Newsgroups: comp.patents
Path: sparky!uunet!usc!rpi!uwm.edu!linac!mp.cs.niu.edu!uxa.ecn.bgu.edu!mflll
From: rc...@netcom.netcom.com (Robert Cain)
Subject: Personal Use
Status: RO
Message-ID: <1992Jan19.050830.28936rcain@netcom.COM>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Sender: mf...@uxa.ecn.bgu.edu (Dr. Laurence Leff)
Organization: Netcom - Online Communication Services (408 241-9760 guest)
Date: Sun, 19 Jan 1992 05:08:30 GMT
Approved: pate...@cs.su.oz.au
Is it forbidden by patent law to build a patented device in one's back
yard for ones own use? If the answer (by chapter and verse) is yes,
has there ever been prosecution of such an endeavor?
--
Bob Cain rc...@netcom.com 408-358-2007
"Systems should be described as simply as possible, but no simpler."
A. Einstein
[ Note from Temporary Moderator
> There was a law magazine article on that question. Some ancient
> precedents at the court of appeals level do say that one can build
> a patented device strictly to satisfy curiosity, etc. However, it
> is not clear whether they would hold up, or whether they would be
> considered pursuasive in other jurisdictions.
>
> Note that this concerns personal use. There was a case where a generic
> drug manufacturer started doing the bioavailability tests on some
> drug before the patent ran out. That was ruled to be an infringing
> "use" even though the drug was not used to cure any disease.
> In case you don't know, the U.S. is divided into court of appeals
> districts. Federal District Court cases get appealed to the court
> of appeals for the district. A court of appeals precedent is only
> valid within that district. They are considered "persuasive" in other
> districts. It happens fairly frequently that the same issue gets
> decided different ways in different court of appeals districts.
>
> In that event, the Supreme court will often take the next case involving
> that issue so that their can be a uniform rule.
I will try to look up the law review article and the relevant cases.
This matter is of some personal interest.
End of Note from Temporary Moderator]
Newsgroups: comp.patents
Path: sparky!uunet!wupost!uwm.edu!linac!mp.cs.niu.edu!uxa.ecn.bgu.edu!mflll
From: brns...@KRAMDEN.ACF.NYU.EDU (Dan Bernstein)
Subject: Re: Personal Use
Status: R
Message-ID: <9201202057.AA07621@KRAMDEN.ACF.NYU.EDU>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Sender: mf...@uxa.ecn.bgu.edu (Dr. Laurence Leff)
Organization: IR
Date: Mon, 20 Jan 1992 20:57:41 GMT
Approved: pate...@cs.su.oz.au
In article <1992Jan19.050830.28936rc...@netcom.COM> you write:
> Is it forbidden by patent law to build a patented device in one's back
> yard for ones own use?
Yes. Special exceptions aside, if you make, use, or sell a patented
invention, you are infringing the patent. The most important broad class
of exceptions is for what's usually called experimental use. I advise
you to take the following paragraph with a grain of salt:
> > There was a law magazine article on that question. Some ancient
> > precedents at the court of appeals level do say that one can build
> > a patented device strictly to satisfy curiosity, etc. However, it
> > is not clear whether they would hold up, or whether they would be
> > considered pursuasive in other jurisdictions.
Here's some real case law. It's neither ancient nor forgotten. It forms
a rather strong precedent which, by comity, other courts must follow.
Should I note that one of the defendants was IBM? (Quotes courtesy of
West's USCA.)
``Experimental use does not infringe a patent.'' Chesterfield v. U.S.,
1958, 159 F.Supp. 371, 141 Ct.Cl. 838.
``An experimental use for philosophical or amusement purposes is not an
`infringement', but an experiment made commercially may be an
infringement.'' Northill Co. v. Danforth, D.C.Cal. 1943, 51 F.Supp. 928.
``Experimental use is not public use if it is conducted in good faith
for purposes of testing invention and for no other purpose not naturally
incidental to experimental purpose.'' Norfin, Inc. v. International
Business Mach. Corp., D.C.Colo. 1978, 453 F.Supp. 1072, affirmed 625
F.2d 357.
``Neither use of patented machine for experiments for sole purpose of
gratifying philosophical taste or curiosity or for instruction and
amusement nor construction of infringing device purely for experimental
purposes constitute `actionable infringement'.'' Kaz Mfg. Co. v.
Chesebrough-Pond's Inc., D.C.N.Y. 1962, 211 F.Supp. 815.
``Use of patented machine without authority from patentee for
experiments for sole purpose of gratifying philosophical taste or
curiosity or for instruction and amusement does not constitute
infringing use.'' Ruth v. Stearns-Roger Mfg. Co., D.C.Colo. 1935, 13
F.Supp. 697.
``Experimental testing by defendant of infringing apparatus for a brief
period before going into commercial production, none of products having
been commercially sold, did not constitute infringement.'' Akro Agate
Co. v. Master Marble Co., D.C.W.Va. 1937, 18 F.Supp 305.
