software patenting [from Manny Farber via Editor ..djf]

[efarber () iiic ethz ch <efarber () iiic ethz ch>]

I chopped the list otherwise this would have been a 250KB message.


Manny




                         TITLES TO 2700 SOFTWARE PATENTS
                             ISSUED IN 1992 and 1993


                                 Greg Aharonian
                          Internet Patent News Service
                         P.O. Box 404, Belmont, MA 02178
                             patents () world std com
                                  617-489-3727


    In January and February, the Patent Office is holding hearings in San Jose
and Washington on the "problem" of software patents.  The hearings will focus
on two aspects: the nature of software patents (legal, economic and software
theory), and the process of awarding software patents (in particular, software
prior art).  To help these hearings be more empirical, I have prepared a list
of 2700 software patents issued in the past two years.  It should help people
base assertions in the actual statistics of issued software patents (of which
there are over 11,000).


     The patents listed below were issued in all of 1992 and the first half of
1993.  The list is based on what I consider to be a software patent, though it
does reflect the nature of software patenting in the US, and does include
software patents from all categories of software for which patents are being
sought.  And given the Patent Office's classification scheme for software
(which they are revising), this was not a very easy list to compile.


     Despite the tremendous numbers of software patents being issued (with
respect to the small numbers of true software innovations each year), software
patenting as of yet is not a really big socioeconomic problem, though unless
the software community acts now, it will become a big problem.


     A series of diskettes with more detailed information on each software
patent listed below (full title, number, assignee, abstract, classification,
application date) are available for $595.  Also provided are diskettes with
9000 titles from my software prior art database, and version 1.0 of my patents
claim analysis program.  Proceeds go to my project of making the Patent Office
APS patent text files Internet accessible.


     For more information on the Internet Patent News Service, send  'help'
in the body of a message to   patents () world std com   on the Internet.


==============================================================================


     Here are some observations based on examining these software patents:


     1) The most minor of software concepts can be patented.


Despite endless series of articles referencing all sorts of court cases and
judicial precedents, you can receive a software patent for the most minor of
software concepts, as long as the patent application is prepared properly. For
example, the Air Force received a patent for nothing more than the difference
between two versions of a public domain computer program.  IBM received a
patent for putting a scrolling database field in a window. Fourier transforms,
spline interpolation, APL-to-Fortran translation, graphics window clipping,
alphanumeric sorting, operating system utilities, and other ideas which have
been written to death in technical journals.  The following list has 350 image
processing patents for a year and a half.
                   Any software concept is patentable. Any.




     2) Extremely broad software patents can be acquired.


While everyone knows about Compton's multimedia patents, and Action's workflow
patents, Digital Equipment Corporation recently received a patent which very
broadly claims all case-based reasoning technology, as did Ehud Shapiro for
all concurrent logic programming technology.  Most independent claims in
software patents are too broad, and should include the Jepson phrase "...the
improvement comprising...", if this style wasn't so discouraged.  Of course,
such broad claims are rarely defendable, so why allow them?




     3) Hardware and software patents are logically equivalent


With one CAE tool you can specify an algorithm in VHDL (a programming language
similar to Ada) and execute the algorithm as you would any other algorithm (a
software algorithm then as worthy as any other to be patented, a software
algorithm usable in any application through a Windows DLL), and / or generate
a circuit diagram and logic cell component list and have an integrated circuit
produced by a foundry (a hardware device then as worthy as any other to be
patented). This ability blurs and eliminates any distinctions between hardware
and software, causing a new set of headaches for the Patent Office.  Take a
software idea, patent it in a hardware form (look at some neural net patents),
and then challenge software infringers in court using the doctrine of
equivalence.


The patent hearings in San Jose as they are now focused suffer slightly from
lack of consideration for this aspect of software.  The Patent Office needs
people inhouse with experience in VHDL, CASE and predicate logic to start
grappling with the problem of software/hardware equivalence.  It will be
impossible to modify the patent rules for software in isolation, since this
equivalence opens up a loophole.


