🚀 go-pugleaf

(Aside: why do I feel I'm having an argument with Zippy?? :-)

In article <9220@cactus.org> ritter@cactus.org (Terry Ritter) writes:

   In <BURLEY.91Sep2162405@churchy.gnu.ai.mit.edu>
       burley@churchy.gnu.ai.mit.edu (Craig Burley) responds:

   >I've never heard of anyone discovering a new
   >approach to something (inventing one) by first reading a software
   >patent and saying "aha, then if I add <x> to it...Eureka!" and I
   >doubt I ever will.

   In one sentence you assume that:

      a) we know what a "software patent" is,

A patent that does not involve/require the transformation of matter, for
example, though there probably are better definitions.  The PTO doesn't
seem to have trouble making the distinction -- just in using the results
of that distinction properly.

      b) never having heard of an experience means that it never 
	 occurred and never will,

I said "doubt" and the remainder of my post clearly showed I accepted that
such things might happen, only rarely enough to not justify the huge
costs of software patents.

      c) that the relatively small patent literature on anything you
	 might call a "software" patent can be compared to the massive
	 conventional literature.

Quite a reasonable assumption, I think.

   Do you find these debating tricks useful to confuse and defeat? 
   Alas, I had presumed that you would seriously address THE ISSUE.

   First of all, very few patents are strictly "software patents." 
   Many of what are commonly called "software" patents cover BOTH
   hardware AND software implementations.

Doesn't matter that they can be done in hardware -- the point is, they
are patents on algorithms, aka software.

   Next, with very few "software patents" issued, it is hardly
   surprising that you have not heard of the utility of building upon
   them.

There are very many software patents issued, too many in fact.  Did you
know that the "pause-one-second, send +++, pause-one-second" sequence
is patented by Hayes?  That is an example of a software patent -- even
though it involves and can be implemented by hardware, it is entirely
within the domain of algorithms and the passage of time, both of which
are "laws of nature" according to patent terminology.  Only since the '80s
has this sort of thing been patentable.  Even then, it's obvious -- we use
pauses in our speech patterns to give otherwise ambiguous words clear
meaning.  I take it you believe that without a patent on this, nobody would
have ever known about it and been able to build on it?

All of the software patents I've heard of were well-understood throughout
the industry before the patents were published.  Please give me a counter-
example of one where the patent truly published a brand-new approach to
doing something and that was in the software/algorithm domain.

   And if software designers rarely use the patent literature
   (as seems likely), it is also no surprise that few developments
   have been based on that literature.

I'd be amazed if even one had.

   If, however, your point is that there need be no patent literature
   because you think you already have access to everything in other
   journals, despite the fact that you are not acquainted with the
   patent literature, well . . . WHAT IS WRONG WITH THIS PICTURE?

What is wrong is I don't need an expensive way of finding new ideas when
I have several cheap or free ways of doing it.  The extent to which this
is true makes software very different from many other fields where patents
are allowed.

   >the huge cost
   >involved in turning software creation into 1% development, 40%
   >search for what is patentable in the development, and 59% search
   >of existing patents to avoid them far outway (sic) any occasional
   >positive result from patenting software

   What is your source?  Where do you GET these figures?  If you just
   make up things as you go, we will have very little to discuss.

I thought it was clear I was estimating.  I'm not aware of any studies
that put the lie to these estimates, in any case.

   >it is becoming increasingly likely that optimizing
   >compiler technology will be capable of violating/inventing a
   >patent given a fairly ordinary description of a problem (I already
   >know how to build such a system, but I can't describe it and must
   >keep it secret until software patents are abolished; in the
   >meantime, if I manage to build it, at least I can use it myself
   >and probably make a small fortune off of it).

   First, enough hysterics:  You CAN build ANY software system FOR
   YOURSELF, and use it BY YOURSELF, with little economic risk. 
   Patents are ECONOMIC rights, and they are really only effective to
   prevent you from seriously manufacturing, selling, or distributing
   a patented invention.

No, if you use it by yourself to do work for others, you are taking a
fairly significant economic risk.  The point is, I am not really interested
in building software for myself -- I want to write lots of software for
other people to use.  That is a widespread tendency in the software field; it
is a major reason computer technology has advanced as fast as it has; and
software patents seriously threaten to dampen it to nearly a residual level.

   Next, if you mean by "software patents," all patents which apply to
   any mechanism realized in software, such patents are quite unlikely
   to be "abolished," because this would significantly impact the
   current protection for HARDWARE logic mechanisms.  You will have
   a long wait.

It is true that patenting a hardware logic mechanism that implements
something realizable in software would be risky, given that once someone
did it in software, they were out of the reach of patent-infringement
charges.  But that raises an interesting question: if the hardware logic
is so simple that somebody can knock off an equivalent version in
software, what's the point of letting it be patentable anyway?  Remember,
hardware that drives RAM, I/O busses, and the like, would still be
patentable because software can't do it itself -- one always needs
specific hardware to accomplish that.

   But if you have a software idea worth "a small fortune," then why
   not:

      a) patent it yourself,

	 This would give you the ability to describe it in open print
	 and yet still allow you to license or use the technique for the
	 benefit of society, instead of keeping it secret and away from
	 society, as you seem to think you must do now.  You could even
	 GIVE AWAY the patent, once you have it.

If I could do this and be assured that a) I wouldn't have to undertake
a hugely expensive patent search to make sure that every combination of
features I happen to use in my product isn't already patented and b)
that patent search would include all patents in process (which it couldn't
according to patent law) and any patents likely to be applied for before
I finally wrap up and deliver mine (or at least document my own inventions
via notary or something), I'd very likely do it.  a) and b) won't happen
until software patents are abolished; a) could happen before then if
someone donated me several million dollars so I could fund the
undertaking, but it isn't likely.

      b) describe it in open print

	 This would constitute full open disclosure, and thus "prior
	 art," preventing anyone else from applying for the patent
	 subsequently (you would have a year to decide).

Same problem -- even being open about it leaves me open to patent-
infringement lawsuits.  Others might ALREADY have applied for patents
on a multitude of aspects of my ideas, some of which aren't important.

   Option (b) would seem to provide all the features you hope for in
   a world WITHOUT software patents.  Moreover, YOU can do this, RIGHT
   NOW, with virtually NO investment.  One is led to wonder why you
   do not do this.

Option (b) does not so provide.  I could try and patent the ultimate
optimizer and find out somebody has a patent on using XOR in a certain
way that my optimizer cannot avoid -- meaning I cannot distribute my
optimizer for free or license my patent for free.

   The WHOLE POINT of patents is to ALLOW people to DESCRIBE their
   otherwise secret ideas so that they can be protected.  And if you
   DO NOT WANT that protection, ALL you have to do is publish.

No, people can already describe their otherwise secret ideas, patents just
persuade them to do it more often.  Point is, released software doesn't
really contain secret ideas -- at least, not for long, if those ideas are
valuable.

   >Right now I know of a personal-computer-based page-layout program
   >whose internal file format violates a software patent.  Who is
   >responsible for this violation if the coders simply used one OO
   >library that provided a

   The courts will help you decide, if you are unable to do so.

That is just the kind of thing I want to avoid -- having the courts decide
that I am liable for paying a license fee for every freely copied version
of a program I wrote, let's say GNU Fortran if I released it myself.
Somebody could turn out to have patented, say, the mechanism I use for
representing parse state (something a bit unique, but far more clever than
the software patents I've heard of, except IBM's register-coloring
algorithm), and sue me for, say, $50 for every copy of GNU Fortran
estimated to be in use, even though I basically gave it away for free.
There is no evidence something like this couldn't happen -- for example,
AT&T apparently can stop MIT and the FSF from shipping X-Windows or
perhaps demand a retroactive license fee.

   Respecting intellectual property rights is a hassle, and WILL
   ALWAYS BE a hassle.  Shall we simply "say yes to piracy?"
   (I believe that the unlicensed use of intellectual property is
   commonly called piracy.)

Respecting intellectual property rights in this case would mean I have rights
to my own ideas even if someone else happened to come up with them first.
You wish to deny me of those rights -- even when extended to the realm of
my own thoughts implemented in computer programs I wish to distribute.
That is far more invasive than in the realm of implementing matter-
transforming machinery and selling/using that.

The word "piracy" has a particular meaning that I am not discussing here.
If I intentionally use someone else's idea illegally, it is piracy.  If
software patents are abolished, it is not piracy for me to use someone else's
software algorithm any more than it is piracy to give away a copy of
GNU EMACS.  The point is that what you call "piracy", if you are indeed, as
you seem to be, extending it to include cases where one simply has an idea
that someone else had before and uses it in a product, happens all the time
in software and far more often than in any other field where patents are
permitted.

   >Why should you be denied use of
   >your own property?  Remember I do support the idea of patents in
   >general, which can and do prevent you from, say, selling the plans
   >to your house even though it's your own design because somebody
   >else patented it first.

   Again, more hysterics.  If YOU implement ANY software technique,
   and USE it FOR YOURSELF, the ONLY risk is YOU (shooting your mouth
   off, which--admittedly--could be a risk).  Within criminal law, this
   is known as "self incrimination."  You don't have to do this.  Don't.

If I use a technique only for my own interest, there is no real risk.
If I use it as a tool to do things for others, there is a definite risk.
IBM apparently already forces various compiler vendors to "prove" they
don't use their register-coloring algorithm (based on things I've read
in this newsgroup) by showing them the internals of their compiler code!
I'm writing a compiler!  Put 2+2 together -- even though I'm distributing
the compiler for free (via the FSF), and even though the back end is
written by other FSF people in this case, what would prevent IBM from
hauling me into court if I just used my compiler to compile programs for
other people and was successful at it, much less if I shipped my compiler?

   However, if you mean that you should be able to MANUFACTURE copies
   of that work, well, the answer is that the design is NOT YOUR
   PROPERTY when someone else has taken the trouble to register the
   invention and make it public.

That is the answer to which I object.  You are saying an idea I have is not
my property just because someone else had it first.  I am saying I can
agree with this only insofar as it applies the way it has for a couple
hundred years -- to matter-transforming machinery, but NOT to algorithms,
because in the former case I can see an overall advantage to society for
allowing such usurping of personal intellectual property rights, whereas
in the latter case I can see no such advantage.

   It is IMPORTANT to recognize what is happening here.  YOU wish the
   RIGHT to keep an invention SECRET, AWAY FROM SOCIETY; someone else
   makes it PUBLIC in a patent application.

I already have that right, patents or no.  I'd like the opportunity to make
a plethora of software-patentable ideas available quickly and easily without
risking being sued because someone else happened to have thought of them
first.  At no point in my post did I ever say I wanted to change laws so
I could have the right to keep an invention secret.  How could I, since I
already have that right, and patents do not in any way change that fact?

   EVEN IF you do not believe in software patents, YOU CAN STILL PUBLISH
   YOUR IDEAS.  You can do that NOW, WITHOUT changes in the law.  And,
   if you do, then anyone coming after you cannot get a patent and cannot
   restrict your ability to use that work.  The fact that you do not do that
   shows that you have a different agenda.

"anyone coming after you" -- what about those that came before me, whose
patents I cannot read and understand and in any case do not need to to think
up fairly obvious and easy-to-implement ideas in a moment?  The idea for
a lightbulb came fairly easily to its inventor, it was the implementation
that took a long time -- with software, the idea IS the implementation,
pretty much.

   >Anyway, patents are NOT economic tools to recover R&D
   >expenses, or at least not primarily so; patents are tools to
   >encourage publication of the fruits of R&D expenses.  Please keep
   >this important difference in mind.

   What do you do, just make this stuff up as you go along?  The obvious
   POINT of a monopoly is to recover economic value.  That is WHY a
   patent is a limited monopoly.

I've read the literature on patents.  You've already agreed before to what
I said in the paragraph -- that patents are tools to encourage publication.
They were not intended, from what I read, to help recover R&D expenses.
It was assumed that sales would take care of that, and it still is!  In
cases where sales expectations are not enough, various funding groups
often step in and do the funding, sometimes insisting the resulting ideas
be placed in the public domain or patented for no licensing fees for
free-access use.

   >There are far better mechanisms for encouraging R&D
   >itself -- look at Japan and at various U.S. government programs
   >that do this.  Some of them even insist that the resulting
   >techniques be placed in the public domain (i.e. not patented or
   >copyrighted) or assigned to the government!

   Ah, yes.  Consider cryptography, for example:  That research is
   secret.  WE THE PEOPLE pay for it, but WE do not get access to it. 
   I think we can afford to forego a great deal of such successful
   government research, thank you.

That is a separate issue, and a relatively minor one at that.  Funding
for generally available R&D probably hugely dwarfs funding for
cryptographic research.  It probably is even larger than funding for
all non-publicly-available research.  My aunt in Houston had in her
possession a huge document containing lots of wonderful technical
information, required for publication by the government because the
discoveries and inventions contained therein were funded by government
money.  Obviously then it is not necessary or part of patentability for
patents to help fund R&D.  Naturally, having an existing patent that brings
in large license fees can help fund future R&D, but that's true of many
things, and does not per se offer a justification for patents (for example,
dropping employee wages below minimum could do the same thing, but that
wouldn't make it desirable to make legal, would it?).  Again, the justification
for patents is that they offer an incentive to publish the details of a
process, something I argue is generally unnecessary as an incentive in the
software field and is far too expensive to support as a mechanism in that
field.

My original point was and is that funding R&D is generally seen as
beneficial, which is why various governments do it.  But that is
separate from the extent to which patent protection is available.
Patents try and ensure publication of techniques.  Publication of
techniques is not sorely needed in software, a new (<100-yr-old)
science, to anything like the extent it was needed back in the 1700s
and before in the sciences of those days, when so much was kept
secret.  There has never been any evidence I've heard of that indicates
an important software (algorithmic) discovery was kept secret and then
lost because software patents weren't available up until the 1980s.
So why do we need them?

I did read an article about medical patents.  The example given (apparently
a real one) was this: a medical company started development on a
medicine of some kind by funding four separate efforts, labeled A, B, C,
and D.  After the teams on each effort had come up with their solutions,
the company's top engineers performed studies to see which medicine was
most effective (fewest side effects, and so on), and B clearly won.  But the
story doesn't end there -- the company's top lawyers then performed their own
studies to see which medicine was most patentable, and D won.  So the
company shelved A, B, and C, and released D.  And this is in an area where
I'm not arguing for patent reform (though many are, of course) -- in
software, the decision would be much easier to make in an unfavorable way,
and this will begin to happen soon if it hasn't already -- software
companies will begin purposely bringing out various less-than-fully-
compatible and -useful products because their lawyers tell them they are more
patentable, as DEC did with its BI bus (apparently) years ago (though that
was not a software patent per se, from what I know of it).  Do you really
want this to happen with software?

   I WOULD like to see the research environment improve--we could
   sure use it!  In my opinion it would be helpful to concentrate on
   the applied results TO THE PEOPLE of any such research.  If
   research had to show results IN THE MARKETPLACE, research centers
   would be far different than they are today.

See above; though I'm not sure how this pertains to whether software
patents exist.

   Showing results IN THE MARKETPLACE is another thing that patents
   are about.  Patents generally do not return money to the inventor
   until they are ACTUALLY APPLIED in a product, manufactured, and
   then SOLD to someone who thinks the product is a good deal.

Yes, generally true.  But with software patents, many are applied
retroactively to get fees for products already released and successful
and that incorporate commonly understood ideas long before the patents
for them were made public, and this is done by companies that exist ONLY
to hold software patents and squeeze fees out of others, as in Refac
Technologies and Cadtrac (spelling and other errors likely).  Even
spreadsheets are under attack for using "natural order recalculation",
a patent nobody needed as a means of publication but somebody now has.

   >Innocent people getting convicted is something the law bends over
   >backwards to avoid in a case where there is a .001% chance that a
   >particular fellow is not a serial killer and is thereby released
   >on what is commonly called a technicality.

   It must be particularly CONVENIENT for you to be able to just MAKE
   THESE THINGS UP.  Where do you get these figures?  What is your
   reliable source?  I suspect the real value is FAR higher, but I
   don't have your sources . . . .

It is well documented that our legal system would rather let 10 guilty
people go free than risk convicting 1 innocent person (or figures fairly
similar to that).  That is a strength of the system we have, in my
opinion.

   >If we are so willing to avoid unjust conviction in a
   >heinous case, why do you suggest it is ok for someone to be
   >convicted or otherwise taken for all the money he or she has
   >simply because they wrote and distributed compress or X?

   Well, that depends upon whether they have been properly warned that
   they are making use of intellectual property which does not belong
   to them.  If so, and they continue, then I guess they made their
   own decision, and they should be willing to back it up in court.

We ALL are making use of intellectual property that does not "belong" to
us.  The originators of patent (and copyright) law made it plain that
certain things belonged to nobody and everybody: laws of nature; obvious
ideas; typefaces (fonts); and so on.  Laws of nature has always included
algorithms and software, and the PTO never had any difficulty making
the distinction, except that until the 1980s they took it to mean that
no idea that even included a law of nature (algorithm) done in software
as a critical element couldn't be patented, then they went too far in the
other direction.

   >Fine, but with software, I come up with 20 new patentable ideas
   >for every 3 hours of typical programming, based on what the PTO
   >has been awarding as patents.

   Wo, SPACE CITY!  Again, if you have no idea, JUST MAKE SOMETHING
   UP!  I guess you reserve the right to just make up whatever values
   you want, as long as they benefit your argument.  Not good form.

Based on what I've seen patents awarded on in software, I'm being
conservative with this estimate.

   I would like to see YOU, or ANYONE, actually come up with, and
   SUCCEED IN PATENTING 20 ideas from 3 hours work.

If I had the tens of thousands of dollars and significant time, I or
any decent programmer could do it.  Do you really think it takes time
to come up with "one-second-pause +++ one-second-pause"?  Backing store
as an idea?  Include portions of text files?  Use of canonical byte
ordering?  Given that these already are patents, it would amaze me if
anyone could manage to spend 3 hours writing a useful new program WITHOUT
coming up with 20 or more new patentable ideas.  The point is, taking
the trouble to figure out which ideas should actually be patented and then
patenting them takes far more time and money than the 3 hours did.

   You must have a pretty good opinion of yourself to imagine that
   NOBODY else has either OPENLY DESCRIBED or FILED FOR PATENT on any
   trivial little thing that you come up with.  Others may not share
   that opinion, however.  Obtaining even ONE patent may be just a
   bit trickier than you think . . . .

