On Wed, Aug 21, 2013 at 4:38 AM, Brion Vibber <bvibber(a)wikimedia.org> wrote: The last point isn't accurate. The original plan was to exempt certain languages from the login redirection, and those projects would be "home" wikis. When someone logged-in there, they'd also be logged-in everywhere else via central auth. The current plan is to disable the HTTPS redirect using geolocation for countries that have a > 5% error rate for HTTPS requests. This discussion is technical, so I'm going to move back to wikitech-l, now. - Ryan

On Aug 21, 2013, at 1:39 AM, Pierre-Selim <pierre-selim(a)huard.info> wrote: The reason why is outlined in Ryan's blog post as well as his previous post and the Wikipedia entry on https linked from that post. The short answer is the current state is known to present a number of privacy and security vulnerabilities further emphasized by the now-known existence of software designed to deliberaty target these vulnerabilities in Wikipedia specifically. https://blog.wikimedia.org/2013/08/01/future-https-wikimedia-projects/ The article mentions that anons will still be able to edit. It also mentions that areas that block https will be considered for bypass on redirect even though this reduces the overall privacy protection of all the projects in the manner discussed on the parent thread on wikitech (linked by Ryan below). Putting real numbers behind those "mights" has been part of the hard work put into it by the operations and platform engineers at the WMF and they are adjusting to the real world tradeoffs you mention. (Btw, the issue is not that https is filtered but that it is BLOCKED by certain countries BECAUSE it cannot be filtered.) They are inviting your discussion here and on wikitech on those details. As for the decision itself, even though that has been on the roadmap for quite a while, I'm sure that even that is amenable to discussion. It would behoove anyone who wants to influence the decision to be well versed in the historical discussion first. Luckily, the standard for the Movement is consensus, not catering to every extremist view with 100% buy-in. The latter standard is impossible as people would be affected either way. The technical component is informing the decision and helps to hash out some of the details, but this is a case where parts of the Vision are being compromised today, and a different (hopefully better) compromise is being reached through this rollout. Take care, terry

Hi, <tl;dr> I do not really enjoy the way the mandatory-for-editors HTTPS was introduced, mainly for time frame and communications (still) reasons, although I’m globally really enthousiastic about a better security and particularly the activation of HTTPS. Generally speaking I do _hope_ in the future WMF will give more time and more discussion space to handle major changes. <end tl;dr> History: (I concede I may lack some readings, but I think I have the big picture) After the PRISM scandal in June (2.5 months ago) everybody condemned that program and the Internet security became a major concern for Internet users. HTTPS is in important means to improve the security (although concerns about the protocol and the way it is implemented appear) and since it was a matter of time before it could be globally activated the blog post published on August 1st announced HTTPS will be activated for logged-in users 20 days after, with solutions about the blocked China HTTPS to be found [1], after a discussion on wikitech-l [2]. Some Chinese editors made petitions [3] (starting on 08/08) and Iranian users raised a similar problem [4] (on 14/08). In parallel these last two weeks there were discussions on wikitech-l about some way to opt-out by user and/or geographically. And in parallel the last two weeks there were discussions on wikitech-l whether some opt-out mechanism should be implemented with two opposed points of view: 1/ this security about the protection of the password must be for everyone else it is unuseful (which is true in a perfect world), no matter if China and other HTTPS-unlucky people cannot login (and hence must edit under IP or not edit); 2/ although security is very important, not to allow HTTP logins in China (and other HTTPS-unlucky people) will destroy etablished parts of the community and should be avoided, so implementation of work-arounds is needed. And this last discussion had not to be on wikitech-l because it is political, and was only a few raised elsewhere (where HTTPS is technical and should be discussed on wikitech-l.) Finally some work-arounds were implemented; first it was a list of wikis where HTTP login will be allowed (this decision became public on Monday [5]) and yesterday (sic) it was announced a geolocalised solution [6]. Secondly there will be a preference for the users, although until yesterday it was not clear for everybody how exactly it was implemented. In parallel the central notice was set up two days ago with an English-only page, pywikipediabot was announced to be ready some hours ago. And in some hours there should be the deployment target. [1] http://blog.wikimedia.org/2013/08/01/future-https-wikimedia-projects/ [2] http://lists.wikimedia.org/pipermail/wikitech-l/2013-July/070981.html [3] http://meta.wikimedia.org/wiki/Requests_for_comment/Petition_of_HTTPS_defau… [4] https://bugzilla.wikimedia.org/show_bug.cgi?id=52846 [5] https://meta.wikimedia.org/wiki/HTTPS?diff=5731209&oldid=5728786 [6] http://lists.wikimedia.org/pipermail/wikitech-l/2013-August/071348.html Conclusion: I know the fact we now know we are spied is disturbing, but… Why the hell HTTPS is so truly *urgent* we cannot spent more than three weeks (at all) to think about the problem, investigate related problems (including political and communitical here), think about solutions and user interfaces/interactions, implement solutions, widely avertize the problem and solutions, and peacefully deploy the patches? I would have loved some RFC and some discussion elsewhere than on wikitech-l with structured problems and solutions, and more time allowed for discussing all that with the community -- because I guess it was widely discussed internally in technical and operations teams, but the community discovered these plans and had to report potential problems in a time frame of 3 weeks. More generally speaking, I would love the WMF share more their internal plans long before rollout -- even if I concede writing and discussion is more time-consuming than oral speak and introduce latencies -- and probably in some digest and expanded forms (I know there are already both, it’s probably to be improved and perhaps more targeted to avoid everyone’s burnout). And perhaps slow the rhythm of the technical changes to have a more stable environment (I understand this is personal and there are other PoV). Thanks, ~ Seb35 Le Wed, 21 Aug 2013 11:37:35 +0200, Pierre-Selim &lt;pierre-selim(a)huard.info&gt; a écrit: