Commentary
It's Time For A Content Identification Standard
- by Ben Weinberger , March 29, 2010
Look around: it's no longer the Wild West out there. Yesterday's TV Everywhere experiments and pilot programs are morphing into today's company-wide business priorities for cable providers.
Digital delivery of premium video content is a real business, one that is generating real money for media companies. Those same companies need to get together with cable operators and other stakeholders to develop a uniform content identification standard.
The existing "system" -- not much of a system at all, really -- largely consists of manually identifying and logging individual television episodes or films. That workflow is simply unsustainable. With the volume of digital delivery set to rise exponentially over the next few months and years, it is imperative that media companies act soon.
By doing so they will save massive amounts of dollars, hours, and headaches.
Let's take "How I Met Your Mother" as an example. 20th Television produces the show, which airs on CBS and has recently been syndicated to Lifetime. This is a hit sitcom, and an extremely funny one at that. It will have a long life in syndication, overseas, and in the digital media realm.
When 20th and CBS deliver a new episode of "How I Met Your Mother," that half hour is bundled with basic, asset-level metadata: episode name, description of the storyline, guest stars, running time, etc.
That information lives in a data feed and is delivered to everybody receiving the episode - a list that includes programmers, cable operators and data providers like Tribune and Rovi, which need to accurately and regularly provide information to viewers for electronic programming guides. Sometimes, the providers create that data as well.
But here's the problem: with a broad and expanding universe of content delivery destinations, not having a consistent way to identify content causes unreliable metadata and an inability to traffic dynamic ads in an IP-based environment. Moreover, you can forget attempting to engage your audience through coherent search results and user experiences.
Let's do away with manual content identification. It is a downed tree lying across the express lane to progress.
The industry needs to automate content identification to ensure the same universal and accurate information travels with the content no matter where it is delivered (online, TV, On Demand, etc). This will streamline the process, increase accuracy, and approve the end user experience.
So, what would a future with a unified framework look like?
To create this digital "ID" for each piece of content, we would want to avoid watermarking, since these methods are considered destructive and alter the content by placing pixels into the digital asset. A separate, non-destructive, digital content ID could be an excellent choice.
That signature, housed externally to the digital content, would include a link to all of the contents' metadata (i.e. title, summary, and appearances). It would also have the content owner's digital signature verifying that this is indeed accurate episode information. Bottom line, the ID would enable instant and consistent access to the valuable metadata that facilitates ad placements, search, discovery and targeting across all IP platforms.
Universality of this kind would simply make life easier for any company or technology that is attempting to distribute video and generate revenue for media companies.
The more you know about your content, the easier it is to monetize and distribute. This is my "Yes we can!" challenge to the digital media ecosystem: let's get together, hammer out the details and make this happen. It will be a vast improvement for all concerned.
Ben, you are in serious danger of losing your "video insider status." There is ALREADY an international standard that is usable to uniquely identify content, and it is in widespread use.
It's called International Standard Audio Visual Number (ISAN), and it even includes "versions" of content: such as different primary languages, edits, etc. http://www.isan.org. There are even regional and industry-specific registrars. And, this has been a quite settled area for at least half a decade (the last change was to eliminate the distinction between ISAN and "version-specific ISAN" -- then called V-ISAN.) ISAN has been around in one form or another, for quite some time.
There are existing mechanisms for including ISAN in DVDs, transmitting in broadcast/satellite/cable transport streams, linking to web sites (much of the metadata you mention is outside of the scope of that provided by the ISAN.org web site.) There are even private numering ranges. And, ISAN only applies to finished works; there are other systems for identifying content in process.
The "complaint" is that ISAN registration costs $10 per item. Boo hoo.
Best Regards;
John Willkie
EtherGuide Systems
Ben,
To quote your conclusion: "The more you know about your content, the easier it is to monetize and distribute."
Assuming content distributors adopt your suggestions, how much do you think How I Met Your Mother will reap in online syndication?
My guess is zero. Same as Doogie Howser.
This already exists, it's called MPEG-7 (aka "Multimedia Content Description Interface"):
http://mpeg.chiariglione.org/standards/mpeg-7/mpeg-7.htm
Ben -
Let me guess, this is a service your company would offer your clients, correct? WOW - that would make the WHOLE ARTICLE Shameless Plug of The Day. I am impressed!