The operative phrase (at least in California, New York, and Colorado :-))
is ``for the sole purpose of gratifying philosophical taste or
curiosity, or for instruction and amusement.'' I find it extremely
difficult to believe that any court will ever rule such use to be
infringement. Mod?
On the other hand, be aware that (e.g.) making your own RSA keys for
day-to-day communication, not just to satisfy philosophical curiosity,
is almost certainly infringement. ``Infringer of patent cannot escape on
ground of experimental use where it used machines to operate upon
customers' products in the ordinary course of business.'' Spour, Waldron
& Co. v. Bauer Bros. Co., D.C.Ohio 1938, 26 F.Supp. 162.
---Dan
[Note from Temporary Moderator
These appear, from the citations, to be District Court cases and thus
have no precedential value outside the immediate district (usually
part of a state).
I will look up my references and followup.
THANKS FOR AN EXCELLENT POST!!!!
End of Note from Temporary Moderator]
Newsgroups: comp.patents
Path: sparky!uunet!wupost!psuvax1!uxa.ecn.bgu.edu!mflll
From: ji...@microsoft.com
Subject: Re: Personal Use
Status: R
Message-ID: <1992Jan29.030422.4228@microsoft.com>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Lines: 28
Sender: mf...@uxa.ecn.bgu.edu (Dr. Laurence Leff)
Organization: Microsoft Corp.
References: <9201202057.AA07621@KRAMDEN.ACF.NYU.EDU>
Date: Wed, 29 Jan 1992 01:06:33 GMT
Approved: pate...@cs.su.oz.au
In article <9201202057.AA07...@KRAMDEN.ACF.NYU.EDU> brns...@KRAMDEN.ACF.NYU.EDU
(Dan Bernstein) writes:
|``Neither use of patented machine for experiments for sole purpose of
|gratifying philosophical taste or curiosity or for instruction and
|amusement nor construction of infringing device purely for experimental
|purposes constitute `actionable infringement'.'' Kaz Mfg. Co. v.
|Chesebrough-Pond's Inc., D.C.N.Y. 1962, 211 F.Supp. 815.
I wonder aloud about the applicability of these exemptions for people
who write software and distribute it is some form of freeware, copycenter,
copyleft, or other category of software. Say someone chooses to
distribute software under something like the following terms:
"Copyright 1993 Cooperative Computer Club
This software is provided solely for experimental, educational, or
amusement purposes. Any use of this software for commercial purposes,
distribution for profit, or for any other purposes besides the above
expressed purposes is strictly prohibited and violates this licensing
agreement blah blah deblah blah Violators of this licensing agreement
may be subject to copyright or patent infringement proceedings blah blah blah"
You know what I mean. Would legitimate attempts to distribute such software,
under such terms say between college professors, pass muster? Or would such
software be considered enticement to infringe? What do people think?
[Seems to me that legitimate not-for-profit software *should* pass muster
under the above exemptions -- not saying that it *would*]
Newsgroups: comp.patents
Path: sparky!uunet!zaphod.mps.ohio-state.edu!uwm.edu!psuvax1!uxa.ecn.bgu.edu!
mflll
From: n...@Think.COM
Subject: Re: Personal Use
Status: R
Message-ID: <kof5hjINNmko@early-bird.think.com>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Lines: 30
Sender: mf...@uxa.ecn.bgu.edu (Dr. Laurence Leff)
Nntp-Posting-Host: godot.think.com
Organization: Thinking Machines Corporation, Cambridge MA, USA
References: <9201202057.AA07621@KRAMDEN.ACF.NYU.EDU>
<1992Jan29.030422.4228@microsoft.com>
Date: Thu, 30 Jan 1992 06:07:51 GMT
Approved: pate...@cs.su.oz.au
In article <1992Jan29.030422.4...@microsoft.com> ji...@microsoft.com writes:
>In article <9201202057.AA07...@KRAMDEN.ACF.NYU.EDU> brns...@KRAMDEN.ACF.NYU.EDU
(Dan Bernstein) writes:
>|``Neither use of patented machine for experiments for sole purpose of
>|gratifying philosophical taste or curiosity or for instruction and
>|amusement nor construction of infringing device purely for experimental
>|purposes constitute `actionable infringement'.'' Kaz Mfg. Co. v.
>|Chesebrough-Pond's Inc., D.C.N.Y. 1962, 211 F.Supp. 815.
>
>I wonder aloud about the applicability of these exemptions for people
>who write software and distribute it is some form of freeware, copycenter,
>copyleft, or other category of software.
Note that the quoted text says "use", not "distribution". The intent is
clearly to allow a person to read a patent, make an implementation for
personal use just to see how it works first hand ("gratifying ...
curiosity") or for a limited demonstration ("for instruction"). For
instance, before committing yourself to paying the license fee for a
device, you can do some experiments to see whether it's really appropriate
for your application; in the case of software, you might want to write a
version of the program in order see whether its performance meets your
requirements.
Once you start distributing the patented device, I think you're outside the
scope of the exceptions. I don't think restrictions on the use of the
software by the recipients matters in this case.
