[whatwg] Removal of Ogg is *preposterous* , SHOULD, and other matters

[ddailey]

> On Tue, 11 Dec 2007, L. David Baron wrote:
>>
>> In this case, most implementors following the SHOULD and implementing
>> Theora might help companies whose concern is submarine patents become
>> more comfortable about shipping Theora, especially if some of the
>> implementors are companies similar in size or wealth to those
>> non-implementors.

Hixie replied:
> As it stands, all the vendors who would implement Theora due to the SHOULD
> in the spec already are implementing Theora.

David's note got me to wondering if inclusion of the SHOULD language by the 
W3C might ultimately reduce the liability to companies who actually 
implement Theora. That is, a judge who discovered that the W3C acted in good 
faith in attempting to find an unencumbered codec when it in turn 
recommended to Big Company A that it should use Theora might be quite 
receptive to A's defense against Scavenger S's claim against A. I don't know 
if patent law (like copyright law, at least in the US) makes allowances for 
"innocent infringement," but if it did that would certainly lend some 
protection to both W3C and those who might follow its advice. This would be 
a question for the attorneys who I gather will ultimately be called in to 
help the W3C WG with its deliberations.

Another question of a similar nature: while I understand that Big Company A 
might indeed extend its vulnerability by actually conducting patent searches 
(various aspects of law seem to be likewise counter-intuitive, even 
paradoxical), would that remain true if Big Companies A, B and C were to 
underwrite a large-scale patent search by W3C? W3C might be able to shield 
the sponsoring companies from whatever incidental discoveries it made midst 
its deep search and hence limit their liability.

Re-iterating some things I said earlier, either there is wiggle room 
remaining to create a new video (or audio) formats in the gaps between 
existing patents or there isn't. It seems unlikely that all available space 
has been carved out particularly given that JPEG and GIF87 are already out 
there and given the requirement that a patent be nonobvious. Conceptualizing 
sequences of video frames as a time-based spatial frequencies analysis seems 
obvious. From there it would seem that almost infinitely many data 
compression schemes exist. For example, one ordinarily would tend to match 
the redundancies of frame i with those at allied locations in frame i+1. 
Suppose we consider an arbitrary frame to be a clipped rectangular subregion 
of a larger realm over which the camera actually moves. Then the compression 
technique might consist of first building hypotheses about the larger realm 
and then calculating interframe redundancy based on those hypotheses. With 
sound we have strings (of sinusoidal amplitudes) being concatenated together 
in each discrete moment in time; string similarites across moments maybe 
recast as multidimensional substring problems hence transforming what might 
look at first like a conventional Fourier analysis into something based more 
on discrete mathematics in very high dimensional space. I guess all I'm 
saying is that the number of methodologies that could be applied to the 
problems seems large and that one outcome that should not be foreclosed is 
the development of an obvious (hence non-patentable) codec from scratch with 
the collective talents of those so inclined to cooperate. If each step in 
the production of such a format is "obvious", then all of its components 
would, by definition, be patent free. If no such wiggle room exists then the 
granters of those patents have arguably been overzealous and at least some 
of those patents must, it seems, be invalid.

Something that is suggested to me in the arena of international treaty work 
on IP harmonization that the W3C may be interested in adding its voice to 
would be large scale indemnification -- WIPO working in conjunction with W3C 
or some such thing. Certainly, reform of patent law is apparently mandated, 
though doing such work on a country-by-country basis seems like slow work. 
In the world of Real Property, the common law concept of eminent domain or 
compulsory purchase is extended as a power to governments to allow for 
creation of technologies (like roads or utilities) that would otherwise be 
encumbered by known molecular obstacles (like barns or fast food 
restaurants). When those obstacles become invisible and non-molecular (in 
the world of IP) and when they fail to have coordinates in Euclidean space, 
the regional jurisdiction of the "government" seems ill-suited to deal with 
those obstacles. Creating a treaty which allows the W3C to "condemn" a 
patent that I might hold might give a bit too much power to some folks (and 
I can imagine a zillion folks, and twice that many bots, voting against such 
a treaty) but in the long run. it might be necessary to think such thoughts 
in order to allow interoperability on our info-highways.

cheers,
David Dailey