John - thanks for bringing up ISAN. I am the CEO of Secure Path Technology, the primary ISAN Registration Agency in North America. You're right in that ISAN is the solution to the problem which Ben references. We work throughout the ecosystem to evangelize and implement ISAN, an ISO standard which was designed from the ground up to be the universal content identifier. We agree there is a need, and are working every day to bring the solution to market. We have done a number of deals (including with companies referenced in Ben's article) which bring the cost per code down to a fraction of published pricing. It's also worth mentioning that the cost of procuring an ISAN code is a one-time cost, as the code remains unchanged for the life of the work. Here's the shameless plug: if you would like any further info on ISAN, feel free to contact us at info_at_secpath.com
Ben,
This idea was thought of years ago for internet content (darpanet back then). It was created in the 1960's by Dr. Timothy Nelson as Project Xanadu. Dr. Nelson's idea was to have hyper link back to the copyright holders be they video, text or audio. This idea was also reference that Tim Berners-Lee who helped develop the WWW used while at CERN.
In the current Broadcast world there is ISAN, but in the VOD world there is a CableLabs specification that everyone uses that supports a 'package' consisiting of the content and an XML file that contains all the metadata information of title, genre, rating, artists, description and dates content can exist and not exist. When you use cable , satellite and/ or telco iptv vod this is the format everyone uses for on-demand titles and it is created directly by the studios. An XML file would be easy enough to publish other places, but why duplicate efforts that already exists?
A requirement for content owners and providers is that it would need to fit into the current framework for the studios and operators to accept it. I would bet they would use the CableLabs model for this type of on-demand content since it easily fits into their current model they already use and their various ecosystems (content management, billing systems, encryption, clients, guides, etc.) support. If it is not an advantage to the current system, they probably would not adopt it given all the moving parts that would also need to adopt it.
@Brian, It only took over a DECADE for the studios to accept MPEG-4. Good luck with acceptance of mp7, 21 and others.
Jim
I agree, the studios were quite slow to adopt MPEG-4, but much of the rest of the ecosystem picked it up rather quickly when desktop/mobile computing power got to the point where decoding was realistic.
ISAN and the CableLabs metadata solution do not address temporal metadata (i.e. the notion of descriptors of objects that appear within time based media such as audio and video) - which is something that Ben and Digitalsmiths like to talk about. Also, XML is great for data interchange but inefficient at scale. MPEG-7 addresses this issue with both XML and binary formats.
I would argue that the sheer volume of content that is available on IP networks will require much finer-grained search and filtering capabilities than a basic library-style content description schema will allow for. And the web will naturally migrate towards open standards, which is why I believe MPEG-7 will eventually gain traction.
Actually, Ben, ISAN now (and has for about 5 years) provided for version-specific implementations, and they are supposed to accompany the content. You could have one version with spanish language subtitles for the airline market (cut appropriately.)
And, the ISAN web site provides for searches. More functionality can provided by the distributor or producer. Methinks you might just be 'pimping' to be a service provider. However, to do so, I think you would be better off knowing what is working today.
MPEG-7/21, by the way, has -- to my knowledge -- zero market penetration. It's more granular than is ISAN, but much harder to implement, and there doesn't seem to be much of a need for it.
It seems that I have hit a hot button with my digital media colleagues (some of you have even contacted me directly). It’s an important issue and I’m glad to see that so many people are engaged and interested. <br><br>
I want to take this opportunity to clarify a few points. As many of you have commented, there are systems like ISAN and MPEG 7 in existence today. I am not suggesting that we abandon these systems, but rather, extend their capabilities as part of a complete, fully automated visual identification system. <br><br>
Consider this. A single video often will have multiple versions (ex. Airline v. full broadcast v. director’s cut). Today, there is no central way to identify it as it gets re-transcoded and undergoes multiple iterations in the standard digital media workflows. As more content is distributed on more platforms, this problem will continue to grow in scope and complexity. <br><br>
The current identification systems are based on pre-set numbering schemas, and as a result, they are hard coded for specific assets. If you don’t know what that asset is, the ID number won’t help. By using computer vision to automatically identify an asset in real-time, you can nearly eliminate these issues. You can then reference a database like ISAN and/or other schemas to provide a complete record of that asset – and, you get it instantly without relying on pre-defined or embedded tags. <br><br>
I’m not ringing the death knell for the current systems. Rather, I propose that we take the identification process a step further and incorporate BOTH an intelligent metadata management framework with systems like ISAN in a complete, fully automated system. Imagine a system that provides a consistent method for identifying a video asset or part of an asset regardless of its resolution, bit rate or version. Then, combine that with a way to connect the asset to a standard numbering system like ISAN. It’s not quite video nirvana, but it’s a significant step forward. <br><br>
The real “shame” would be to continue to live with what we have today. Organizations are constantly challenged to manage assets without an ISAN, duplicate assets or assets distributed through advanced networks without a consistent way to determine what the asset is and associate the correct metadata. <br><br>
Let’s take a leadership role and evolve our digital media standards to meet our needs today and in the future. If we do, we have the potential to unlock significant value and new monetization opportunities – improved discovery, better ad targeting and performance, personalized experiences, higher engagement and increased viewership – and change the industry in the process.