Re: We must fork the SVG standard (was: SVGA 1.0 uses RAND ->DO NOT ! implement it, DO NOT ! use it)

Claude Zervas wrote:
> 
> > Chris Lilley wrote:
> >....
> >You have not shown that SVG is encumbered.
> 
> But neither you sir, nor the W3C, have shown that the SVG is ***NOT*** encumbered.

Correct. There is neither proof nor disproof; instead there is
information, which  developers must interpret as best they can. Allow me
to digress a little into software patents in general.

If software patents exist, it would be better if they were only waared
for astonishing clevernes, ingenuity and originality. But they are not -
the bar for consideration of prior art and obviousness to someone
skilled in the field is extrodinarily low. 

Plus, they are not written by engineers, but writen an a deliberately
vague and obscuring legalese so that when it is callened in court, there
is easy 'cannon fodder' bogus claims that can be given away and agreed
invalid to protect the core claims. This is legal manouevering, not
technology.

Lastly, patents are not really validated when issued. They are validated
if people assume they are valid and act as if they are valid; and their
ultimate test is unfortunately left to the legal system.

In addition, W3C producers specs and developers produce software
implementations of them. So it might be that one reasonable way to
implement something is subject to a patent claim and another, equally
satisfactory way is not.

So W3C cannot say "there is a patent on such and such, therefore clearly
it is valid and applies to a section of the spec." We can only say
"there is a patent claim and here is the number, go look it up and see
if you believe it is valid and if so, whether it applies to the way you
have written your code."


> This makes the SVG proposal suspect and unattractive to implementors 

Only because people have chosen to focus on it. Thats a public relations
issue, not a technical or legal one. The same is true of any other
specification from anywhere else.

> until the
> remaining RANDs have been converted to RFs.

I agree that having those few licenses which are RAND be converted to RF
would make me feel better and would make the developer community feel
better. It does not  however offer a guaranteed of immunity - just
increases the statistical likelihood. Exactly the same as any other
specification frm W3C or from anywhere else.

> I would be very hesitant to invest a lot of development effort when there is a > non-negligible
> risk of litigation (and hence financial ruin) down the road.

So, tell me, which other specifications have you implemented and what
assurances were there for those? Or did you just assume that no knews is
good news?

> As long as there are RANDs attached to a W3C proposal there exists such risks.

Well, I would counter and say that as long as there are software patents
granted for obvious and trivial bits of computer science then there is a
risk.

>  I don't
> care how much anybody bleats on about how the patents may not encumber
> the proposal. 


That sounds very much like "I want life to be simple, don't confuse me
with the facts"

> We really have no idea what the patent holders may do in the future.

In the general sense and for patent holders in general and for
specifications other than SVG in general, that is true. You have no
ideas what patents are there, who holds them, and whether they intend to
actively get money from them or passively hold them in reserve against
when someone sures *them*.

However for SVG you do have some more information because W3C called for
patents that might apply from the members of the working group that
developed the specificatio to be declared in advance. This
simultaneoously makes information available sooner and weakens the legal
case against any patents declared later by those companies.

> IBM
> or Adobe may seem friendly enough right now, but if the past is any indication one would
> be a fool to count on a continuing benevolence.

I agree that counting on continuing benevolence is foolish. It may have
worked well enough in the past but incidents such as the Unisyss claim
for GIF, the Intermind patent claim for P3P and the BT claim for
hyperlinks shows that continuing to count on benevolence is foolish.
Hence, W3C is developing a formal policy.

> All W3C proposals should be left unencumbered by at least the W3C member corporations.

Ok, thanks for the precision there. I agree that W3C can 

> If a member will not give up a RAND then the proposal should be tossed. 

You seem to implicitly suppose that all patents are valid and that all
W3C Members are benevolent (in a different way). Suppose some company
stands to loose if there is an open specification for something. They
could just join W3C, make some claim, offer RAND and according to your
proposal, they would achieve their object of immediately halting the
development of the spec regardless of how off the weall, obvious or
plain irrelevant their claim was. Thus leaving the market to be
dominated by a proprietary and possiblt undocumented format or protocol
with possibly onerous licensing terms ....

> Patents are a
> way for BigCos to monopolize, and hence stifle, innovation. 

OK, personally you will not fine me disagreeing with that one, but in
that case get your local politician to start arguing that software
patents stifle innovation, free trade and economic development.

> To support this implicitly 

Now there you have a good point, and while it seems clear to me that W3C
needs a policy to deal with those countries that have allowed software
patents, it must be very careful to not give the impression to other
countries that software patents area good thing or to be encouraged.

> by allowing RANDs attached to proposals is shortsighted and does not seem to fit the W3C
> charter.

I accept that this is your position and that of many others; I would
merely caution that it is declared patents that are the primary problem
when ecaluating a specification. Some company offering RAND that has no
patents that apply does not seem to be a problem, and merely constrains
their future behaviour should they discover or acquire through purchase,
merger etc additional patents in the future. But I agree that if the
company gave RF to start with then the problem does not arise. However,
I would rather see a company give RAND in general and then give RF on
specific patents as it becomes aware of them, than see them give no
assurances at all.

-- 
Chris

Received on Sunday, 7 October 2001 11:39:47 UTC