[whatwg] Removal of Ogg is *preposterous*

David Hyatt hyatt at apple.com
Wed Dec 12 11:12:57 PST 2007

On Dec 12, 2007, at 6:38 AM, Elliotte Rusty Harold wrote:

> David Hyatt wrote:
>> Fear of submarine patents is only one reason Apple is not  
>> interested in Theora.  There are several other reasons.  H.264 is a  
>> technically superior solution to Theora.  Ignoring IP issues, there  
>> would be no reason to pick Theora over H.264.  Everyone wants an  
>> open freely implementable codec, but it doesn't follow that Theora  
>> should automatically be that codec.  About the only argument I've  
>> heard in favor of Theora is that "it's open", but that is an  
>> argument based purely on IP and not on technical merits.
> Openness is a prerequisite. Technical adequacy is a prerequisite.  
> The technically best solution is not a prerequisite. In case it  
> isn't obvious yet, an open, adequate format is preferred over a  
> better proprietary one.

I don't think that is obvious at all, especially when the <video>  
tag's chief competition, Flash, is using the technically superior  
solution.  Why would authors switch away from Flash if <video> doesn't  
offer any technically compelling reason to switch?

>> If you consider mobile devices that want to browse the Web, then  
>> depending on the constraints of the device, a hardware solution may  
>> be required to view video with any kind of reasonable performance.   
>> A mandate of Theora is effectively dictating to those mobile  
>> vendors that they have to create custom hardware that can play back  
>> Theora video.  Given that such devices may already need a hardware  
>> solution for existing video like H.264, it seems unreasonable for  
>> HTML5 to mandate what hardware a vendor has to develop just to  
>> browse Web video on a mobile device.
> Thanks. I wasn't previously convinced we needed to mandate *any*  
> particular format, but you just convinced me. If hardware is support  
> is required for some devices, then it does indeed sound like a good  
> idea to mandate some minimum level of conformance. It is far better  
> that this minimum level of conformance be an open, freely  
> implementable standard such as Ogg/Theora than a known patent  
> encumbered format such as H.264.

Good.  I also believe there should be a mandated baseline.  That's why  
I think SHOULD is too weak, and that we should be working towards a  

>> Or put another way, imagine that GIF was an open format but PNG was  
>> IP-encumbered.  Would you really want to limit the Web to  
>> displaying only GIFs just because it was the only open image format  
>> available?
> Please stop attacking straw men. No one has suggested that. Under  
> those circumstances, I absolutely would support requiring all  
> browsers to display GIFs. This would not prohibit them from also  
> displaying PNGs if they chose to license the relevant patents.

Right, but, continuing the analogy, the issue you run into is if the  
Web at large considers PNG to be superior and just ends up using it  
anyway, then specifying "SHOULD use GIF" is rather irrelevant.  I do  
not think people will switch to <video> using Theora when a  
technically superior alternative exists that will also work in  
Internet Explorer (Flash).  We have to make sure that <video> is on  
par technically with what Flash can do.

>> Technical arguments are relevant here, so take some time to  
>> consider them before accusing people of having shady ulterior  
>> motives.
> Technical arguments are relevant, but do not control. They are  
> neither the only nor the most important consideration.

Similarly an inadequate open standard should not be proposed as the  
only way forward simply by virtue of its openness.  Wanting an open  
standard does not mean that Theora should just be automatically chosen  
to be that open standard.  It is also a logical error to assume that  
openness is not desired by a vendor merely because one potential open  
format is not approved by that vendor.

(hyatt at apple.com)

More information about the whatwg mailing list