--
Barry Margolin, Thinking Machines Corp.
bar...@think.com
{uunet,harvard}!think!barmar
Path: sparky!uunet!news-server.csri.toronto.edu!bonnie.concordia.ca!
garrot.DMI.USherb.CA!uxa.ecn.bgu.edu!mflll
From: bur...@geech.gnu.ai.mit.edu (Craig Burley)
Newsgroups: comp.patents
Subject: Re: Personal Use
Message-ID: <BURLEY.92Jan30153241@geech.gnu.ai.mit.edu>
Date: 30 Jan 92 20:32:41 GMT
References: <9201202057.AA07621@KRAMDEN.ACF.NYU.EDU>
<1992Jan29.030422.4228@microsoft.com>
Sender: n...@ai.mit.edu
Organization: Free Software Foundation 545 Tech Square Cambridge, MA 02139
Lines: 134
Approved: pate...@cs.su.oz.au
Status: R
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
In article <kof5hjINN...@early-bird.think.com> n...@Think.COM writes:
Note that the quoted text says "use", not "distribution". The intent is
clearly to allow a person to read a patent, make an implementation for
personal use just to see how it works first hand ("gratifying ...
curiosity") or for a limited demonstration ("for instruction"). For
instance, before committing yourself to paying the license fee for a
device, you can do some experiments to see whether it's really appropriate
for your application; in the case of software, you might want to write a
version of the program in order see whether its performance meets your
requirements.
Once you start distributing the patented device, I think you're outside the
scope of the exceptions. I don't think restrictions on the use of the
software by the recipients matters in this case.
Aye, there's the rub! How can mere distribution, even in electronic form,
of essentially a _translation_ of a patented process from one language
(Legalistic English) into another language (say, C++) be considered the same
as "distributing the patented device"? If it can, then _any_ form of
translation from patent language to any other (including "plain English", as
in writing up a more readable description) must be considered potentially
infringing distribution. What would prevent me from writing an interpreter
or compiler for the patent "language" that read a patent and implemented its
software components, especially if I was able to achieve this task using
today's technology by restricting the scope of the patents I cared about?
Nothing, of course. What would prevent me distributing it? Again, no present
law could prevent that, any more than if I somehow made a robot that could
build any machine from a description (say, a non-software patent), though that
task clearly is far more difficult. Given that such a program was distributed,
what would prevent people from trivially getting this program to implement
their favorite algorithms by directing it to one or more software patents?
Only the threat that such _individual_ use _might_ be prosecuted.
If it can't, then _no_ distribution of software can ever be considered
infringing distribution, since no software, whether source, object, or
executable form, can truly be considered a _device_, but rather a
description (or, specifically, containing a description) of the patented
technique. Only the person who actually runs the program on a physical
machine (a general-purpose computer) could be said to be infringing the
patent. The person who wrote and distributed the program couldn't even
be liable for contributory infringement, any more than a person who
(legally) translates a recipe for a patented chemical process could be
considered liable for contributory infringement if a chemical factory used
the translation in violating the patent.
This is one reason I'm very concerned about software-implementation
patents. Unlike most other forms of patents, there seems to be no way
to distinguish between personal use, commercial use, and so on. And formal
language theory seems to suggest that a given idea, expressed in any
formal language, is the same, and that ideas and implementations are
ultimately indistinguishable. (But I'm no formal-language expert.) If
that's the case, it seems very unlikely that patenting of software
implementations is possible without in fact patenting ideas, and thereby
preventing people from even expressing those ideas in different forms.
I guess what I'm saying is that, unlike other forms of patents, software
patents seem to have the problem that the _expression_ of the patent (the
patent language, so to speak) is essentially equivalent to the
_implementation_ of the patented device (or _any_ device that accomplishes
the same thing). If the PTO decided to require that all software patents
be written in, say, Eiffel, to improve clarity, make patent searches easier,
and so on, which might be entirely reasonable (substitute your favorite
language, e.g. Haskell, ML, Prolog, if it helps :-), then how can one be
infringing on a patent by doing _exactly_ what the patent process promotes --
publishing the expression of the patent (i.e. copying the Eiffel code)?
And, as I pointed out earlier, I know of _no_ computer that actually
implements the operations it is claimed to implement in the ways we normally
expect (i.e. the ways we do as humans). The ADD operator doesn't add; it
accomplishes the same thing using, say, boolean XOR operations. The MULT
operator doesn't multiply. LOAD and STORE are examples that are basically
implemented as expected. But forget about divide, square root, floating-point
operations of all kinds -- I know of no computer in wide use that uses the
"canonical" (expected, as ordinary humans would expect it) implementations
for those operations.
Given that, even when a program is run on a machine, if the patent it is said
to be violating is written in terms of addition, subtraction, and
multiplication, and even though the source code (only; we can't say the object
code or executable code specifies "add" when it really only says "do whatever
is agreed by the architecture description when opcode 506 is seen") is
written in those same terms, how can the patent be violated when the
actual _implementation_ of most or all of the underlying operations is
quite different from the way a human would do it -- especially the human that
invented the patented process in the first place? For example, if the RSA
patent(s) are expressed in terms of multiplcation, exponentation, and division,
how can either source code "implementing" RSA algorithms truly be considered
infringing (when the code is just a published translation of the already
published patented algorithm) or machine code "running" RSA algorithms be
considered infringing (when the actual implementations of the algorithms are
_fundamentally different_, except in result, from that specified in the patent,
unless RSA patents are written specifying shifts, XORs, NANDs, and the like)?
