Where is Web Crypto ?

For the ones wondering if the Web Crypto is a dead topic in W3C, they should not worry anymore. The Web Crypto API is safe. It is still trying to define the primitives of cryptographic operations, to be integrated in all browsers, and it is still alive. What we mean by alive in W3C is that : it entered Last Call, and received a serious amount of great feedbacks, from smart reviewers and this is the good sign of interest. The dark side of having productive reviews is that it takes time to address all of them. That is why you have not heard about W3C Web Crypto API since the summer : editors Ryan Sleevi from Google and Mark Watson from Netflix are busy, with the Working Group to solve open bugs (see our bugzilla repository).

Yes, some bugs are more complex then others…

At the moment the Working Group is reviewing interesting items :

– Making sure the specification is extensible to integrate new algorithms or new type of algorithms (see related bug)

– Trying to find a way to include specific algorithms such as NUMS curves and 25519 curve (see related bug). That discussion is obviously related to the idea of specification extensibility. But those candidates are special in a way that some related discussions are happening in IETF. IETF is currently discussing the set of algorithms that will be recomended for next TLS version. We know that several candidates are running, and IETF and W3C should on that matter stay synchronized, nevertheless the sensitive aspect is that IETF is not scheduled to make decision before its next meeting, in early November.

Once those bugs will be resolved, with few other ones, we might be in a position to exit Last Call, lets say in September. This means that we could start moving toward Proposed Recommendation. Another great adventure…

Proposed Recommendation ? What does it mean ?

It means that W3C is calling for implementations, experiments, based on the specification as edited in its Last Call exit. That specification may embedd some feature at risk. Meaning some features that my be dropped after implementations demonstrated it is not reasonnable, technically viable or feasible. But this is where we will see implementations happening. The implementers will have to demonstrate their interoperability, that is why we will develop tests (note that a repository is already existing, we just need to fill it with smart and relevant tests). Several implementations have been referenced already in Chromium (see related issue), in Firefox (see related bugzilla) and pending in WebKit (see related bugzilla) and Internet Explorer (see IE11 preview announcement).

Those beginning of implementations on the browser side make that specification on good track to become an actual W3C recommendation, with associated prooducts. This should happen, lets say… in 2015 (standards are living in a timezone where milestones are barely predictive, but are supposed to follow a process, see W3C one)

Do we close the Working Group, after that ?

During the last 2 years, the Web Crypto WG has been addressing a variety of problem and features, some of them have been parked. That is why even if being in Last Call for the Web Crypto API, the working is already thinking about its next version. The discussion about prioritization of next web crypto API will happen during a 2 days workshop, held in Mountain View on the 10 and 11th of September. Two days for that topic may sound long, but it appears that the topic of cryptographic is also related to other key aspects such as usage of secure token, like hardware or trusted software are, or web authentication operation – where FIDO Alliance is currently working on, or again certificate management and in general linked to browser security. Gathering all actors having some interests in that topic will hopefully help W3C to identify realistic features to be added in the open web platform. The workshop will be quite crowded as 70 people coming from 47 companies, all discussions will be accurately minuted and a report will be issued. You may hear about it on that same blog…

Note : you can also read more about the Web Crypto API scope by reading an old post here : https://poulpita.com/2013/07/31/first-web-crypto-implementations-expecting-your-imagination-to-play-with/

Let me get you jealous of me… Summer time. South of France. Ideal conditions to attend the worldwide conference of the free software (Rencontres Mondiales du Logiciel Libre, in french). Lets call it RMLL2014. What is it about ? All about free software and open source, covering broad topics such as society, internet, art, collaborative tools, software suites, security, open data…

All free software, offering alternatives to proprietary software, together with its designer, coders, evangelist and their free mindset were present. One may ask ‘what the hell was she doing there ?’. That happened because Christophe Brocas @cbrocas, Mathieu Blanc @moutane, and Philippe Teuwen @doegox organized a security track. And it happens that free software, open source, W3C security do have something in common. My presence here, was also the opportunity to attend some amazing talks. Here are some benefit of sitting here and listen.

