data.

when the issuing/client financial institution eventually gets the 8583 financial transaction ... they can directly verify the integrity of the financial transaction with the x9.59 digital signature.

an abstract of that is at
http://www.garlic.com/~lynn/8583flow.htm

a reference to the NACHA pilot for debit/atm network can be found by following the pointers at (even tho it was a nacha sponsored event ... it was with debit/atm network which is an ISO 8583 operation, as opposed to electronic ACH message format):.
http://www.garlic.com/~lynn/x959.html#aads

there has been some work by the people that worked on the FSTC electronic check in ACH to perform the same mapping from X9.59 to ACH electronic messages .... getting the same end-to-end intetrity for ACH electronic transactions (checks) at for 8583 networks (credit, debit, stored value, etc).

Note all of this work is for electronic retail transaction providing end-to-end integrity of security of the actual financial transaction (not some other end-to-end transaction that is different than the actual financial transaction). It was designed to be useable for all electronic retail transactions in all environments.

note the current predominate WEB infrastructure send credit card information from the client to the merchant. The WEB merchant takes that in transforms it directly into a 8583-like message that is pumped into the credit card network (via the merchants acquiring processor interface). The original WEB merchant implementations effectively emulated the same process that the electronic point-of-sale terminal used to interface into the merchant's acquiring processor interface. misc. refs:


http://www.garlic.com/~lynn/aadsm5.htm#asrn2
http://www.garlic.com/~lynn/aadsm5.htm#asrn3

The x9.59 specifies the data elements and the format of a signed message ... that the client/customer needs to send to the merchant (whether it is a POS terminal or a client internet machine).

The reference
http://www.garlic.com/~lynn/8583flow.htm

describes how the merchant processing (web or POS) takes the additional x9.59 information and includes it with the standard 8583 transaction that the merchant is already generating for the financial transaction.

what is not described is how the hundreds of thousands of web merchants and possibly tens of millions of POS terminals already work. what is described is the additional change to how it currently works to support x9.59 end-to-end authentication and security.

X9.59 wasn't attempting to create a brand new process and infrastructure. X9A10 group was attempting to define a ubiquituous standard for electronic retail payments (regardless of the kind) and the minimum changes necessary to all the existing electronic retail payments to add end-to-end integrity and security.

note that X9.59 standard also doesn't specify the actual client-to-CFI process .... in part, because X9.59 was being targeted for every retail electronic payment (not just internet & not just credit). A lot of work was put into X9.59 standard so that it could be incorporated into every existing retail payment ... regardless of internet, credit, debit, point-of-sale, stored-value, etc. and provide end-to-end security and integrity for that financial transaction .. as opposed to creating some other transaction which might or might not be in anyway related to the actual transaction.

Proposal: A replacement for 3D Secure

Refed: **, - **, - **, - **, - **, - **, - **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/27/2002 03:01 PM
To: Anders Rundgren <anders.rundgren@xxxxxxxx>
cc: internet-payments <internet-payments@xxxxxxxx>
Subject: Re: Proposal: A replacement for 3D Secure
<resend, possibly mailer problem, getting bounced error response to fstc.org listserve>

if we were worried about not doing something that wasn't already supported ... we probably wouldn't have worked on the initial implementation
http://www.garlic.com/~lynn/aadsm5.htm#asrn2
http://www.garlic.com/~lynn/aadsm5.htm#asrn3

note that today there are already a number of pretty transparent proxy operators for virtual POS terminals (aka in effect most current web e-commerce credit card transactions are run thru what is essentially a virtual point-of-sale terminal to generate an acceptable 8583 financial transaction message for the merchant's acquiring processor). Instead of the merchant running the actual the virtual POS terminals (or virtual cash registers) .... that part of the operation is outsourced to a 3rd party. Some number of these operations have provided client password based operation that filled in all the relavent credit-card information on behalf of the client/customer.

This 3rd party proxies are in use today. The mechant passes the customer off to the 3rd party "proxy" and they do their thing .... including any password oriented stuff that automatically injects the appropriate credit card information into the 8583 message before sending it off to the merchant's acquiring processor.

Applying the proxy approach discussed in PKIX ... extended it to public key ... and x9.59 ... there is nothing preventing any of these 3rd party "proxies" from creating a public/private key pair for each client .... and performing a x9.59 digital signature and including it in the 8583 message that eventually makes it way to the customer's (issuing) financial institution. The client would not see any difference to what they see today using password authenticated 8583 transaction (the 3rd party proxy would just be adding all this additional stuff to it). In fact, the client wouldn't even have to know that the 3rd party proxy has injected x9.59 digital signature information into every 8583 transaction. It is relatively simple and straight foward enhancements to the existing 3rd party proxy virtual POS terminal operations (and the extension would be totally transparent to the client).

Now, functionally I don't see any difference between that proxy implementation and some of the bifurcated proxy proposals (having two different proxies .... one responsible for client authentication that generates the actual 8583 financial transactions .... and a totally different proxy that signs a totally different transaction and tries to figure out how to route it to the client's financial institution ... as well as somehow connect it to the real financial transaction). Having it all in a single 3rd party process would appear to be quite a bit more secure and less failure prone as well as KISS. The difference is that while it might be more secure, less failure prone and more KISS ... somebody might be more likely to ask why do it at all (as opposed to doing the same thing in a much more complex, complicated and failure prone operation).

misc. postings in the PKIX proxy thread:
http://www.garlic.com/~lynn/aadsm11.htm#19 IBM alternative to PKI?
http://www.garlic.com/~lynn/aadsm11.htm#23 Proxy PKI. Was: IBM alternative to PKI?
http://www.garlic.com/~lynn/aadsm11.htm#24 Proxy PKI. Was: IBM alternative to PKI?
http://www.garlic.com/~lynn/aadsm11.htm#25 Proxy PKI. Was: IBM alternative to PKI?
http://www.garlic.com/~lynn/aadsm11.htm#26 Proxy PKI
http://www.garlic.com/~lynn/aadsm11.htm#27 Proxy PKI

anders.rundgren@xxxxxxxx 5/27/2002 1:45 pm wrote:
Lynn,

Although security-wise probably perfect, such a solution is incompatible with current COTS (Commercial Of-The-Shelf) software, while neither 3D Secure, nor its proposed replacement, demands anything new on the client side.

