[Foundation-l] more thoughts on openness and collaborative media

[Shay David]

hi everybody,

i'm writing to respond to the heated threads of the last few days concerning the 
collaborative video technology. other than a few isolated cases of vandalism on 
  our wiki, and a few comments in the blogs that spread half-truth that were 
simply incorrect, and putting religion and the Nazis aside, I've found this 
discussion quite useful at articulating the core of the issues that will need to 
be dealt with in order to ever bring this technology to wikipedia.

I apologize that this post will be a bit long, but these issues are complex, and 
I want to address several of the concerns raised, so bear with me.

First let's start with the understanding the problem space. Video collaboration 
is hard. Text collaboration is hard enough, audio harder, but video is really 
the 'black sheep' (for reasons I discuss below). Despite some promising early 
attempts, to date there simply is no open system for online video editing let 
alone one that supports collaboration in the wiki-spirit. Now you can look at 
this situation and say 'so what? who needs it anyway' but I think that anyone 
that has been paying attention to recent trends on the web knows that in  a 
YouTube era, rich-media is an essential part of the web and of culture, and that 
its share is only growing. if wikipedia aims to share with every human being the 
sum of ALL human knowledge, then we can't really dismiss the need to include 
video in that spectrum.

Kaltura is a small startup with big ambitions. It's mission is to build the 
first global and open network of legally remixable video and rich-media content, 
to be licensed under CC licenses. We are partnering with organizations the world 
over in order to build this network. These include for-profits and non-profits, 
universities, libraries, museums, NGOs and governments. While our partnerships 
extend well beyond encyclopedias, it was natural for us to seek partnership with 
the WMF and the Wikipedia community, from whom we take great inspiration and 
seek guidance on the standards of openness. As far as we are concerned, the 
announcement of the open experiment is an invitation to start a conversation. To 
this extent all the threads on WMF mailing lists and the blogs are a very good 
start to this conversation.

To enable open video collaboration several challenges need to be overcome including:
a. Technological
the system needs to be able to ingest content in a wide variety of formats, to 
convert them all to one commensurable format, to be able to trim, sequence, 
annotate, and overlay content, and lastly to be able to store and stream large 
volume of huge content files efficiently, all while preserving metadata 
information for permissions, revisions, and attribution.

b. Legal and openness-related
to be truly open, or free in the sense of the free software movement, the system 
needs to do all this while overcoming three distinct problems
1. the tools used need to be open sourced
2. the formats and codecs need to be free of proprietary claims
3. the content needs to be licensed under free licenses that can accommodate 
remixing (which greatly complicates the notions of attribution, derivative 
works, and transformation)

As mentioned above, there is no system that can do all this today. I hope that 
we are on track to deliver such a system before the end of 2008. But it's a long 
route, and we are not there yet. Some of the comments on this thread remind me 
of Zeno's paradox that "proved" that you cannot get from point A to point B 
because there will always be a point C in the middle that you need to get to. 
Today we might be closer to our starting point than to the end point, but we are 
not stuck in any point C. Building this system is doable and with the right 
support, it can be accelerated. When Richard Stallman set out to build GNU, he 
started with EMACS and GCC, not with the kernel. It took more than six years for 
Torvalds to develop the Linux Kernel. What was common to both of them was that 
they made their system compatible with the closed UNIX, giving an organizing 
principle to a disparate development environment. In much the same way, we hope 
that we can use Kaltura's existing system, that uses several proprietary 
components, in order to build a truly open system.


I want to refer to a few specific points below.

1. The use of ogg/vorbis/theora. As you can read on Wikipedia and elsewhere, 
these formats are not magical. They are simply the older generation of ON2's 
technology, freed from their patent claims. Right now they are the best option 
for an open format, but we cannot and should not design the system to rely only 
on them. The system needs to be open enough to work with any available format 
that is free. One would hope that one day Flash will be free. Adobe is seeking 
its way in this regard. Perhaps with sufficient pressure from projects such as 
this, Flash will be free one day. Perhaps this is wishful thinking, but if we 
don't try, we'll never know.