Obtaining one patent is so tricky, I intend never to try it unless I come
up with a non-software idea that is worth patenting (because then I'd
presumably invest significant time and effort implementing it, and I'd
want patent protection for the process that resulted).

   >In any instance where it can be shown that a patented-style
   >publication is more informative and useful and build-upon-able
   >( :-) than releasing the source code or even the executable image
   >and using a disassembler, I'll agree that, in that instance,
   >patentability might well be preferable.

   A patent is a CONTRACT, a DEAL between inventor and society.  If
   the SOLE aim of a patent were to benefit society, then there would
   be no monopoly awarded.

The sole aim of patents IS to benefit society, as we all know from reading
about patents.  To accomplish the aim, monopolies are awarded.

   Alas, such an approach ignores human nature.  If there were no
   reward for disclosure, then only trivial items would be disclosed. 
   It is one of the little ironies of market economics that deals
   which are not profitable for BOTH sides simply do not happen.

What approach?  Other than that, I agree with the above para.  Note that
pretty much only trivial items have been disclosed via software patents
even though there IS reward for disclosure.

   Since your background is in "freeware," it is NO WONDER that you
   think all new ideas are trivial!  Economically speaking, that is
   what we would expect.

Did I say my background is in "freeware"?  I've been a professional
systems programmer for around 15 years, only the last two of which I've
been doing freeware.  And I DON'T think all new ideas are trivial, I've
had enough to know better -- I just think it is effectively impossible
to regulate them in a field where the distance between conceiving an
idea and implementing it is already so small and is growing smaller
all the time.  Used to take me 20 minutes to type in assembler code
to convert a binary number to decimal via a recursive algorithm on an
old teletype; now if I needed to do that, it'd take me 2 minutes or less
in C.  That was not an original idea of mine, but it is representative
of the kinds of ideas programmers put into practice every day.  Much of
"creative" programming (writing new programs, not just following other
peoples' specs) is just the creation of tons of nifty little ideas.
All of which may be patentable, based on what the PTO has been doing for
the last ten years.

   >I really get the impression you're not a computer programmer, or
   >at least not a serious (or efficient) one.  I can hardly keep up
   >with the tiny

   Well, having decimated my argument (HA!), now you wish to demean
   ME.

What's demeaning about suggesting you're not a computer programmer?  You've
been making far more demeaning statements about me in this post and I
haven't taken offense.  If you really are a computer programmer, could
you send me one of your recent programs (that you're willing to send) so
I can look through it and send you back a list of possibly patentable
(or already patented) ideas just to give you an example of how many ideas
you've created in a (hopefully simple, I don't have much free time) program?

   This used to be called an "ad hominem" attack, and is one of the
   fallacies of irrelevance.  It nicely demonstrates your competence
   to argue, as well as your disposition to do so fairly.  Real nice
   work.

It wasn't an attack, simply an observation.  I'm showing that, based on
my experience in programming combined with what the PTO considers
patentable, I can't possibly program at the same speed to which I am
accustomed if I have to worry about possible patents (mine or somebody
else's).  The fact that you seem to think this is possible suggests you
aren't well-acquainted with the art itself.  It's no more an attack than
if you were writing that football quarterbacks should be able to avoid
getting hit if we let the defense field 15 men instead of 11 and I said
"I get the impression you're not a quarterback, or at least haven't played
one in the NFL or a top-rank college team".

I notice you haven't outlined your background, by the way.  I am interested
in what you do and what drives you to suggest that software patents are
necessary -- do you have a personal stake in this as well?

   >If you compare the lifetimes of various popular products in
   >various different fields, I think you'll find that software dies
   >so fast, it isn't worth worrying about patents for it.

   This is an argument FOR patents.  Society has an interest in
   PROTECTING inventions so that they may (eventually) be re-used by
   all of society.  Simply RECOGNIZING such an invention, so that it
   may be catalogued for the future, may be an important event.

I argue that this isn't necessary for software.  TECO was an invention of
sorts; most people who need to recognize it do indeed; we hardly need to
reuse it; and it wasn't patented.  Yet it was a super-hot piece of
software in its day, especially MIT's version.  We have about as much
need in the software field to maintain documentation on a 15-year-old
product as we need to maintain it on Gutenberg's original printing
press.  When its historically interesting to somebody, the information
will hang around anyway (as in the case of the printing press and TECO),
otherwise we really don't need the info, not at the huge expense software
patents incur on development time.

   Note that it is NOT true that all inventions last 17 years; if an
   inventor fails to pay the "Maintenance Fees" (as might happen if
   the company goes out of business), the patent could lapse after
   only 4 years.

I'm not sure I see the point of this, although I don't disagree with it.

   >And, in any case, the whole point is that
   >it is nearly impossible to succeed at keeping any released
   >software's internal design secret, unless it is perceived as
   >generally useless and therefore not worth disassembling.

   This is ANOTHER argument FOR patents, for if an invention CANNOT be
   protected, it may not even be applied.

It will if they want sales!  And unpatentable inventions have been applied to
a vast extent for the last 50 years.  I know of very few cases of companies
holding up releasing software until they can do a patent search and start
applying for patents.

Correct that -- I know of NO such cases, although IBM's register-coloring
algorithm might be one (they might have waited on shipping a compiler
for all I know).  I do know that if I suggest the basic requirements for
patents in software to any of my friends that are in charge of software
companies -- that they must take the near-final copy of the product's
output assembler code, do a simultaneous search of the assembler code and
published software patents and see if their code inadvertently (or purposely)
contains patent violations; and then either publish their source and
assembler code openly, or re-search their code and figure out all the
possibly patentable things (which will number in the hundreds, thousands,
possibly tens of thousands, based on PTO actions to date) and then start
the patent process on each of them -- they will skip it.  For now.  Once
it becomes clear that not taking these steps exposes them to litigation
or paying retroactive patent fees over which they have no control (and
on ideas they may well have invented, perhaps even "first", but didn't
happen to document, as in the case of backing store to name one example),
they might well start doing that.  Needless to say, that'll slow software
development to a crawl, but it'll only happen as it becomes generally
recognized as necessary (so as to meet competitive deadlines).  Already
more and more such companies are seeing far enough ahead that they join
the LPF and/or oppose software patents.

