Subject: Re: Patent prior art database
From: "Rob Cameron" <robc@international-characters.com>
Date: Thu, 3 Jul 2008 07:28:09 -0700
Thu, 3 Jul 2008 07:28:09 -0700
As I'm the CTO of the patent-based open source company that Larry
Rosen is assisting, perhaps my comments will be useful.   I can also
give a bit of an update on what we're doing.

In answer to the main question, the answer is to publish in multiple
ways, much as Tom suggests.   A USPTO filing is ultimately the
best to ensure that your technology can found and used as prior
art.   However, I believe that it must be a full, nonprovisional filing.
The USPTO fees for this are less than $500.

A provisional filing must be followed up within one year with a full
application, or it is considered abandoned.   It would not then be
considered prior art (by itself) because it is not published.

Filing a full application will ensure that the application will actually
be catalogued according to the USPTO classification system and
published.   The published application database is routinely used by
examiners in the prior art search process.  Publication occurs
18 months after initial filing.   (So make sure that you also publish
in another forum at the time of filing the application.)

However, if you use this strategy, I encourage you also to make
a patent dedication that your patented technology (if granted) will be
available free for open source.

I can also comment on peer-to-patent having participated on both
sides (reviewing applications and submitting prior art; putting two
applications through the process.)   This is currently a pilot program
to test the idea of oprn submission of prior art before applications
are granted.    It is a good idea, but difficult to implement well.
However, I suspect that the results will be good enough, and the principle
important enough that the project will continue.

As for what International Characters is doing with its high-performance
XML technology, we are continuing the patent-based open-source
strategy.   We now have two code repositories available under OSL 3.0.
Our high-performance UTF-8 to UTF-16 transcoder is available at
http://u8u16.costar.sfu.ca/ and is quite stable.  It demonstrates some
of the most complex algorithms with our technology.   Our ongoing
development of a high-performance XML processing stack is available
at http://parabix.costar.sfu.ca/ - a work in progress.   However, we
are able to demonstrate end-to-end performance in XML applications
that is 2.5X to 10X better than Xerces or Expat on various measures.

I've recently returned from making an invited presentation on our technology
at the IBM TJ Watson Research lab.   We are looking forward to research
collaboration to further develop our technology (particularly in coupling
the intraregister parallel bit stream technology with the intrachip
parallelism
of multicore processors).   We hope to continue to do so in an open way.

I must say, however, that we have not yet been successful in engaging
the open source community in public participation with our projects.
Several factors are probably in play, including the difficulty of SIMD
programming in general, the added difficulty of dealing with parallel
bit stream technology, our own inadequacies with respect to public
documentation, as well as continuing distrust of the patent-based
open-source model.   Nevertheless, we still think the model has value -
it allows small companies and their partners to develop novel technology
in an open way with significant protection against being beaten to market
by proprietary competitors with deeper pockets.

Robert D. Cameron, Ph.D.
CTO, International Characters, Inc.
http://international-characters.com/


As I'm the CTO of the patent-based open source company that Larry
Rosen is assisting, perhaps my comments will be useful.   I can also
give a bit of an update on what we're doing.

In answer to the main question, the answer is to publish in multiple
ways, much as Tom suggests.   A USPTO filing is ultimately the
best to ensure that your technology can found and used as prior
art.   However, I believe that it must be a full, nonprovisional filing.
The USPTO fees for this are less than $500.  

A provisional filing must be followed up within one year with a full
application, or it is considered abandoned.   It would not then be
considered prior art (by itself) because it is not published.   

Filing a full application will ensure that the application will actually
be catalogued according to the USPTO classification system and
published.   The published application database is routinely used by
examiners in the prior art search process.  Publication occurs
18 months after initial filing.   (So make sure that you also publish
in another forum at the time of filing the application.)

However, if you use this strategy, I encourage you also to make
a patent dedication that your patented technology (if granted) will be
available free for open source.

I can also comment on peer-to-patent having participated on both
sides (reviewing applications and submitting prior art; putting two
applications through the process.)   This is currently a pilot program
to test the idea of oprn submission of prior art before applications
are granted.    It is a good idea, but difficult to implement well.
However, I suspect that the results will be good enough, and the principle
important enough that the project will continue.

As for what International Characters is doing with its high-performance
XML technology, we are continuing the patent-based open-source
strategy.   We now have two code repositories available under OSL 3.0.
Our high-performance UTF-8 to UTF-16 transcoder is available at
http://u8u16.costar.sfu.ca/ and is quite stable.  It demonstrates some
of the most complex algorithms with our technology.   Our ongoing
development of a high-performance XML processing stack is available
at http://parabix.costar.sfu.ca/ - a work in progress.   However, we
are able to demonstrate end-to-end performance in XML applications
that is 2.5X to 10X better than Xerces or Expat on various measures.

I've recently returned from making an invited presentation on our technology
at the IBM TJ Watson Research lab.   We are looking forward to research
collaboration to further develop our technology (particularly in coupling
the intraregister parallel bit stream technology with the intrachip parallelism
of multicore processors).   We hope to continue to do so in an open way.

I must say, however, that we have not yet been successful in engaging
the open source community in public participation with our projects. 
Several factors are probably in play, including the difficulty of SIMD
programming in general, the added difficulty of dealing with parallel
bit stream technology, our own inadequacies with respect to public
documentation, as well as continuing distrust of the patent-based
open-source model.   Nevertheless, we still think the model has value -
it allows small companies and their partners to develop novel technology
in an open way with significant protection against being beaten to market
by proprietary competitors with deeper pockets.   

Robert D. Cameron, Ph.D.
CTO, International Characters, Inc.
http://international-characters.com/