Despite all the discussions about software-implementation patents pro or con,
I've seen _nothing_ that even _begins_ to answer this set of issues, except
perhaps the suggestion that any program source code that contains an
implementation of that patent is potentially infringing or co-infringing.
I totally object to this concept, if that's what permitting software-
implementation patents means, since that's a complete disregard of freedom
of speech _and_ the foundations of patent law (i.e. widespread dissemination
of patented techniques). Despite the term "implementation", published source
code _never_ actually implements any algorithm, it merely describes how it
might be implemented in a different language than the original patent.
I don't think my objections necessarily pertain to patents like the Hayes
"+++" patent (though I haven't read that one), though I'm curious why
someone can't get around that patent by pretending their modem implements
an entirely different character set, replace "+++" with "JJJ", and have
the character set map "J" to the same value that ASCII maps "+". But
despite the lunacy of permitting "+++" to stand up as a valid patent, I'm
more concerned about "pure" software patents -- patents entirely about
software, or mental, processes and implementations, for the above reasons.
Either they are fairly easily (and unknowingly) violated with today's
technologies, and trivially violated with tomorrow's, or they are, basically,
_never_ violated. In any case, they are useless and worth dispensing with
altogether, despite the apparent temporary ability for a few of these patents
to stand up in court.
By the way, I think there was someone out there who claimed that a software
implementation of, say, an automatic transmission, having all the elements
of the transmission but only in software (and thus in relation to, essentially,
a "virtual world" containing a road, an engine, a car, &c), should be
considered an infringement of a patent on a real automatic transmission.
Needless to say, I totally disagree with this notion, and in fact don't even
care to hear from anyone who thinks virtual implementations of real-world
things, where the virtual things cannot _possibly_ achieve the same real-world
results as the real-world things, should be considered to infringe on patents
on the real-world things. But someone very "pro" software patents might well
believe this kind of thing. If you really think you can reasonably defend this
sort of thing, go ahead and give it a try, I guess.
--
James Craig Burley, Software Craftsperson bur...@gnu.ai.mit.edu
Member of the League for Programming Freedom (LPF)
Newsgroups: comp.patents
Path: sparky!uunet!zaphod.mps.ohio-state.edu!uwm.edu!psuvax1!uxa.ecn.bgu.edu!mflll
From: mcgre...@hemlock.Atherton.COM (Scott McGregor)
Subject: Re: Personal Use
Status: R
Message-ID: <45295@athertn.Atherton.COM>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Lines: 67
Sender: n...@athertn.Atherton.COM
Organization: Atherton Technology -- Sunnyvale, CA
References: <BURLEY.92Jan30153241@geech.gnu.ai.mit.edu>
<9201202057.AA07621@KRAMDEN.ACF.NYU.EDU> <1992Jan29.030422.4228@microsoft.com>
Date: Fri, 7 Feb 1992 01:45:23 GMT
Approved: pate...@cs.su.oz.au
Craig Burley makes a number of interesting comments and indeed a very
interesting and tight argument about the inconsistency of software patents.
However, it seems to me to be based upon some assumptions about what a
patent is that are different from what the patent laws suggests, and
what the PTO seems to act on, and what inventors seeking patents expect.
Craig writes:
> I guess what I'm saying is that, unlike other forms of patents, software
> patents seem to have the problem that the _expression_ of the patent (the
> patent language, so to speak) is essentially equivalent to the
> _implementation_ of the patented device (or _any_ device that accomplishes
> the same thing).
The assumption that Craig seems to be building on is that patents cover
only devices, not processes. Unfortunately the aforementioned groups
seem to think that processes, and not simply the devices that perform
them are patentable material. Secondly, Craig seems to assume that
patent should cover what is in the patent application itself. However,
typical patent practice holds that the patent application is only a
description of process or device, not the actual working model. In the
case of software, it is acceptable to submit actual source code as part
of the description of how to build the device or process, but it is not
required, a high level flow chart might be a quite acceptable alternative.
Because Craig assumes that patents applies to devices not processes, he
ignores the fact that most of his arguments are not limited to software
patents alone, but to any process (i.e. a process described by set of
IDEAL parts, steps and procedures followed in a formal way). Most of
Craigs arguments would argue just as persuasively against any patent
process, even one for rubber curing. Indeed that might be viewed as one
explanation why the justices decided Diamond vs. Diehr the way they
did--the alternative might have undermined all process patents. It may
be that because Craig ignores the interpretation of a patent as applying
to processes (referring instead to hardware devices) that he overlooked
these other class of patents, or maybe he thinks they shouldn't allowed
to. I can't tell from his remarks. But since his part of his argument
seems based upon this point, a point not conceeded by those in the
patent infrastructure--he is at the risk of arguing past others who make
opposing assumptions.
