On Sat, Sep 27, 2008 at 4:59 PM, Michael Snow <wikipedia(a)verizon.net> wrote: [snip] [snip] There are so many distinct classes of problem with flash that it can be hard to have a productive discussion about it. For example, you can assault flash for being ruinous for accessibility to the disabled, you can criticize the problems of security created if you allow uploading flash software, flash lacking freely licensed authoring tools, for it being browser crashing, being mostly used for toys and trash, so widely blocked in many business environments, etc. For the purpose of this discussion I'm going to talk about flash-as-a-video-format. For that purpose the important problems are almost entirely caused by the codecs used inside flash. Adobe does not own the patents those formats and couldn't freely license if they wanted to. Not only has this situation not gotten better recently, but it has gotten worse as Adobe has incorporated a newer highly non-free video format. I nominate this for the most misguided statement of the year award. :) Flash video requires the some of the most encumbered and expensive video formats available. The rights holders of H.264 have created a "your first hit is free" short time offer where web use is possible at no-cost until 2010. While I do not know what it will cost post-2010 H.264, today the licensing for non-web use of H.264 involves per-use/per-device licensing fees which are the most expensive of any widely used codec that I am aware of. Microsoft's OOXML office format is closer to a free format than the video used in flash. :) It's important to get details like this correct in a public message since mistaken statements like this harm the hard work done by many people working to improve the situation. This is a discussion that no one can rationally have without an objective look of what the benefits would actually be. It needs to be backed up with measurements. But no where in your message do I see you raise a serious question of what advantage would be gained in return. It seems to be just accepted that there would be some great benefit ("It would make it much easier to achieve wide distribution"), with only an airy an unquantifiable "evangelism" lost as though there were no pragmatic benefit. The existing system correctly plays back for an overwhelming majority of users (I posted some stats on this two years ago or so), though my ability to measure the success rate was lost when it was made a part of the site-proper. The 'much easier' assertion is often made, but not backed up with facts. While I think we can agree that there would be some benefit to using Flash Video, the evidence I've seen so far would indicate that the benefit would not be that great. The primary usability problems on our site have been pointed out to be things mostly or totally unrelated to format choice. For example, we have no software facility for converting arbitrary uploads from users to whatever we'd like to distribute. This is a format-agnostic problem and it results directly from Wikimedia's historic non-investment in this area of functionality. The same inability to transcode on the fly means that if a user uploads a 10mbit/sec video,your DSL connected PC will sputter as we attempted to cram 10mbit/s into a 1mbit/sec pipe. The typical bitrate used by youtube has historically been around 250kbit/sec for audio + video, which is why the quality is poor, but which is also a major reason why it works fairly well. When last I checked the average bitrate of videos on commons was about 4mbit/sec. So you're making a significant error if you compare the results obtained by a company whos bread and butter is working video with the results we've obtained while targeting a very different use case, and with video largely as an after-thought. So I think the benefits question is still far from closed. There is a real practical loss associated with offering formats that require payment for their use even in parallel to free formats, one which extends beyond the licensing costs that Wikimedia itself will likely have to pay. Consider: the only reason we (or almost anyone else) would ever consider using a slightly non-free format is because there is some 'costly' shortcoming in the free-format. For free audio and video formats the shortcoming is that they are not widely adopted. For media types where a free format is widely adopted: hypertext (HTML), uncompressed audio (wav), still photographs (jpg), absolutely no one suggests that we even consider using non-free formats, although proprietary formats do exist in these areas. Right now there does not really exist a free format for low bandwidth video: There is Flash/H.264 which requires payment to the rights holders. Alternatively there is Ogg/Theora which you pay for in reduced compatibility. The difference is that if we offer only the latter we will cause (and demonstrably have caused) people to install the latter and encourage vendors to support the latter which decreases that cost for everyone, eventually contributing to the format becoming truely without cost. I don't see value in evangelism for formats, and I don't expect you to. I see value in creating a world where success in our mission, which ought to include people being able to exchange the knowledge we have recorded without paying format taxes, is possible without compromises. Getting there requires something that looks a little like evangelism, but the motivation is entirely different. Offering "Widely-adopted-non-free-format *and* not-widely-adopted-free-format" is a minor compromise for someone who's motivation is promoting "not-widely-adopted free-format". But if the motivation is "make the free format widely adopted, so no one is forced by the market to use the non-free format" then it doesn't make much sense as a compromise because it completely loses the ability to use Wikimedia as an argument for including free formats in browsers, and completely loses the encouragement for regular users to click-install and get support themselves. But until the free formats have enough adoption that people feel no reason to consider this using non-free for reasons of adoption, there really is no such thing as a free format, only formats with different kinds of costs, and Wikimedia needs to help change that if it wants it's work to live up to the licenses the work is distributed under. Once adoption is no longer a factor in the use of non-free formats, I'll gladly write the flash-video support for Wikimedia myself if Wikimedia still wants it. ... but it won't want it, because it will no longer be a consideration just like non-free text formats are not a consideration. The fact that it is a consideration is all the evidence you should need to see that format-freeness is still a problem which is obstructing Wikimedia's mission. If you want to fix this problem you could start by investing in the solution: There are organizations like Xiph and events like FOMS (http://www.foms-workshop.org/foms2009/) which would benefit greatly from Wikimedia's support. Every Wikimania seems to have attendance by people in the proprietary video world, but folks working on free tools seem to get little audience because they can't afford the travel, and can't sponsor PR blitzes.