(And to save the Patent Office from having to deal with one more headache, I
won't mention how within twenty years you will be able to implement software
algorithms using gene sequences).


     4) Mathematical algorithms are patentable


Forget whatever anyone says on this issue - the following list of issued
patents shows that mathematical algorithms patents are being issued.  Whether
or not they are defendable, beneficial, or relevant - they are being issued
on a large scale.  Check out the Numerical Analysis, Compression, and Signal
Processing sections of the following lists - mathematical algorithms pure and
simple.  Thus when the Patent Office says (quoting from their announcement of
the hearings):


     There are three general categories of exclusions to patent eligibility
     that are particularly relevant to software-related inventions. The first,
     and most commonly applied exclusion, is the exclusion of mathematical
     algorithms, per se, from patent eligibility.  For a summary of the law
     governing this exclusion, and for guidance on how the PTO applies this
     exclusion in the context of its examination procedures, see "Patentable
     Subject Matter, Mathematical Algorithms and Computer Programs",1066 OG 5,
     (Sept. 5, 1989) and "Note Interpreting In re Iwahashi," 1112 O.G. 16
     (March 13, 1990).


well, this statement conflicts with the Patent Office's own awarding of what
I consider to be mathematical algorithm patents.  And few at the Patent Office
have experience with VHDL to realize how VHDL can undermine the above stated
examination procedures by allowing algorithm patents implemented in hardware.


     5) Methods of doing business are patentable


Forget whatever anyone says on this issue - the following list of issued
patents shows that methods of doing business patents are being issued. Whether
or not they are defendable, beneficial, or relevant - they are being issued
on a large scale.  Check out the Financial, Office Automation and Networking
sections of the following lists - many are business methods pure and simple.
Thus when the Patent Office says (quoting from their announcement of the
hearings):


     Second, methods of doing business are excluded from patent protection.
     While no cases have directly applied this exclusion to deny patent
     protection for software-related inventions, the exclusion is relevant for
     questioning the patent eligibility of processes that are modeled upon
     existing business processes but are implemented through a software-based
     system.  See, Paine, Webber, Jackson & Curtis, Inc., v. Merrill Lynch,
     Pierce, Fenner & Smith, 564 F. Supp. 1358, 218 USPQ. 212 (D. Del. 1983).


well, this statement conflicts with the Patent Office's own awarding of what
I consider to be method of doing business patents.  Technologies such as ISDN,
groupware, and traditional process control all enable these patents to
infiltrate the system.  This is another example of the hardware/software blur,
as patented ISDN applications, while traditionally considered hardware due to
the networking aspects and ISDN's hardware infrastructure, are essentially
software patents due to the programming of ISDN applications.  ISDN patents
matching incoming phone calls to customer information databases are methods
of doing business patent implemented in software.


     6) Software patent litigation is not very active


Compared to other technologies, software patent litigation isn't big business.
I see more cases dealing with biotechnology, electronics, mechanical devices
and chemistry, than I do with software.  Most software patents are either
easily work-aroundable or are easily challengable for prior art (though Roger
Billings is having more success against Novell than I thought he would).
Equivalently, software patent infringement lawsuits are no more an impediment
to conducting business than patent infringement lawsuits in other industries.
The notoreity for software patents is probably due to the Internet and the
ease of electronic mail.


Also, despite all of the talk about software patents, no one has ever prepared
and distributed the following list - the Patent Office, the bar associations,
the Software Patent Institute, etc. No one. (Of course one reason may be that
looking at this many software patents is somewhat mind-numbing :-).  Serious
discussions and policy planning can only be done with the empiricism of
knowledge about existing software patents and software prior art. If software
patenting was a really big problem, my phone would be off the hook.  It isn't
because they aren't. For now.


     7) 25% of all software patents could not survive reexamination


I maintain the largest software prior art database in the galaxy.  Based on
public software components and algorithms I have reviewed over the last ten
years, numbering in the tens of thousands, I can successfully challenge at
least 20% of most software patents being issued.  Another five percent are
faulty for not providing enough information to implement the software concept
being patented - if I can't create working source code from your patent
description, you shouldn't get a patent.