Craig's second assumption, concerning expression of a thing vs. the
thing itself, can also lead to him arguing past others used to the view
that there is no "thing" in a patent, ONLY a description of that thing.
In that respect others would not see his point about translation, since
ALL patent text would have to be translated (implemented) to another
form to be operable. In such a view, it won't matter if that
implementation is in eiffel or C any more than it would matter whether
an implementation of a patented brake was in hardened steel or titanium.
I bring these points up not because Craig is right or wrong, but merely
to point out that software opponents sometimes fail to convince others
because they start out with assumptions not generally recognized by
those others. Without common starting points there is no ensuring that
all people will reason to the same conclusion. Indeed, I believe that
this is most of what is going on in the debates of this topic here. I
wonder for example, whether the same inconsitancies result if you assume
that ALL patent text must be translated/implemented to achieve the
device or process covered. Or if it is possible avoid risking all
process patents when attacking software patents. The patent office
distinguishes between algorithms and processes in that the latter is
patentable and the former not. Is there a consistant anti software
patent, pro process patent position that maintains the distinction, even
if devious author attempts to clothe a software patent in references to
changes to electromagnetic states in magnetic media or chips?
Scott McGregor
Newsgroups: comp.patents
Path: sparky!uunet!wupost!psuvax1!uxa.ecn.bgu.edu!mflll
From: egg...@twinsun.com (Paul Eggert)
Subject: distinction between algorithms and processes
Status: R
Message-ID: <#+_*&y6xh@twinsun.com>
Temporary-Moderator: mfl...@uxa.ecn.bgu.edu
Sender: use...@twinsun.com
Nntp-Posting-Host: farside
Organization: Twin Sun, Inc
References: <45295@athertn.Atherton.COM>
Date: Mon, 10 Feb 1992 06:00:00 GMT
Approved: pate...@cs.su.oz.au
mcgre...@hemlock.Atherton.COM (Scott McGregor) asks:
Is there a consistant anti software patent, pro process patent position
that maintains the distinction [between algorithms and processes], even if
[a] devious author attempts to clothe a software patent in references to
changes to electromagnetic states in magnetic media or chips?
Yes. There are several reasonable ways to make a consistent distinction
between software patents and other patents. Perhaps the best known is the
position taken by the League for Programming Freedom: (1) The development,
distribution and/or use of a computer program should never be held to infringe
any patent. (2) A combined hardware/software system should infringe a patent
only if the hardware alone does so, or nearly. (3) If the hardware is a
general-purpose computer, then the combination should infringe only patents
covering the construction of such computers--no matter what the software does.
Another consistent way to distinguish software patents is the position taken by
Dan Bernstein: that you shouldn't be able to patent mental processes, i.e.
processes that in principle you can do in your head.
More details about these two positions can be found in my 1991/11/25
posting to this newsgroup entitled ``definition of software patents''.
... software [patent] opponents sometimes fail to convince others
because they start out with assumptions not generally recognized by
those others.
The same can be said for software patent proponents. Another problem is that
software patent proponents often seem to ignore earlier postings in this
newsgroup (:-). However, since this is Usenet, we should expect some noise,
and not let it distract us from the important issues.
Path: sparky!uunet!munnari.oz.au!metro!cluster!mcgre...@atherton.com
From: mcgre...@atherton.com (Scott L. McGregor)
Newsgroups: comp.patents
Subject: Re: distinction between algorithms and processes
Message-ID: <4044@cluster.cs.su.oz.au>
Date: 18 Feb 92 21:37:20 GMT
References: <#+_*&y6xh@twinsun.com>
Sender: n...@cluster.cs.su.oz.au
Lines: 144
Approved: pate...@cs.su.oz.au
In article <#+_*&y...@twinsun.com> egg...@twinsun.com (Paul Eggert)
writes:
>Yes. There are several reasonable ways to make a consistent distinction
>between software patents and other patents. Perhaps the best known is
>the position taken by the League for Programming Freedom: (1) The
>development, distribution and/or use of a computer program should never
>be held to infringe any patent. (2) A combined hardware/software system
>should infringe a patent only if the hardware alone does so, or nearly.
>(3) If the hardware is a general-purpose computer, then the combination
>should infringe only patents covering the construction of such
>computers--no matter what the software does.
So, let's consider a rubber curing process ( E.g. add 2 parts of this
chemical, then 1 part of this chemical, mix at 5 rpm for 4 hours at 150
degrees F, then cool for 14 hours in a water bath at 38 degrees F). This
is just a process description. Note that there is no equipment yet. Now,
my original question asked for solutions that would not overturn the
patentability of all processes. Here's a process description, let's
assume it is patented. Now what happens if someone builds a factory, with
a vat with a mixer, heater, and cooling system, plus two valve controlled
additive tanks. This is a physical system for making rubber. This is
different from the process, and also from the resultant product (the
rubber). None of equipment (vat, valves, tank, mixer, heater and
refridgerator) may be patented, nor may their arrangement in the factory.