2008/9/27 Michael Snow <wikipedia(a)verizon.net>et>: For the moment: as others have noted in this thread, there's no need for us to even consider compromise on audio and video formats, because we're winning. Theora and Vorbis will be supported in-browser by Firefox 3.1. (What's the stats for browser user agents on Wikimedia?) For the future: we may have a reason to compromise temporarily at some times. As I understand it - and I strongly welcome correction if I'm wrong! - we put some Java software on the servers and loaded the Cortado Java viewer in browsers before Java was actually 100% free software, because Sun's programme for freeing Java was progressing nicely and there was little likelihood we'd be embarrassed by them failing to complete it. So it may not be a good idea to straightjacket ourselves. - d.

Michael Snow wrote: My understanding, from a previous mailing list discussion, was that we can and hopefully will support Flash video. As long as the encoder license fees are reasonable in proportion to our other operational costs, adding proprietary video formats alongside free ones can only increase the dissemination of knowledge and assist in our educational mission. -- Tim Starling

I will quickly weigh in on this thread: First context: (I developed metavid.org (a wiki congress video archive) and presently working with kaltura & wikimeida on the early stages of a ogg theora html5 collaborative video editing solution) We at the (much much smaller) metavid project faced a similar issue of accessibility vs free/freedom a few months ago. As an explicitly free & patented unencumbered software project for the first 2 years we had exclusively used ogg Theora for our video archive. There was subsequently a strong push to improve accessibility and support flash. The solution was to do a fall back distribution of the flash codec decoupled from the player. Presently on the metavid site you can play back the flash video streams with either VLC or the flash plugin. If your client supports it you can play the ogg stream with vlc, native browser support, or cortado. Key to this fall back solution is the user interface is identical regardless of whatever method your playing back the content with. All the play head controls, scripted interactions with transcript editing, remote embed scripts, transcript playback are abstracted away from the player. This is in some ways comparable to canvas emulation in flash that brion mentioned. I think if the fullback approach is properly implemented it facilitates larger accessibility and hence entices much more wide scale usage of the video functionality. Ultimately allowing you to more actively promote free software solutions as an experience with identical or improved quality without the costs of proprietary codecs. As Erik mentioned in this thread more data about playback would be helpfull and by request I have hacked up a simple video player data collector and survey on the ogg / flash playback support situation. Should be ready to deploy shortly. --michael Michael Snow wrote: