[theora] Thusnelda Video Quality

[Gregory Maxwell]

On Tue, Mar 31, 2009 at 8:46 PM, Remco <remco47 at gmail.com> wrote:
> On Tue, Mar 31, 2009 at 6:42 PM, Michael Dale <mdale at wikimedia.org> wrote:
>> As Gregory point out we need to shape the message as theora is "free",
>> much higher quality than before and patent unencumbered. Not theora is
>> better in absolute "quality" terms.
>
> I would like to see how it holds up though. Is it better than MPEG2?
> MPEG 4 ASP? H.263? Quality is a factor that needs to have some hard
> data, because bandwidth costs money just like patent licenses do.

Do realize that you're asking about "hard data" for something which is
inherently subjective. You can only speak accurately of quality as
trends and aggregates, "I.e. X tends to be preferred over Y in this
situation by many people". You're also talking about a soft metric.
Once you're already in the realm of highly lossy video is no hard
quality X which is clearly unacceptable where X+ε (some slightly
higher quality) is acceptable.

Don't misunderstand me: Those sorts of measures are *completely valid*
ways of measuring and evaluating things, but because they are fuzzy
and soft they are categorically different from codec usage fees.

They are different in other ways as well: There is genuine and
vigorous competition in transit pricing
(http://gigaom.com/2008/10/07/wholesale-internet-bandwidth-prices-keep-falling/)
while patent licenses are, by definition, a monopoly. If you are
trying to optimize your total cost, you will have significantly more
freedom to optimize on the bandwidth side (including this like
tit-for-tat P2P file transfer; which allows the recipients to pay for
all the bandwidth, effective)

… But I wasn't failing to consider your line of reasoning when I said
people need to consider the freeness of the format. In fact, I'm
pretty confident that its still good for you both personally and for
society, to use Theora if the total cost is currently worse.

I think that I'm really failing here to make an important point about this.

I'm generally happy for you to use whatever formats you want to use.
I'd have to be a pretty boring person to get worked up over someone
elses selection of codecs. Unfortunately, your choice of formats is
not an independent personal decision, it has significant effects on
others.

If everyone had the freedom to pick the formats they use based on the
direct merits (like your proposed total cost comparison) and their own
personal preferences, and they could choose codecs which don't require
payment of monopoly license fees… then I'd say Xiph.Org's mission was
accomplished and we could just go home now. But if publishers do not
use Theora clients won't get support, if clients don't have support
publishers won't use it regardless of the merits for their
application. If publishers… A catch-22 results.

This network effect means your decision to use or not use Theora has
an unavoidable effect on my ability to use the format and by extension
my ability to avoid paying for monopoly codec licensing. (Which I may
wish to do because my cost factors are different than yours, or simply
for philosophical reasons) Moreover, the resulting "effective
non-viability" of unencumbered alternatives helps keep the monopoly
licensing fees high:  MP3 licensing costs dropped like a rock after
Vorbis was released (and enforcement diminished as well).  So even if
the specifics your application drive you to a proprietary format the
success of formats like Theora is important to your bottom line.

For most content producers the 'cost' of compatibility dwarfs other
factors.  Quality? Who cares about quality if it doesn't even *play*?
 Format decisions are *local* decisions with global effects, so for
video (and a lesser extent audio) the groups issuing monopoly licenses
are able to set their prices just below the point where it would be
less expensive for single powerful organizations to switch and pull
their users with them. As a whole society is paying far far too much
for codec licensing because the licensors benefit from intrinsic
inefficiencies in the system.

Take a look at areas where free formats have achieved widespread
adoption: Still images (JPEG, for example), Hypertext (HTML), etc.  In
these areas proprietary formats receive no air. Even though superior
proprietary alternatives exist, virtually no one bothers to use them
because compatibility is the 'cost' that matters most for almost
everyone, not quality.

If we had the mindset that people should be making their decisions
based on the short term local total cost then we might as well give
up:  Today Theora clearly loses to proprietary formats on the basis of
compatibility to enough of an extent that quality/bitrate isn't really
a factor.

In order to escape the situation we have today where free media
formats are effectively not viable due to compatibility reasons,
regardless of their technical merits, the internet community must
collectively choose to optimize *globally* when we make our local
format decisions.  We actually need to choose to use free formats even
if they don't always win on their individual merits so that we will
collectively have the freedom to choose them when they do win. Only by
doing that can we keep compatibility from dividing us and reduce the
total cost to society.

This isn't such an odd line of thinking: It's not so much different
than the enormous numbers of people choosing to pay a premium for
renewable energy simply because they know that by sponsoring it today
they can get a better result long term and for everyone. It's
precisely this kind of reasoning we need to encourage people to
consider for open media.

By adopting unencumbered formats today even though they currently
'cost' more you will be contributing to a better society, one which is
more efficient and which has more freedom, and will lower your future
media-transport related costs.  Can you see the merit in this
position?