by Tom Lord, 19th September 2008
Multimedia-Attached Metadata Expression (MAME) is a simple idea: allow all types of media files to include publisher-supplied meta-data (such as copyright licenses) and browsers to make this metadata available to users.
We often think of the web as a system of two-party, point-to-point communication: a user asks for some resource from a page author and the page author transmits that content in reply.
That's a simple conceptual model but it doesn't reflect the reality. Today's web pages often include several separately authored, often separately published elements. The page author may control the overall look and content of a page but pages also include various media files such as images, video and audio, and increasing even more esoteric media such as font files.
In that one to many communication, page authors are privileged to have complete control over what messages reach the user. Authors of linked media files are nearly silenced. For example, even if the publisher of an image wants to includes important copyright or authorship information, the web provides no mechanism for this. The only thing browsers want from an image file is a set of pixels.
MAME, which stands for Multimedia-attached Metadata Expression describes a mechanism for ending the censorship of media file publishers. The idea is quite simple: allow all types of media files to include publisher-supplied meta-data, which user agents then make available to users. We call such meta-data, since its purpose is to inform users, notices.
Notices have an open ended set of potential uses including timely topics such as conveying licensing information to users and carrying ads.
This document does not claim to convey a fully worked proposal for MAME but the remainder of this document does sketch what such a proposal would include. It is intended to be a starting point for discussion, particularly in the context of seeking alternatives to the media-file meta-data approach taken by the recent EOT proposal from Microsoft.
What new technologies are needed in order to make MAME a standard part of the World Wide Web?
The idea of "media-attached meta-data" is motivated by considering various common kinds of meta-data that it would be useful to attach to media files. For example, in the overview, above, we remark that MAME is potentially useful for conveying licensing information, carrying ads, parental guidance information and more.
It is undesirable that there be a needless proliferation of media meta-data definitions. For example, if several different media publishers each seek a meta-data format for machine-readable licensing information, and if the legal structure of that information is similar in all cases, then each party benefits if all can agree to use a common format and promote a single interpretation of that format by user agents and other applications.
The W3C Media Annotations Working Group is one example of from where such standards may come.
It would be disastrous if each type of media file required it's own separate and special rules governing the representation of this new form of media-attached meta-data: If each media file format required its own unique rules, user agents and other applications would have to include additional meta-data parsing code for each media file type supported.
Rather, MAME needs a standard mechanism for bundling media-attached meta-data in the same file as any traditional media type.
We suggest the idea of using MIME for this purpose, although that is certainly not the only reasonable solution possible.
The essay Resource Meta-data and User Notices explores this in more technical detail.
Finally, MAME requires standards for the processing and display of media-attached meta-data, the "Expression" element of MAME.
Open questions exist for such display such as:
We leave these questions open, at this time, since they require a group effort to think through various security implications and weigh these against possible applications.
Please read the essay Resource Meta-data and User Notices which explores this topic in more technical detail.