2. The status of the open source release. It's perplexing to me that people 
didn't even bother to look at the code before calling it a  'Trivial MediaWiki 
plugin'. We have released what I believe is the most extensive online editor 
framework ever to be developed, and the smartest online media player out there. 
It took over a year of development and thousands of hours of some very bright 
engineers.

These components are written in ActionScript/Flex, and in order to compile them 
and run them, one can use an open source SDK developed by Adobe,
http://labs.adobe.com/wiki/index.php/Flex:Open_Source
or other IDE's such as Eclipse.

The main challenge of this project is to take this code, and make it compatible 
with GNASH. We have been in touch direct communication with Rob Savoye and the 
GNASH team, and are in the process of doing so. Any developers that can help us 
in this goal are invited to contact me directly, or sign up on sourceforge.

note that the editor and player are not simply tools to play or edit video. they 
are a whole framework that represents a new approach to video on the web. the 
editor embodies a new meta-data schema that allows one to describe a set of 
rich-media assets and operations on them such as transitions and overlays. it 
also support layers of dynamic data such as RSS feeds, or any other XML data 
streams.

the framework is built in a way that abstracts the core from the plug-ins, so 
that plug-in development is greatly simplified, making the framework very 
extensible. we are seeking help from developers that are interested in 
developing plug-ins for transitions, overlays, effects and dynamic data layers 
(such as, for example. internationalized/localized sub-titles)

what has not yet been released as open source are several important components 
on the back-end that handle content ingestion and streaming. these where 
originally developed on a system other than media-wiki and are now in the 
process of convergence to the mediawiki architecture, and will be released as 
open source during the term of the experiment.

in the meantime, for anybody that wants to test it, we provide a hosted back-end 
solution that grants FREE access to developers and non commercial use. it 
requires a 30 second registration process after which each developers gets a 
key, as is customary on every web-service.


3. the relationship between kaltura and metavid. I think that at the basis of 
this thread lies a misunderstanding of what each project is. while similar in 
some respects metavid never intended to be the framework that kaltura is 
developing. Michael Dale and I have been in touch for a while, and we are now 
trying to figure out what are best integration points, and how we can join 
forces. it seems as if a common XML structure for sequences is the first step, 
and we'll take it from there.

4. on the point of free formats and fair use, and some claims that the 
collaborative video technology will encourage copyright infringement. It's 
important to understand that fair use does NOT exist in the abstract. For 
anybody interested in it, http://en.wikipedia.org/wiki/Fair_use  actually gives 
quite a good overview of how the courts have been swinging on the issue. Fair 
use can only be interpreted within the context of specific uses (for example, in 
past cases, making thumbnail image copies for the purpose of indexing, was 
upheld as fair use, whereas reusing short music clips was not). There are 
various factors to determine when use is fair. I strongly believe that for its 
clear transformative value, the collaborative video technology represents an 
opportunity to push the envelope on what uses are considered fair.

that being said, for cases that are clearly infringing, the tools should support 
all the standard mechanisms that will help avoid inclusion of copyrighted 
material in the free content repository, and will allow the community to 
self-police, as it has done in the past.

5. regarding the Terms-of-Service. here too I think that there was a lot of 
confusion of what was covered under the TOS. thanks to the commentators that 
have noted that there were some inconsistencies between the TOS and the CC 
license that governs user submitted media. We took note of this, and it will be 
corrected immediately. There has been a learning curve for us in deploying all 
this, and any further comments are welcome.

at the end of the day, I think that this discussion proves why it was so 
important to start this project with an experiment in a way that will allow us 
all to crack the hard nuts BEFORE any problematic technology is deployed on WMF 
sites. I look forward to seeing more comments, and to getting ideas for creative 
solutions to some of the thorny issues that were raised.
I encourage you to experiment with all this on the kaltura demo wiki, or on 
wikieducator, and make your feedback public on our feedback page
http://www.kaltura.com/devwiki/index.php/Feedback_Page

happy remixing,

Shay David
CTO, Kaltura