But if this equipment is used according to the process, then it is an
infringement. The controlling of the equipment, in accordance to the
patented process, by management is what infringes.
Now lets see what happens if we introduce a computer. Let's say we hook up
a general purpose computer in the factory. It has a number of RS232
ports. One port is hooked up to the electric control for the heater, one
for the refridgerator, one for a clock, one for each of the valves, etc.
Still the factory now including the computer (but no software) still is
not infringing--as long as they are not running that process. But someone
writes a program for this computer. It does things like send to various
different RS232 ports strings like "open valve" "38 degrees F", etc. It
reads from the clock to determine when to send these strings down. Now
when the program runs, it causes the factory to perform the patented
process.
Now here we have done only one thing, writen and run some software, and
apparently a non software process is infringed. What are the possible
responses we could take? There are two:
1) By the LPF view, we have met all 3 rules--it is definately software
now, so it must be okay to infringe the rubber making process, because LPF
says infringement should not apply to software. We can also conclude this
using Dan Bernstein's rule. Is the process itself abstractly describable
and is it able to perform the abstract description in one's minds with
mental representations? Definately--you can imagine the whole factory and
process in your mind, and probably just did above. Of course, this
creates a gaping hole for every would be physical process infringer to
drive a truck through: Just rewrite the process as a program, and you
can't be held for infringing. If these sorts of rules for separating
software algorithm patents from process patents are used then we undermine
the starting goal that I set--namely to define a rule that prevents
software patents WITHOUT undermining other process patents.
2) Of course, we could go the other way, and say look--you can't get
around the patent on the physical process just by taking the written
description and rewriting it as software. This protects the physical
process patent status--but what has become of the goal of preventing
patents from affecting software from patent issues? We have lost the LPF
and Bernstein models. In fact, it may be that the programmer who wrote
the code that sent "38 degrees F" to "/dev/refrig" hadn't the slightest
idea that this was going to be used to cure rubber at all, and yet that
programmer is still smack dab in the patent suit.
My reason for talking about rubber processes above is because this is the
same kind of question that was raised in Diamond vs. Diehr, a suit about a
rubber process, but where the question of infringement came down to
whether using a computer let you off the hook. But lets be clear about
this. If I build a black box and it takes signals in and produces signals
out, is the process it uses to transform those signals patentable or not?
Let's say it is a Dolby s/n reducer. It's analog electronic circuits. So
let's say it patentable. Now I tell you that I've built a new version
that takes the same inputs, generates the same outputs, uses the same
process--but this time there is an A/D converter connected to a general
purpose computer running an FFT connected to a D/A convertor. That FFT is
just the digital analog of what the old analog hardware did, but now in
software. Can I now safely infringe Ray Dolby's patents? If so, doesn't
that mean his process patents are meaningless?
The problem with the LPF and Bernstein models is that they break when they
are applied to software embedded in physical processes. That someone can
do an algorithm in their mind, or in a computer, is of little utilitarian
value if the results stay strictly in their mental state, or in registers
in computers, and so no one ever acts on them.. It is only when they are
output, changing physical states on disk or on output ports, or printed or
displayed that they can be used to achieve utilitarian objectives. But at
that state they are physical processes--the only difference is what goes
on inside the black box.
>> ... software [patent] opponents sometimes fail to convince others
>> because they start out with assumptions not generally recognized by
>> those others.
>The same can be said for software patent proponents.
Yes, that's most likely true. And from a debate standpoint that leaves us
at the proverbial standoff. But just as software is valuable only after it
leaves the mental only world and begins to effect the physical world, so
the debate will only matter when it is translated into real world actions.
The reality is that only the congress, the courts and the PTO are in a
position to affect changes to what is patentable or is not. They aren't
starting from no position and trying to choose one for the first time.
They already have one, and it allows software patents at present. Rolling
back the clock mentally doesn't make it so physically for these people. If
you want them to change the rules, it is not enough to say that the
proponents case as faulty as the opponents case. It is not enough to say
that you have noble goals. It is necessary to demonstrate the causal
connections that prove the calamities that have not yet caused the
software industry from growing spectacularly. This is not so because I say
it must be this way. This is so because that is how the congress and
courts work. If software opponents don't care about what happens in
reality and just want to vent their frustration--that's fine. But if they
want to have an effect, as I believe the authors of the LPF viewpoint do,
they need to sharpen their argument a bit more. They are off to a good
start, but they haven't reached causal proof yet.
> Another problem is
>that software patent proponents often seem to ignore earlier postings in
> this newsgroup (:-).
Perhaps this is directed to me. I certainly don't mean to ignore earlier
postings. I've studied the LPF rules closely, and Bernstein's rule as
closely as I could from the postings that I have received. Both do seem
good at making sure that they protect all of the things people would think
of as software. But I think that they fail in that they over restrict
process patents. I don't think this has been well defended in past
postings. For some people, who don't think process patents are a good
idea either, that may seem fine. But the LPF viewpoint suggests that its
authors goals are more limited, and that they don't want to undermine
other patents with their changes. Certainly some software patent
opponents here have claimed that they actively support patents in other
areas including process patents. Because every process description can be
transformed into a software algorithm, this continues to be an area of
concern for people who would oppose software patents, but support process
patents.