Partly I blame other agencies of the government for this problem.  They fund
billions of dollars of software research and development each year, yet they
neglect to inform the Patent Office of the fruits of their results.  At least
two thirds of all software prior art is due to government funded research,
none of which is shared with the Patent Office.  I have closely examined the
stacks of the Patent Office research library - you would never know that the
US government funds research of any kind based on the contents of the stacks.
These agencies owe the Patent Office a lot more assistance.


     8) Software patent examiners are being asked the impossible


A software patent examiner has an average of two to three hours per patent
application to verify and/or conduct the software prior art search.  Even
working in Boston, where I use some of the best technical libraries in the
world, along with my own software prior art and software patent databases,
it takes me at least a few days to investigate a software prior art problem.
If you figure a few weeks for others, including software patent examiners,
one of the most important problems of the software patenting process arises:
software prior art searching.


As in observation 7), some blame is due to other government agencies providing
little technical assistance to the Patent Office, especially in providing
organized information on results of their software research and development.


==============================================================================


     I use 42 categories of software patents, listed below, though a few are
so big that I will probably split them up.  Some patents are listed in more
than one category to simplify your search for relevant patents.  Many of the
titles have been abridged without losing their meaning, to fit on one line.
Some titles start with '--'.  This indicates the assignee is IBM.  I do this
because IBM has 1/8 of all software patents, has made money off its software
patents, and it is interesting to see their software patenting strategy.






                        IMAGE PROCESSING
                        OPERATING SYSTEMS
                        NETWORKING and COMMUNICATIONS
                        PROCESS CONTROL and NUMERICAL CONTROL
                        GRAPHICS
                        GRAPHICAL USER INTERFACES
                        DATABASE
                        WORD PROCESSING and DOCUMENT HANDLING
                        FINANCIAL
                        COMPUTER AIDED SOFTWARE ENGINEERING
                        ARTIFICIAL INTELLIGENCE and EXPERT SYSTEMS
                        ENGINEERING
                        CIRCUIT DESIGN / COMPUTER AIDED ENGINEERING
                        NEURAL NETWORKS
                        SECURITY AND ENCRYPTION
                        MULTIPROCESSING
                        PATTERN RECOGNITION
                        MEDICAL AND HEALTH
                        SIGNAL PROCESSING
                        SPEECH RECOGNITION and SYNTHESIS
                        COMPRESSION
                        NUMERICAL ANALYSIS
                        COMPUTER AIDED DESIGN
                        AUTOMOBILE
                        OFFICE AUTOMATION
                        GEOPHYSICAL
                        NATURAL LANGUAGE ANALYSIS
                        CHARACTER RECOGNITION
                        PARALLEL PROCESSING
                        BIOLOGY
                        VISION
                        DISTRIBUTED PROCESSING
                        ALGORITHMS and DATA STRUCTURES
                        OBJECT ORIENTED PROGRAMMING
                        ROBOTICS
                        FUZZY LOGIC
                        SIMULATION
                        EDUCATION
                        PHYSICS
                        CHEMISTRY
                        SPREADSHEETS
                        VIRTUAL REALITY




------------------------------------------------------------------------


                        IMAGE PROCESSING


Inspection control system and method
Reading a document and processing the image
Registering image data
Manipulable three-dimensional projection imaging method
Eliminating scattered .gamma.-rays and reconstructing image
Method and apparatus for image reproduction
--Pseudo-bar code control of image transmission
Adaptive coding and decoding of frames and fields of video
Transmitting encoded image data with quick image expansion and contraction
Phase corrected complex-difference processed magnetic resonance angiograms
Technique for representing sampled images
Generating image data representing integrated image
Image recognizer and method for recognizing image
Detecting the position of an object pattern in an image
Making image conversions with error diffusion
Motion vector detection and band compression apparatus
Calibrating an x-ray scanner from image of at least one calibration standard
Image processing for moire reduction and/or random dot generation
 ...