Getting free software essence, from Richard Stallman, himself

Mister Free Software Foundation, the daddy of GNUproject, the man who showed the path to open source, collaborative wealth and promoting citizen control over software. After selling few goodies for the FSF and taking pictures with fan, Richard Matthew Stallman exposed his theory. Simple and efficient. Any software which is not open is killing freedom and is a potential danger. As a consequence, a relevant software has to embed the following merits : the freedom to run the program as you wish, for any purpose (freedom 0), to study how the program works, and adapt it to your needs (freedom 1), to redistribute copies so you can help your neighbor (freedom 2), to improve the program, and release your improvements to the public, so that the whole community benefits (freedom 3). This implies access to the software source code. The ideology behind that business model proposal is to give back power to the people, reducing technology attacks on citizen freedom. Having by default the possibility to analyse collectively a software gives, from Stallman perspectives,  more chance to refrain editors to performed actions without user consent (understand, implementing tempting things such as backdoors, user tracking systems…).

Discovering Caliopen, the new private messaging project, by Laurent Chemla

Another free thinker. Laurent Chemla, french, on the internet since ages, promoter of individual freedom and collective activism, founder of Gandi (a hosting server service). Laurent wrote ‘Confession d’un voleur‘ (confession of a robbery) in 2002, where he confesses how he made money because he was curious and visionary, because playing days at nigh on internet while other did not realize it was existing. This gave him a competitive advantage he transformed into money, but he did not feel comfortable with it, as his vision was more that internet was about changing the world, and giving a chance for the citizen to take power back on politics and locked systems. Since few months, Laurent is working and thinking about a new service related to messaging, that would fit his dream, aka being privacy centric and secured. Rationale for designing this offer is the following : freedom is at risk those days (and since a long time, by the way, but lets pretend Snowden was a new thing, says Laurent). Except few geeks ready to get technical skills to protect themselves and suffer the pain of using PGP, noone is really taking serious action. He re-assessed that freedom must be protected, by political and social actions, but also with technical tools (another domain where he feels he can do something). Laurent does not want to build another complex solution, or a like-gmail-but-more-secured solution, but rather offer to users an attractive email service. Here is Caliopen. Caliopen promise is to manage conversation, whatever is their media (email, sms, pigeon, …) and associate trust level to every piece of the communication. A user, a conversation, a device, a context for using Caliopen, any central piece has a trusting level. Which allows power to the user to decide. Shall I send this sensitive email to my friend Stéphanie who is well known for opening her mails on her crappy device in the train, on public wifi ? Part of Caliopen exists, the back end side, and the project is looking for skilly web developers to code the client side.

Hearing recent news about Tor Project, by Lunar

Lunar is one of the voice of the french association Nos Oignons, which aims to raise fund and maintain some exit nodes of Tor project. Lunar shared during the security Track of the RMLL2014 principles, weaknesses, and stakes behind Tor Project. For the ones not yet familiar with Tor, it is a network allowing users to access to internet resource, with an anonymized address. The aim is to counter the filters over internet imposed by application, services, ISP or government. Tor Project relies on a network of 4000 nodes, which are driving the communication on a random path, inducing the scrambling of the IP address of the machine which initiated the request. lunar is clear. Anonymity is not perfect. For example by spying data entering the Tor network, and data getting out of the network and making correlation. For example, some information about traffic can be guessed by observing the Tor Network life, or by entering the guard allocation entry and exit nodes… But the network is used and maintained by people having a vision, always trying to improve its reliability, usability, supported by a researcher community. What was also discussed were the conditions for hosting an exit node. As it is a special node, exposed to a large traffic, and to potential legal attacks, it requires to have a fix IP address, with reception of notification and better if it is an association behind, rather then a specific person. Integration of Tor access into software is also a key discussion point and the conclusion was that only Tor Browser or Tails should be used. One should note that Mozilla and Tor Project are working together on the definition of the next privacy feature of Firefox, in order to make ‘private navigation’ a feature with a real value proposition. Lets stay tuned… A lots of usefull references for understanding Tor projects and limits can be found under https://pad.riseup.net/p/lsm2014-tor.

Next episod of that post will tell you more about other projects discussed in the RMLL2014 Security track…