--
Scott L. McGregor mcgre...@atherton.com
Atherton Technology fax: 408-744-1607
1333 Bordeaux Drive
Sunnyvale, CA 94089
Path: sparky!uunet!munnari.oz.au!metro!cluster!riv...@theory.lcs.mit.edu
From: riv...@theory.lcs.mit.edu (Ron Rivest)
Newsgroups: comp.patents
Subject: Software patents and self-revealing inventions
Message-ID: <4167@cluster.cs.su.oz.au>
Date: 20 Feb 92 02:45:25 GMT
References: <45295@athertn.Atherton.COM> <1992Feb17.150629.28895@vicom.com>
Sender: n...@cluster.cs.su.oz.au
Organization: MIT Lab for Computer Science
Lines: 85
Approved: pate...@cs.su.oz.au
I'm not convinced that there is really a well-founded argument against
software patents per se. At least, I've never heard one. There are certainly
many bad software patents that have been issued, but what is the argument
against ALL software patents?
Let me give a thought example.
Suppose someone proves that P = NP, and in doing so develops a
polynomial-time algorithm for solving any problem in NP. Depending on
whether or not his algorithm was patentable, he would very likely (if
he were motivated to capitalize on his efforts), do one of two things:
(A) if his algorithm was not patentable, he would set up shop solving
other people's problems, keeping his algorithm a trade secret.
In other words, you bring your traveling salesman, knapsack
or integer programming problem to him, and he solves it for
$5,000 a problem. The algorithm might remain secret for
many years, and might even die with him, not to be discovered
for centuries. (What was Fermat's proof for his Last Theorem,
anyway?)
(B) if his algorithm is patentable, he discloses it in a patent,
collects royalties for 17 years, and then the idea is in the
public domain.
I think scenario (B) is arguably preferable to scenario (A), in terms
of any reasonable measure of social utility. (We may presuppose that,
since it is not our invention, no other choices are open to us. We
have to define the rules for inventors to live by.) Achieving (B) is
exactly the kind of thing that the patent law is intended to
do---require someone to disclose his invention in return for a
limited-time exclusive use (or licensing rights). Society is better
off if such an invention were, after a limited time, in the public
domain. I think any argument against this interpretation would
probably have to be against all patents per se. But I would like to
hear opinions.
My example was carefully chosen to be an invention that is not what
you might call ``self-revealing''. We can define an invention to be
``self-revealing'' if practicing or using this invention in the
specified manner is more-or-less guaranteed to make its operation and
key inventive idea clear to someone skilled in the art.
For example, consider pop-tops on soda cans. Once you see one, it's
obvious what the idea is; it is ``self-revealing''. Same goes for
overlapping windows on computer screens, say. (Or XOR cursors.)
One plausible argument against most software patents is that they are
generally self-revealing. Self-revealing inventions are a little bit
strange to patent, because its not clear why the goverment should
choose to grant a patent for one. If the goal of the patent process
is to get inventions in the public domain, eventually, then modifying
the law to exclude the patentability of self-revealing inventions
would also meet that goal. Once the invention is made, the only way
to practice it is in some self-revealing manner that effectively
places it in the public domain.
I realize that the notion of ``self-revealing'' may be a bit difficult
to define precisely. Probably you want to distinguish reverse
engineering (disassembling object code, or doing a chemical analysis,
say) from ordinary operation, and so on. I don't know if this can be
done in a reasonable manner. It's probably an easier question to
answer, however, than ``what is an algorithm'' or ``what is a mental
process''.
So I raise the key question: should ``self-revealing'' inventions be
patentable?
I suppose the strongest argument in favor of patenting self-revealing
inventions is to protect the investment a company may have made in
making the invention in the first place. Are there good examples of
areas where self-revealing inventions are expensive to create?
Pharmaceuticals come to mind, but to my mind they are not really
self-revealing (you have to be a chemist, not a patient, to figure out
what's in the pills). Certainly a policy against ``self-revealing''
inventions would outlaw the patenting the ``look and feel'' of a
software system.
Perhaps we are asking the wrong question when we ask ``should software
be patentable?'' The notion of ``being software'' is a technical one
that is not directly related to the goals of the patent system. Other
notions, such as utility, obviousness, or self-revealingness are more
germane. We might focus on identifying these criteria more carefully,
rather than trying to argue for or against the patentability of
software per se.
