Hi Martin, I think we are basically on the same page :-) Here is a recent document that may be of some interest: http://webpki.org/papers/mobilephone-pki-options.pdf
I'm obviously not a fan of WPKI in spite of that it "Works". But I have also given up on PC-browsers since signatures are already used by MILLIONS of people but the browser vendors still don't get it. Associated effort that also will be dwarfed by the limited browser vendor support: http://www.idmanagement.gov/drilldown.cfm?action=openID_openGOV Anders member of http://TrustedComputingGroup.org ----- Original Message ----- From: "Martin Paljak" <martin.pal...@gmail.com> To: "mozilla's crypto code discussion list" <dev-tech-crypto@lists.mozilla.org> Sent: Thursday, October 08, 2009 14:20 Subject: Re: mobile phone certificates. Re: why client certs FYI: Estonia has WPKI, eID keys on SIM cards, a SIM application triggered via OTA messages. This is often suggested as either an overall replacement or additional method for cases when browser based PKI (SSL+server conf+signature plugins/applets) miserably fails, for whatever reason. Unfortunately it often fails for either technical or usability reasons, because for historical reasons, in Estonia the picture is messy with several bits and pieces on different platforms and browsers, which changes with either OS or browser updates and require re-installation of the client software to make it work again. It is also messy because of the "freedom" of every web site owner to choose the components he likes - everyone can roll their own applets for example, that all behave differently on different platforms for example, or play with SSLVerifyClient mod_ssl option. WPKI on the other hand Just Works, the same way, with all browsers, on all platforms, without any additional software. And there is just a single culprit to blame when things fail - the GSM operator, unlike smart cards where one sometimes has to explain why the fscked up OS X 10.5 PC/SC stack causes a useless "ssl_handshake_failure_alert" page in Firefox and that closing the browser (to clear the SSL cache) and re-plugging your reader and smart card helps. There are also downsides: mID (mobile-ID) is not as universal and available to application developers as it is a server2server application which requires a subscription deal and costs money with every transaction. eID smart cards on the other hand are free to use, offer better end-to-end security and cost no extra money per transaction. The problem is that SSL, which has some very nice properties from security point of view, is not suitable for an end user friendly authentication mechanism on the web. It is a nice technical mechanism that is buried deep in the protocol stack for already many years, designed by security technologists and programmers, and now "exposed" to users because *technically* smart cards fit in that framework. Loading PKCS#11 modules to make things work? Whaaat? It can be mitigated by extensions and installers, but hey... A user on some forum in Estonia suggested a really nice "it is possible with software" solution: "why don't you already build a browser-like application that would work on all platforms and would contain all the necessary software to make it work ???" Which would be indeed a much easier to maintain solution with predictable usability, instead of providing "bits and pieces", plugins, framweworks and what not to please the USB/CCID/expresscard/pcmcia/PCSC/PKCS#11/CryptoAPI/tokend/applet/plugin/firefox/IE/Opera/Chrome/whatnot monster. Ask anyone who deals with interaction design about what they think of smart cards and browser usability and then count the used swear words... -- Martin Paljak http://martin.paljak.pri.ee GSM:+3725156495 2009/10/8 Ian G <i...@iang.org>: > On 07/10/2009 22:17, Anders Rundgren wrote: >> >> I don't believe that client certificates in PCs will ever become >> mainstream since >> credential mobility and distribution issues have proved to be >> insurmountable; not >> technically but politically. >> >> However, in mobile phones at least the mobility issue is solved >> (phone=token) which >> is also the reason why useful distribution schemes will start there as >> well. >> >> Since phones also have limited screen resources, PKI GUI improvement is >> also called for. Certificates will most likely be represented as cards >> through an >> image that either is embedded in the certificate (PKIX) or be supplied as >> a >> separate attribute during provisioning (KeyGen2). >> >> Will this one day reach the PC? No, you will still use the phone as the >> token >> (and token selector/executor) while the PC crypto will be bypassed. NFC >> does the connection together with Wi-Fi. > > > Hmmm! Interesting thoughts. > > iang > -- > dev-tech-crypto mailing list > dev-tech-crypto@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-tech-crypto > -- dev-tech-crypto mailing list dev-tech-crypto@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-tech-crypto -- dev-tech-crypto mailing list dev-tech-crypto@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-tech-crypto