   >If I knew less about computers than I
   >do,

   I guess you must be a real EXPERT!  Gee, how can anyone hope to
   oppose the opinion you have of you?

   Yet another instance of irrelevance.

It's not a matter of opinion.  I know quite a lot about computers and
writing programs for them.  That is not a subject open to debate.  There
is, of course, a vastly larger amount I DON'T know.  And what is a matter
of opinion is how much I know, or whether I know enough to really
understand the impact of software patents on the field.  But I make
no claims to the latter two items, just that "I know a lot about computers"
and "based on what I know, I think software patents are a really bad
idea".  I look forward to having a reasoned discussion with someone who
clearly knows more than I do about software and yet sees software patents
as vitally important for the field (vs., say, their own wallet).

The thing I was saying was that if I knew LESS than I know -- if I didn't
know what it was like writing reasonably complex software by myself and
fairly quickly, what it was like coming up with new ideas and being able
to basically instantly implement them, and what it was like to hear the
PTO awarding patents on things a lot like what I dream up while taking a
shower, then indeed I could be very worried about the impact of abolishing
software patents.  But it is my own experience that shows me the complete
impossibility of effectively implementing software patents and still maintain
a viable software industry in this country.  I hardly want to see the U.S.
lose its lead in one of the few areas it still (mostly) leads in -- software
-- just because some lawyers want to line their pockets, and too many
programmer and management types get excited about the fast-money
possibilities to care about the long-term impact.  We already went through
that in the auto and television industries, a lot of people got rich quick,
but now we've lost the leads as a result.

   >and thought indeed important secrets behind a popular program
   >could be lost forever, I'd agree that patents were important to
   >ensure publication, but that's not the case.  In fact it amazes me
   >how often totally useless or rarely sold software programs that
   >attempt to keep their internals secret via various techniques
   >(zapping debugger registers, copy protection, et cetera) are
   >nevertheless hacked at by intrepid types who just want to
   >know how they work -- and effectively published.  This happens
   >with games and such, and often they're more compdgx than "real"
   >application programs!

   The issue is NOT whether a few "intrepid types" know what is going
   on, the issue is whether new ideas are WIDELY PUBLISHED and MADE
   AVAILABLE to PROFESSORS and NEW GENERATIONS OF PROGRAMMER.  After
   the program dies, it hardly can be said to be available to
   analyze . . . .

This has been happening for decades without needing software patents, and
software patents show no promise of improving the state of the art, at
least not to me.

   Frankly, I think you enjoy the POWER of being SOMEONE WHO KNOWS,
   and, consequently, see no problem in keeping that knowledge SECRET
   from your "lessers."

No, I enjoy the power of being someone who can learn and apply techniques
I've learned freely to the benefit of others, and I know that software
patents will take that away from me.

   On the other hand, I am for OPEN PUBLICATION and KNOWLEDGE, with
   restrictions only on ACTUAL PRODUCTION USE of that knowledge. 

We already have open publication and knowledge without (prior to)
software patents.  Shipping a software program constitutes publication,
albeit typically in a fairly inscrutable language (object code).
Yet actual production use would appear to include far too many
restrictions, based on what I can gather of your definition of the
phrase.  Please elucidate that definition if you have a moment.

   Which of these attitudes do you think is more true to the goals of
   education?

I think mine is.  I expect full release of GNU Fortran along with source
code will be much better than patenting away lots of ideas in it and
not releasing the source, for example.  In fact, the presence of
software patents actually discourages the release of source code, to make
it harder for a patent holder to determine whether your product
is infringing on his/her patent.  If that is true, you appear to be
effectively arguing for a mechanism that discourages the release of
source code.  Is that correct?

   >And as I suggest, I cannot risk publicizing my ideas for much
   >better optimization and management of software programming because
   >it is very likely they will violate 100s of trivial patents even
   >if I tried to patent my

   And I suggest that the REAL reason that you do not PUBLISH your
   ideas is that you want to keep FULL CONTROL over them.  PUBLICATION
   OF YOUR IDEAS DOES NOT--AND CAN NOT--VIOLATE ANY PATENT!!!

Sure it can, and does, quite often.  compress was fully published, but
turned out to violate a patent, and now people are (or at least were --
I doubt I'm up-to-date on this) trying to abandon compress.  X-Windows
was freely and openly published, and turned out to violate AT&T's
backing-store patent.

   >Instead I must keep my ideas a personal
   >secret,

   "Personal secret."  That makes it pretty plain.  You want to have
   it, and you don't want anyone else to have it.

It was very clear in my original post that the opposite was true.  Why
do you intentionally distort what I said?  Abolish software patents, and
I can publish freely, else I am taking a risk I cannot afford to take
by publishing.  So I must keep it a secret or stop thinking about it, the
latter of which I can accomplish only by leaving the field entirely, since
it is so generally useful.

   Yes, sir, it sounds like a fine educational environment . . . .

Yours, with software patents, is a lousy one.  Mine, without them, is much
more like the environment of the '70s and before -- an excellent
environment for learning how to write software.  It's the environment I
learned in, for example, as I entered the field.

   >I'd like
   >to publish what I've learned, but I can't take the risk in this
   >society -- software patents basically mean I don't really own my
   >own ways of doing things.

   Oh, NOW I see!  You don't want to publish because then someone
   might figure out that you intend to use THEIR intellectual
   property.  Pirates rarely do like to draw attention to themselves!

Why do you call me a pirate?  If I have an idea and I implement it, I don't
want to have my life-savings taken away from me just because somebody else
happened to patent it first.  I've had some of these ideas kicking around
in my brain for 10-15 years without writing anything down because it was
clear, based on existing PTO policies at the time, that it wasn't necessary.
Now it's too late -- and I can't afford a patent search on all the possibly
patentable ideas I have, nor can I honestly say where I got them all from
(maybe most of them are "original", but what does that term mean in the
context of ideas and a society where people learn by observing?).

   However, on another level, if you REALLY think that software ideas
   are SO TRIVIAL, why don't you also assume that you have the ability
   to GET AROUND any trivial patent?  Generally that assumption would
   be correct.  Certainly, hardware designers have had to deal with
   this on a continuing basis, because THEY EXPECT to honor patents.
   So what are YOU whining about?

