Re: Image Provenance

Hi David,

On May 13, 2008, at 7:34 PM, David Dailey wrote:

>
> A couple of weeks ago, in discussions surrounding "Issue 54,"  I  
> suggested that perhaps user agents could assist with a part of the  
> problem by providing access to metadata contained in certain file  
> formats (like SVG and PNG) that allow authors of images to embed  
> provenance data.

This suggestion has come up several times and I'm in completely in  
favor of it. I think we want both client UAs and authoring tool UAs to  
make use of the metadata embedded in the image files themselves.  
Client UAs should do so to allow users to query the metadata and  
authoring tools should do so to streamline the creation of longdesc or  
aria-described-by referenced document fragments and setting alt and  
title attribute values as appropriate.

> A colleague who is interested in GIS data (and particularly in  
> relationships between images, data, and the origins of images) --   
> approached me with the following suggestion (without any prompting  
> from me I will have you know ). It seemed within the purview of the  
> broader discussion of how browsers respond to images and what those  
> images represent.
>
> Rather than having browsers merely copy images to the clipboard  
> ("Save picture as..." and "copy" in IE, "save image ..." and "copy  
> image to clipboard" in Opera, etc.) , hence streamlining copyright  
> infringements and making the job of plagiarizing that much easier  
> for our students, how about the browsers would, instead, copy the  
> entire <img> tag, including src and alt (if it exists), as well as  
> the URL in which the image were found, as well as (possibly) some of  
> the textual context, to the clipboard?
>
> For those working with images that are in the public domain (like  
> geographers working with US govt data, or art historians), the  
> suggestion is no less relevant: images are inextricably bound  to  
> their provenance for most academic purposes. In the case of  
> copyrighted images, the browser makers would seem to be on far more  
> solid footing in the eyes of the putative courtroom should they make  
> innocent infringement a bit less likely.
>
> Now, how applications such as web authoring packages, word  
> processors, presentation software, etc. would deal with the richer  
> clipboard data, would be up to the application developers, but  
> presumably, the already rich data structure of the clipboard in both  
> Windows and Mac environments can already deal with a properly  
> structured bundle of imagery-with-data.
>
> In cases of images being rebroadcast from site to site, the ruling  
> in the US case Kelly v Arriba would seem to support the notion that  
> the copying of provenance along with images strengthens what in the  
> US is known as "fair use" and in other jurisdictions is sometimes  
> called "fair dealing." If such metadata were incorporated with the  
> copying act enabled by browsers, then it is possible that the number  
> of alt-less images floating around on the web would, over time,  
> decrease, since access to reliable origination data would be  
> enhanced for both client and server.

On this suggestion, my feeling is more that this information should be  
included more and more in the media file's embedded metadata. In this  
way its made exceedingly difficult for the relevant metadata to get  
separated from the image itself (unless done so deliberately). IPTC  
defines a fairly complete set of metadata properties for inclusion in  
web related media formats. Others have extended this to include  
copyright. We may want to liaison with other groups to specify  
metadata properties for subject description, visual description and  
aural description that accept HTML rather than merely plain text  
content.

Take care,
Rob

Received on Tuesday, 13 May 2008 22:10:37 UTC