🚀 go-pugleaf

Patents: created to encourage publication of inventions, such as chemical
processes, and that still is a very valid and useful reason for them to exist.
But it is less important in two cases: where publication happens anyway,
as is often the case with software and algorithms; and where reverse
engineering is easy, as is the case with software moreso than some other
fields such as chemistry.  These are generalizations, however, and there's
no reason someone couldn't invent something highly useful, original, and
not a law of nature that is a software program and that would be nearly
impossible to reverse engineer.

Ham radios: used, primarily, as "entertainment", or so I understand.  No
such thing as patent violation with this form of use, as far as I can tell
from my reading on the topic.  On the other hand, if one of those many ham
operators took his "design", incorporating lots of already patented
but unlicensed processes, and sold it as instructions or as kits or prebuilt
units, I suspect he'd be meeting with lots of lawyers soon.

Patent searches: ridiculous for software given what currently is patented.
I know that, in C, "x ^= y; ... x ^= y;" is an effective way to save space
using the XOR technique.  Do I have to do a patent search every time I (or
my compiler) happen to use this technique in: compiler design (Gnu Fortran);
windowing code (X); display management (cursor: BINGO, here's a patent); and
so on?  Answer: with the current patent system, YES.  Is this efficient by
any description or in line with the philosophy and reasoning behind the
patent system?  NO: it takes me far longer to do a search than to apply a
law of nature (XOR) to a new problem space; and there is little or no need
for the public to be aware that XOR has now been used for display cursors.

Optimizers: The software patents I object to are cases where laws of nature
and/or obvious things are combined in a given situation.  Optimizers are not
yet good at doing this generally, but are getting better at it.  Global
optimizers based on knowledge-base systems are very likely to do this sort
of thing within, say, 10 years.  Note that the human-knowledge data base,
CYC, is claimed to have "discovered" (invented?) a few new interesting
things when its nightly daemons look over the knowledge entered during the
day and look for discrepancies, redundancies, and "interesting" combinations
of knowledge producing "nonobvious" results.  (See some issue of, I
believe, Communications of the ACM -- send email to me if you want specifics,
my stack of mags is too high to look it up right now.)  If this project is
able to produce these kinds of results for human knowledge, obviously (to me
at least) forthcoming optimizers are likely to be capable of doing things
currently patented.

Hysteria: I'm not hysterical, I'm simply concerned that the patent office,
in tandem with various corporations, are abusing the purpose of patents in
such a way as to ultimately make the computer software industry too
expensive to be in as an independent designer, without yielding sufficient
advantages in the process.  End users will end up having to pay for all those
patent searches (they DON'T cost $10 if they take time -- time costs $$),
avoidance of advanced optimizers or time spent analyzing their output for
patentable techniques, and so on, and this will artificially constrain the
market for software.  I have a vested interested in keeping the computer
software market as free, open, and LARGE as possible, and growing.

Vested interests: interesting that one of the people most outspoken FOR
software patents writes them for a living.

Personal knowledge: I am about as certain that the patent office refuses
to hire examiners whose qualifications are a degree in Computer Science
as they would those with a degree in Chemistry as I am that the U.S.
recently participated in military activities somewhere in the middle east.
I've read many articles in many publications on this topic that stated this;
this is a natural outcome of having spent a day at the library reading up
on patent, copyright, and trademark law when I left my full-time work to
do volunteer work for the FSF and contract work.  Why does someone who
apparently doesn't read articles on the topic ask me if I have personal
knowledge of something?  All he has to do is read articles.  If he has
read them and seen mention of this claim about Patent Office hiring
practices and disputes it, then why not phrase the question accordingly?
We can't all have "personal knowledge" of everything we state publicly;
else we couldn't say much of anything at all.

Motivation: I'd like to be able to use my own inventions in my work as
a contractor, volunteer, whatever, and be free to contribute or sell
software incorporating these inventions.  To me, they're not inventions
as defined by the patent office -- simply using XOR to update a ROM-based
cursor on a display screen to me is no invention.  I don't think it is
enough to say "don't worry, you won't likely get sued if you stay small",
because even though I intend to stay small, I don't live my life breaking
the law only as long as I can get away with it: for example, I don't litter,
and I even try not to waste water even though there's no law against it.  If
the US considers it wrong for me to use any arbitrary number of my own
inventions because someone else patented them (remember the patent process
awards the patent independent of how many people came up with the idea around
the same time without knowing of each others' work), and if the set of
possible violations is too large (as is clearly the case with the current
set of software patents, given the examples I know of), then for me to feel
comfortable, I must either incur the huge expense (time and money) in doing
all the necessary patent searches (and still bear the risk of patents
pending I don't know about) or, more likely, get out of the software
business.

More to the point, I put the odds at greater than 40% that I've violated
a published or pending patent in my work on GNU Fortran, without knowing it,
simply because it is fairly large (over 100,000 lines of code in the front
end, the part I've written from scratch).  But to determine whether I have
indeed violated a patent, the expense of the search would be such to prevent
me from doing any more volunteer work for the FSF.  Yet there would be
no tangible benefits to society from the patentability of the inventions
I reinvented, certainly insufficient to offset the loss of a free Fortran
compiler for UNIX systems.

But I also want to be able to patent truly unique, nonobvious, non-laws-of-
nature things I invent that happen to be implemented in software.  So I'm
not for abolishing all patents, or even all "software patents" (depending
on how you defined the term).  But I am in favor of abolishing the idea
that applying an obvious or law-of-nature technique to a particular
situation is patentable, as is the case with XORing cursors or backing
store.  (Note that it is still not clear to me whether AT&T has a patent
on backing store or on an optimization based on a unique, nonobvious insight
into how windows are actually used in most cases.)
--

James Craig Burley, Software Craftsperson    burley@ai.mit.edu