It's the expense of having to get around trivial patents.  The expense of
not being able to have a communications server recognize <pause>+++<pause>
because it's patented, and therefore not be compatible with 99% of the
modems out there.  The expense of not using a natural and obvious
optimization, backing store, in a free windows system.

And the expense of defending yourself in court against retroactive license
fees on a possibly freely distributed and copyable product.

   "Patents are trivial" but "I'm afraid I might not be able to get
   around them" just does not compute.

I know I can get around them when they're pointed out to me, but it is
too expensive and frankly way too discouraging to constantly have to
worry about this.

   >   Moreover, virtually any electronic device which produces
   >   information could be reduced to Boolean instructions which, in
   >   principle, could be performed mentally.  This certainly includes
   >   computer CPU's themselves.  Are computer CPU's NOT patentable
   >   hardware because a human could, IN PRINCIPLE (at a ten-millionth
   >   the speed and an error rate billions of times higher), perform
   >   "the same" operations?  I think most people would say that a
   >   dividing line has been crossed here.

   >Uh, if you can, via thought and abstractable communication alone,
   >scan a keyboard, drive a display or printer, interpret and
   >generate modem signals, store magnetic data on floppies,
   >read/write/refresh dynamic RAM, then I very much would like to
   >meet you in person!!  You could replace a few expensive
   >workstations I have around here....

   Uh, try again.  Maybe if you were able to follow this, you might
   have seen that I was forming an analogy between your ignorant
   defence of software and hardware devices themselves.

   OBVIOUSLY, we can perform ALL of the actions of a HARDWARE CPU,
   IN OUR HEAD, THE SAME AS WE CAN FOR SOFTWARE.  I claim that this is
   a good analogy, and that,in BOTH cases, the fact that we CAN do such
   instructions, one-by-one, DOES NOT mean that WE are "prior art" to
   CPU's, OR software.  The innovation consists of doing it accurately
   and fast, and software is just a way to tell hardware to do this.
   When software operates, it is an extension of the hardware, and its
   sole purpose is to produce a physical hardware effect.

I'm arguing that software as in "software patent" is not an extension of
the hardware but of the human thought-process.  With no hardware assist,
I can do register coloring as I compile a program.  But when the
purpose of a patented process is to drive a display monitor, a hard-disk
drive, and so on, I cannot replace that, so it is not a software patent.

The PTO has never had any difficulty in making the distinction you claim
is difficult or impossible to make.  They just went from saying "no patent
containing software is valid" to "ok, a patent containing software is
valid" to "all software patents are valid", but from what I can tell from
my research, they've never had much difficulty making the distinction,
where necessary, between a software item and other items.

   >the transformation of the passage of time into information on what
   >time it is is entirely algorithmic (if you allow that time itself
   >is of an algorithmic nature, in that it is a law of nature), and
   >thus not patentable -- until recently.

   Again, since it is so DIFFICULT to understand, digital hardware IS
   patentable.  And there is NO aspect of digital hardware which
   cannot be emulated in the mind.  So the ability to get the same
   results in the mind as in a patented device is NOT new.

It is new to me.  I use digital hardware (say an MC68030 chip) to drive
a monochrome display, access fast RAM, access a hard disk, access two
serial ports, and so on.  It so happens that this configuration also can
reliably add two and two and make four, but I can do that in my head.
I can't drive the display, printer, hard drive, etc in my head.  That
is the distinction.  Register coloring requires no hardware to happen,
same with backing store and various other software patents -- it can
happen entirely in thought.  On the other hand, the process of making
a paper clip can happen in thought, but the result of thinking about it
does NOT result in a piece of metal that will hold pieces of paper together
in a convenient way.  So it is not unpatentable because it can be thought
about.  Obviously ALL patents can be thought about, or they wouldn't be
patents, would they?

To make a useful (I hope) sample distinction, a Turing machine is not
patentable, while a particular implementation of one might be.  A Turing
machine is just an algorithm, basically.  A physical machine that
implements one is patentable to the extent it does physical ("matter-
transforming") things, but the Turing nature of it, while it should be
allowed (and didn't used to be) by the PTO, should not in and of itself
be patentable (which it now is, if there were prior art).

   Software is just a way to make a hardware device operate in a way
   which emulates some OTHER hardware device, made ONLY in hardware,
   WITHOUT software, which does the exact same thing.

The fact that it can be done entirely in software means it should not
be patentable, that it is in the realm of algorithms or "laws of nature",
and the PTO probably would agree with this latter aspect, insofar as they
are aware of the distinction you seem unwilling or unable to make, although
they currently allow the patenting of algorithms.

   Why should a hardware-only machine be patentable, but EXACTLY THE
   SAME MACHINE, implemented with software and a CPU, NOT be
   patentable?

Because the hardware-only machine might transform matter while the software/
CPU combination might just transform representations of matter.  Otherwise,
if a software/CPU combination can do the same thing, it should not be
liable for patent-infringement lawsuits just because the hardware-only
machine was patentable.  Or are you arguing that calculator programs like
bc and dc should be liable for patent-infringment suits from companies that
build hand-held calculators?  Or that makers of hand-held calculators should
be liable for patent-infringment lawsuits by the (now defunct) makers of
slide rules?

   WHAT DOES A HARDWARE PATENT MEAN, if a hardware device can be
   emulated in software with impunity?

Probably next to nothing, if the software solution is as cost-effective
as the hardware solution.  And I have no problem with this.  It is why
computers have been able to so rapidly grow in usefulness -- the replacement
of outmoded machinery with more efficient software that does the same thing
hasn't left makers of the software liable for patent infringement.

This means if I manufacture, say, a neural net chip (hardware) and patent
it, I can't sue someone for implementing it in software.  Sounds harsh, but
that scenario is desirable -- what I should hold a patent on is the
hardware solution (transformation of matter, in this case electrical
signals to electrical signals) but not the algorithm (weighted networks,
even assuming I invented it), so if someone reproduces it in software and
gets it to run as fast, fine, they've beaten me, but I should be able to
make an even faster hardware solution by then.  However, current PTO
practices would allow me to patent the weighted-sum algorithm (despite the
fact that it not only is an algorithm but, apparently, a law of nature --
our own neurons, for example) and sue anyone that used it even
inadvertently in software (in other words, even a transformation of it used,
for example, in statistical research, I'd own).  If someone made a
replacement chip that did the same signal-to-signal transformation but
used a high-speed parallel digital chip with software on board, again I
shouldn't be able to sue, because they've just done a better mousetrap and
I shouldn't own the interface (though, again, current practice differs,
apparently, in copyright law).