If security alone was the "perimeter for success" maybe X9.59 would be the only game in town, but simple deployment is probably the #1 parameter. That's the reason why smart-cards are said to be "coming strong next year". I just wonder what calender these guys are using :-)

Regarding the idea that financial security must consist of an unbroken signature all the way from the client to the end- destination (which is?), I remain unconvinced that this ever will be the dominant way to do things for a number of reasons, where the most important one is, that it typically requires key- generation and/or key-distribution for every kind of account etc. a customer may have the right to use which seems very impractical.

A two-step PKI (client-to-FI, FI-to-Merchant) is more flexible as the client-side PKI may be a single ID-card-like credential activating any number of external Server-PKI resources at any number of TTPs (FIs, Employers etc).

Server-PKI (proxy-PKI) has recently been discussed in the IETF-PKIX list, and I note with pleasure that the resistance is gradually going down. Three years ago, I was [very] alone in that list advocating for such "heretic" uses of PKI. Nowadays even esteemed cryptographers like Peter Guttman and Phillip Hallam-Baker (SAML architect), seem to be "hanging out with the bad guys".

cheers,
Anders Rundgren


Proposal: A replacement for 3D Secure

Refed: **, - **, - **, - **, - **
From: Lynn Wheeler
Date: 05/28/2002 08:08 AM
To: Anders Rundgren <anders.rundgren@xxxxxxxx>
cc: internet-payments a <internet-payments@xxxxxxxx>
Subject: Re: Proposal: A replacement for 3D Secure
as mentioned in previous note ... x9.59 was designed to be used to authenticated the actual financial transaction ... not some other financial transaction. with regard to 8583 (credit, debit, atm, etc transactions) a mapping was provided between what the client provided the authentication for and the actual transaction. Some work was also done mapping between x9.59 and other transaction message formats (like ACH ... checks).

with regard to proxies ... the previous note mentioned that there are already 8583 proxies operating effectively "virtual" POS terminals (i.e. web processing that simulates the message generation that occurs in point-of-sale terminals before passing it off to the merchant's acquiring processing). some of these existing proxies also provide a customer/client password service .... allowing the customer to provide userid/password (or possibly cookie/password) and all the relevant 8583 fields that need be supplied by the customer (PAN/account#, address, name, etc) is automatically filled in.

For those 3rd party proxies that already provide such a service ... the previous note observed that these existing 3rd party proxies could generate public/private keys on behalf of their clients/customers and include digital signatures in the actual transaction (even x9.59 comformant digital signatures) with no requirement for any additional effort or knowledge of their customers.

A possibly side issue is while other documents may not mention 8583 ... if you are doing a credit card transaction ... the actual transaction is 8583 regardless of what else you might do. Now if you want to authenticate the actual transaction ... then if it is a 8583 transaction ... you have to do something about providing authentication for the transaction (whether you bother to mention that it is 8583 or not). Furthermore, if something is provided that claims that it is somehow related to the financial transaction ... and it isn't done by authenticating the actual transaction (whether it is called 8583 or not) ... but is doing some totally different authentication ... then it isn't end-to-end security in the traditional security sense of the term (that was also mentioned somewhere in the PKIX thread).

previous post:
http://www.garlic.com/~lynn/aadsm11.htm#30 Proposal: A replacement for 3D Secure

anders.rundgren@xxxxxxxx on 5/28/2002 12:47 am wrote:
Lynn,

I think we are talking about rather different things, including using different "languages" :-)

But I did not see a trace of any web-based client-solutions on the X9-web-link that your X9-college Tom pointed to.

Because I'm mainly focusing on the client-solution for web-based transactions. Particularly as I feel this is the biggest problem. Therefore a "zero-client" solution seems like a necessity. Zero-client (using my definition), is to not require any hardware or software beyond what the clients already need for connecting to their Internet-banks. Such solutions are currently all over the map, spanning from userid/passwords to full-fledged PKI- solutions using smart-cards and digital signatures. I don't intend to even touch this part, as this would be like begging for a major disaster. Like SET 1.0...

Regarding the transaction from the CFI=>Merchant=>Acquirer, I'm open for suggestions as this is not my "territory". So if X9.59 is compatible with current schemes it should fit right in. Please feel free to suggest how X9.59 could replace the current 3D Secure authenticated payer message and what the pros and cons (there must be such as well), would be with respect to the existing payment structure(s) which you are much more acquainted with than I am.

I noted though that the 3D Secure specification does not even mention 8583-transactions which indicates that there are some problems in this end as well which may call for switching/proxying etc.

cheers, Anders


ALARMED ... Only Mostly Dead ... RIP PKI

From: Lynn Wheeler
Date: 05/28/2002 09:22 AM
To: epay@xxxxxxxx, internet-payments@xxxxxxxx
Subject: ALARMED ... Only Mostly Dead ... RIP PKI
http://www2.cio.com/research/security/edit/a05232002.html

ALARMED ... Only Mostly Dead ... RIP PKI

From: Lynn Wheel