[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [dvd-discuss] Gedanken Experiment -Unix and Norton





microlenz@earthlink.net wrote:

> > Thus the idea "an order 1 sort scheme" shouldn't be patentable, but any
> > actually implementable, (and of course original/novel) algorithm should
> > be IMHO.
> 
> Why? Why do they need protection? WHy should we want to give them protection?
> Is there a strategic reason to do so? Is there something driving the market? Do
> we want to encourage research? 

We want to encourage disclosure.  There is plenty of research, but a
large chunk of those results aren't published.  What is a
state-of-the-art "place and route" algorithm? Something buried in the
guts of Cadence Avanti, et. al. and never disclosed as this is part of
their competitive differentiation.

> I don't see that the state of algorithm
> development is in that bad a shape.  

If this is true, then you have little to worry about.  If the really
cool stuff is already being disclosed, then it can't be patented.  The
only possible change is that MORE algorithms will be disclosed, and
those that are novel and original will be patented. By definition the
"novel and original" is stuff we never heard of, and wouldn't be
available anyway this is a net-zero-loss in the near term (worst case)
and a net-gain after 20 years.

> There are thousands of papers published
> yearly. 

There are millions of lines of unpublished (and from the copyright
holders view) unpublishable source code.  I'll trade off the publication
of source for the patent of algorithms any day.

(1) because the closed source approach is:
(1.1) inherently anti-competitive (compatibility is very difficult at
best)
(1.2) dangerous to "open source" c.f. SCO ("I can sue you for infringing
something I never had to disclose to you I had..." )  Any V.C. looking
at an open source business case is going to want an answer to that
problem
(2) The disclosure of source would establish a huge body of "prior art"
limiting the potential for algorithmic patents
(3) unpublished source does more harm to "promoting progress" with it's
copyright based "infinite inaccessibility" (c.f. WOM (write-only memory)
spoof April 1st EE Times many years
ago) than does the 20 years of "unavailable without license" of patents.

> LOTS of programmers get to use that grist to crank out their programs.
> The biggest fight seems to be to prevent people from getting a strangle hold on
> the field (e.g, LZW patents and GIF formats) but I also see no reason that
> having not had protection before, they should get it now Algorithms are the
> building blocks of programs. You want protection for the building blocks?

Yep, but in trade for being able to see them all.  The only potential
bad news is that some "really cool algorithm" we never heard of won't be
freely available for 20 years.  This is however better than this
self-same "really cool algorithm" never seeing the light of day (or not
for 95 years).  The building blocks we know (anything disclosed more
than 1 year past), won't be patentable, only the stuff that we didn't
know before. This is the fundamental bargain of the patent.  At the (now
meager seeming) 20 years this is a good value (certainly compared to
copyright) for society in non-software inventions. 


> The other problem that I have pointed out before is that there is no such thing
> as an ALGORITHM? What's that he type?  There is no such thing as an ALGORITHM.
> There are only classes of algorithms and the variations on them. Do you REALLY
> want to open up the notion of patenting of algorithms to the intellectual
> property community? Just imagine the interlocking lawsuits as the algorithm you
> didn't know was patented was the nth improvement! Now you've infringed all of
> them!

There's no such thing as an invention in exactly the same sense.  Read
patents sometime.  They are "device A which implements device B with
additional claims C, D, E, and F".  However, the point is that with the
large body of prior art that would flow from the mandatory publication
of source, there will remain very little that is truly novel.

> There is no evidence it is needed now and there are reasons abound
> for why it should NEVER be allowed

"The defense would like to enter these millions of lines of closed
source programming into evidence... unfortunately we can't" ;->

.002