On the other hand, if I manufacture a washing machine and patent it, I have
little fear that someone will make a computer program that'll replace it!
Maybe one that'll replace the hardware timing control in my machine, but
no way they can write a program that just by running on an IBM PC will
manage to actually wash dishes.  Current PTO practices might allow me to
get a patent on the idea of using jets of water to wash dishes, however,
thus disallowing a multitude of useful products, and this is what I object
to -- even though everyone "knows" high-speed water is good at washing
dishes, so much so that nobody might actually ever publish that knowledge,
the PTO would (in this scenario) let me obtain a patent for it just because
I made a machine that happened to use that knowledge as a critical element.

   >   I doubt that there can ever be a consistent legal definition
   >   which would separate "software" from "hardware."  "Software"
   >   is just the term we use to describe the practice of forming
   >   the instructions TO BE USED BY HARDWARE.  Without hardware,
   >   software ALREADY does not infringe patents; patents are about
   >   the USE of software TO PERFORM HARDWARE ACTIONS.  It is these
   >   actions which are patented.

   >Then nobody who writes a program violates a patent, it is only
   >those who actually run the program?  Hardly.

   Of course not.  You missed a little step between WRITING and
   USING:  simply DISTRIBUTING a program containing patented mechanisms
   violates the patent.  Subsequent USE would ALSO violate the patent,
   although a court might require notification of deceived users
   before they would be liable.  The distributor would probably be
   liable for such notification, but this would probably not be the
   extent of his (or her) liability.

You said "patents are about the USE of software TO PERFORM HARDWARE
ACTIONS.  It is these actions which are patented" but now you say
"simply DISTRIBUTING a program containing patented mechanisms violates
the patent."  How do you reconcile this with "Without hardware, software
ALREADY does not infringe patents"?  What do you mean "without hardware"?
Code not typed in?  Code not compiled?  Code not running?  I'm confused
about what you consider to be a patent violation.

   >But leave users holding
   >the bag for actually running the program to get work done?  Nobody
   >who wants software patents will agree with you on this one!

   They WOULD side with me, for I am not on "that" side.  YOU put me
   there so you could argue against it.  

You seemed to be saying that I could write a program that contains a
patented algorithm and distribute it as long as I didn't actually run
it myself.  I'm afraid you haven't sufficiently clarified this for me.

Maybe what you mean is this: I can write all the code I want, but if I
either distribute it, or use it myself for profit (as in using a compiler
I write to compile programs for others but without distributing the
compiler itself), I might be liable for patent infringement.  This would
be consistent with patent law as applied to other fields.  If this is what
you mean, then you are proposing that it is ok for me to write free
software as long as I don't distribute it or use it, unless I'm willing
to incur the expense of a patent search or the risk of not doing one,
correct?

   SELLING programs containing patented mechanisms violates the sales
   and distribution monopoly.  Even GIVING AWAY such programs violates
   the distribution monopoly, unless there is a SUBSTANTIAL reason to
   believe that programs will only be used for researching those ideas.
   Just SAYING "This program should only use this program for research"
   will probably cut little ice if the program is obviously set up for
   production use.  Production use in other research activities is still
   use.

What is the difference between a program set up for research vs. production
use on, say, a UNIX system?

And, again, do you really want most or all distribution of free software
to halt, given that it very likely violates software patents but the writers
of it cannot afford patent searches to make sure?

   >With traditional guidelines for a patent, it was always
   >sufficiently clear who was doing the violating -- with
   >matter-transformation processes, somebody has to build and/or
   >run a machine to transform the matter.  With software, the program
   >exists even when no machine exists to run it,

   If a patented machine was sold without license, the manufacturer
   and/or seller would be responsible.  Once the machine was in place,
   if the user knew the machine was unpatented, continued use might
   well give the user some responsibility.  JUST LIKE SOFTWARE. 
   NOTHING HAS CHANGED.

Except that the process of designing and building a machine to carry out
a patentable process is inherently far more difficult than building a piece
of software to carry out an equivalently complex patentable software process;
and that, unlike hardware, software is easily copied and incorporated into
new "devices", and this is a frequent "use" of the software.

By the way, I don't think that users of patent-violating hardware devices
are ever liable.  A lawyer responded to a post of mine about a year or so
ago.  I had been wondering whether I would be in any way liable or have to
worry if I bought, say, a new kind of hammer, and it turned out it violated
a patent.  He said the manufacturer and distributor of the hammer could
be sued for retroactive licensing fees, but I didn't have to worry -- not
only would a house I built with the hammer be safe, but I wouldn't have to
return or pay an additional fee for the hammer.

Problem is, with software, I could do the equivalent of: use the hammer;
make hundreds of copies of the hammer so I could use a whole bunch of them
at once; incorporate the hammer in my house so the toolroom and basement
would have their own built-in hammers right where I need them; then make
copies of my house for friends to use, or even sell it.  I could do all
of this very easily compared to the reality of the hardware hammer, and in
the software field this all would be considered "use" of the hammer.  (And
of course the distributor of the hammer would have no problems with all this
if it were, say, the Free Hammer Foundation.)

So if GNU C (gcc) incorporates a patent-violating algorithm, not only is
the FSF liable for distributing it, but so is anyone who makes copies of it
for friends, distributes it in turn, OR anyone who recompiles it (with itself)
and/or pulls out the source code for sections of it (including the algorithm
in question) to use in their own product.  Just as people who run compress
are considered liable for violating the LZW patent, from what I've been able
to determine.

Again, software patents ultimately means no way for free software to
continue, at least not without massive support by the proprietary software
industry (not very likely).

