Re: [rtcweb] Stephan Wenger's choices

[Ron <[e-mail]>]

On Sat, Dec 28, 2013 at 03:17:51PM -0800, Eric Rescorla wrote:
> On Sat, Dec 28, 2013 at 3:03 PM, Ron <[e-mail]> wrote:
> > None of which helps people who have a device they bought in the last year
> > which could perfectly do multiple streams of H.261 in real time, but has
> > no hope at all of keeping up with even one stream of a higher complexity
> > codec.  Even if they have a true colour retina display available to them.
> 
> I don't see this as the tragedy you seem to. And I would note that
> FirefoxOS actually runs on devices that are at least potentially in
> this class, so it's not because it's a situation I don't have to deal with.

I never called this a tragedy.  I just pointed out that if people are going
to say "won't somebody think of the mobile devices!" (which have hardware
H.264 that pretty much every WebRTC app will be unable to use) -- then we
either need to actually think of those devices, and consider a codec that
has a low enough complexity that they'll actually be able to use it -- or
we need to decide they are in fact irrelevant, and any arguments based on
them are then equally irrelevant too including for H.264 or any other codec.

I'm happy to let consensus decide which it will be.  But I'm going to call
bull when people try to ride both sides of that to make blinkered claims
that strain to find any coherent connection with reality.


> > We can joke about absurdities, but I'm completely serious about the real
> > benefits that a low-complexity, tiny-footprint, completely accessible
> > codec would have as the MTI.
> 
> Yes, I realize you're serious. But it turns out that there is a difference
> between seriously believing something and actually convincing people.
> 
> So far you have yet to convince me, at least, that it's better
> for users to have an MTI bad codec on every device than to have
> an MTI good codec on most devices and no video on really low-end
> devices.

And that's fine, but if the limit of your technical rebuttal is reduction
to absurdity and "I'm not convinced", then you're not making a very strong
case for what is so bad that you feel it must be avoided at any cost.

How can a codec that provides perfectly legible sign language, and that a
significant portion of users probably couldn't pick for being worse than
any other alternative, and that has none of the IPR problems that everyone
is deeply concerned about regarding both the first round options, possibly
be worse than nothing at all?

I'm not rusted on to any alternative here, so I'm inviting you to convince
me if I'm not convincing you.  But you're not offering up much more than
264 legs good, 261 legs bad.  What is so terrible about the video equivalent
to G.711, that's been shipping in successful commercial products for years,
that makes it too repulsive to even consider as the fallback of last resort?

If your reasoning is strong, that shouldn't be a hard question to answer
with plain engineering facts.

  Cheers,
  Ron