   >   Preventing the application of patents to software would imply
   >   that anyone   who wished to get around a patent on (for example)
   >   a new chip (hardware) could try to reproduce the design in
   >   software (i.e., MICROCODE, in a fast single-chip processor) and
   >   thus avoid the hardware patent.  In this way, eliminating software
   >   patents must inherently damage protection for hardware logic
   >   designs (including computer hardware).  Such protection has been a
   >   basis for the development of the computer industry, and
   >   elimination of such protection could have extremely serious
   >   consequences far beyond you not being able to use your favorite
   >   technique in your compiler.

   >Wrong, since all computer hardware innovation up to this point has
   >not required software patents per se to be protected!  Again, software
   >is not, by itself, going to be able to drive a display or scan a
   >keyboard; the chip design that does these things is indeed hardware
   >and thus patentable, inclusive of any software that is part of the
   >patented process.  But the

   Wrong, since this is NOT the LPF position.  Your point of departure
   is the phrase "inclusive of any software."  (How tough could it be
   for you to do this research?)  Shall I quote the LPF position paper
   for you?

I've read it; by "inclusive of any software" I meant that the presence of
software in a patent should not invalidate the patent, but as I clearly
said elsewhere in my post, the software itself is not patented or
patentable, so someone can use that part of the patented process and not
infringe.

	 "We recommend the passage of a law to exclude software from the
      domain of patents.  That is to say that, no matter what patents
      might exist, they would not cover implementations in software;
      only implementations in the form of hard-to-design hardware would
      be covered." -- Against Software Patents, p. 10

I've read this many times before, thank you, and wholeheartedly agree
with it.

   We now have FAST processors on a chip.  This is new.

   But IF the fast processor emulates another chip with SOFTWARE (for
   example, microcode), the LPF says that this would AVOID any HARDWARE
   patents protecting the other chip.

Yes, and I agree with this, and I believe I made this sufficiently clear.
However, if the chip used patented hardware stuff on it (which it might
do, to drive RAM, displays, et cetera), then it wouldn't avoid that.

   Although the worth of many chips is in their speed, other chips may
   implement complex protocols, such as normally-slow I/O operations. 
   Such HARDWARE designs would no longer have effective patent
   protection because another company could copy the design (WITHOUT
   paying for research into error-rates, protocol formats, etc.) in a
   fast microcode-driven single-chip processor.

Yes, they could, but that's a fairly small part of the whole arena, and
by the time fast-enough general-purpose processors become available that
can replace a special-purpose chip in a cost-effective way (usually a
few years), the company that owns the hardware patent can easily have
used their special knowledge to come out with the software solution
themselves.  Remember that a lot of hardware design happened before the
1980s when there was no effective patent protection against replacement
with a software emulator, but this didn't stop innovation.

The problem I see that might happen is that a company that is coming out
with a special-purpose hardware chip to do something that conceptually
could be done in software might be unwilling to publish (patent) their
chip because it'd make it that much easier for someone else later on to
use faster general-purpose chips and emulate the special-purpose one in
software.  So they'd keep the internals of the chip as a secret, instead.
That might occasionally happen, but it's preferrable to the current system
of allowing patents on anything algorithmic.  And, in any case, where the
chip manages an open interface (as in the Hayes patent), it is hardly
difficult to reverse-engineer the "secret" process.  But as I've said
repeatedly, I recognize such cases can indeed happen, I don't pretend
abolishing software patents will make a perfect world -- just that it'll
make a much better one than the direction in which we're headed with
software patents.

   >It is amazing to me that you can have written an article for a
   >publication (admittedly one I've never heard of), presumably done
   >all the necessary research,

   Someone who has absolutely no compunction about MAKING UP statistics
   for his argument, and who has no idea what the LPF position is on
   software patents, has some damn nerve criticizing MY research in an
   article he has never read.

I don't see why you think I don't know what the LPF position is; and
no I have no compunction about estimating statistics in a field I know
fairly well.  Heck, I've spent almost as much time arguing the usefulness
of software patents with you as I'd have to spend worrying about them in
new products if we don't cancel them, and that's pretty close to one of my
estimates -- and here I'm not worried about a lawsuit as I will be with
software patents!

   >and yet state that "[software patents have] been a
   >basis for the development of the computer industry"

   Jeez, maybe they had better teach you to read again:

   >   Preventing the application of patents to software would imply
   >   that anyone who wished to get around a patent on (for example) a
   >   new chip (hardware) could try to reproduce the design in software
   >   (i.e., MICROCODE, in a fast single-chip processor) and thus avoid
   >   the hardware patent.  In this way, eliminating software patents must
   >   inherently damage protection for hardware logic designs (including
   >   computer hardware).  Such protection has been a basis for the
   >   development of the computer industry, and elimination of such
   >   protection could have extremely serious consequences far
   >   beyond you not being able to use your favorite technique in your
   >   compiler.

   You will note the words "and thus avoid the hardware patent."

   Then, "Such protection has been the basis for the development of
   the computer industry . . . ."

   Clearly, I refer to HARDWARE PATENTS as ONE basis for the computer
   industry, and I stand by that.

But disallowing software patents won't disallow hardware patents; if the
PTO went with the LPF position, it'd be a return to the pre-1980s PTO
position (but not quite as extreme, i.e. more favorable to those seeking
patents on hardware incorporating software elements), so it is hard to
understand why you say the post-1980 PTO position is one basis for the
computer industry.

My point is that before the PTO started allowing software patents in the
'80s, development of the computer industry, especially items that could
(and would) be later emulated in software, was already rampant.  Talk to
Intel about desirability of patents on microprocessor designs -- and that
one isn't even a software patent!

   >Anyway, if you are willing to send me a copy of your article,

   Oh, yeah, I really look forward to that.  Do your own "research."

That is a very uncooperative attitude.  You name a magazine I've never
heard of, refuse to send me an article you earlier assailed me for not
reading, and don't even send me the address of the editor of the magazine
in question so I can obtain a reprint!

   We can ALREADY see just how much "research" you do . . . .

Speak for yourself -- or are there more than one of you reading this (I
refer to "We")?  Meanwhile, if you really care about my research, why do
you respond to my polite request for a copy of your article or information
on how to obtain it with such venom?  What have I said to earn that?  Is
this how you normally conduct yourself in an open, public debate?  Hardly
typical of a native Texan, especially one living in the usually laid-back
Austin area!  (Next thing you'll be considering my assuming you are possibly
a native Texan or living in the Austin area an insult! :-)
--

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