Path: sparky!uunet!munnari.oz.au!metro!cluster!mcgre...@atherton.com
From: mcgre...@atherton.com (Scott L. McGregor)
Newsgroups: comp.patents
Subject: Re: distinction between algorithms and processes
Message-ID: <4178@cluster.cs.su.oz.au>
Date: 21 Feb 92 18:14:58 GMT
References: <1992Feb17.150629.28895@vicom.com>
Sender: n...@cluster.cs.su.oz.au
Lines: 49
Approved: pate...@cs.su.oz.au
In article <1992Feb17.150629.28...@vicom.com> egg...@twinsun.com (Paul Eggert)
writes:
>Yes. There are several reasonable ways to make a consistent distinction
>between software patents and other patents. Perhaps the best known is the
>position taken by the League for Programming Freedom: (1) The development,
>distribution and/or use of a computer program should never be held to infringe
>any patent. (2) A combined hardware/software system should infringe a patent
>only if the hardware alone does so, or nearly. (3) If the hardware is a
>general-purpose computer, then the combination should infringe only patents
>covering the construction of such computers--no matter what the software does.
Note that in the Diamond vs. Diehr rubber curing process case, the process
patent enforcement would be denied by the above, because the case involves the
*use* of software (rule 1). Additionally, it is a process patent, not a device
patent, and uses standard hardware (vats, valves, etc.) and general purpose
computers (rules 2 and 3). Thus this particular definition undermines a
physical process patent, when that physical process is mediated by a general
purpose computer under the control of software. Since effectively any process
can be put under computer control, effectively all process patents would be
undermined by these rules.
> Another consistent way to distinguish software patents is the position taken
by
> Dan Bernstein: that you shouldn't be able to patent mental processes, i.e.
> processes that in principle you can do in your head.
Here too, we find that the Diamond vs. Diehr rubber curing process would be at
risk. You can perform the rubber curing process in your mind, imagining
suitable vats, valves, connected in appropriate ways, operated for appropriate
times under appropriate temperatures, and yielding mental rubber. Or you could
simulate the process on computer. This is not fundamentally different from
performing an algorithm that includes changes to a video display, data line or
disk in your head. That you have the mental results doesn't matter until you
implement them in reality (by changing the video display, signals on the data
line or magnetic polarities on disk) any more in this case than in the former.
Again, the rule would undermine all processes patent, because as above all
processes which are describable (a requirement for patentability) are in
principle possible to do in your head. This is a primary difference between a
device and a process. Of course, doing them in your head doesn't get you real
results--it is USING the process in the real world that get's real results, but
both the LPF and Bernstein models would deny enforcement to REAL WORLD USE as
well as to the merely mental simulations.
--
Scott L. McGregor mcgre...@atherton.com
Atherton Technology fax: 408-744-1607
1333 Bordeaux Drive
Sunnyvale, CA 94089
Path: sparky!uunet!munnari.oz.au!metro!cluster!brns...@KRAMDEN.ACF.NYU.EDU
From: brns...@KRAMDEN.ACF.NYU.EDU (Dan Bernstein)
Newsgroups: comp.patents
Subject: Patents v. Trade Secrets (was Re: Software patents...
Message-ID: <4232@cluster.cs.su.oz.au>
Date: 24 Feb 92 05:12:50 GMT
References: <45295@athertn.Atherton.COM> <1992Feb17.150629.28895@vicom.com>
Sender: n...@cluster.cs.su.oz.au
Organization: IR
Lines: 27
Approved: pate...@cs.su.oz.au
Ron Rivest writes:
> (A) if his algorithm was not patentable, he would set up shop solving
> other people's problems, keeping his algorithm a trade secret.
> In other words, you bring your traveling salesman, knapsack
> or integer programming problem to him, and he solves it for
> $5,000 a problem.
Nobody in his right mind would use this service. Simulated annealing
works wonders for the NP-complete problems people care about. The
simplex method is adequate even for very large instances of linear
programming problems; Khachian's method and its subsequent improvements
provide guaranteed bounds if you need solutions in real time. These
advances have all come out of universities or very large companies,
which have always published their results as a matter of course. Your
hypothetical inventor wouldn't earn a penny---unless you give him a
patent, which lets him extort fees for 17 years.
Another argument: If your inventor (think of Karmarkar if you want)
really thought he could make money selling his invention, WHY WOULD HE
USE THE PATENT SYSTEM ANYWAY? Sorry for shouting, but your argument
seems to be ``If we give people algorithm patents, they'll make less
money than they would with trade secrets, so that's good for everyone.''
If this is true, why wouldn't your inventor keep his algorithm secret?
Do you think he's stupid?
---Dan
Path: sparky!uunet!munnari.oz.au!metro!cluster!brns...@KRAMDEN.ACF.NYU.EDU
From: brns...@KRAMDEN.ACF.NYU.EDU (Dan Bernstein)
Newsgroups: comp.patents
Subject: Re: distinction between algorithms and processes
Message-ID: <4234@cluster.cs.su.oz.au>
Date: 24 Feb 92 05:53:21 GMT
References: <1992Feb17.150629.28895@vicom.com>
Sender: n...@cluster.cs.su.oz.au
Organization: IR
Lines: 36
Approved: pate...@cs.su.oz.au
Scott McGregor writes:
> You can perform the rubber curing process in your mind, imagining
> suitable vats, valves, connected in appropriate ways, operated for
> appropriate times under appropriate temperatures, and yielding mental
> rubber.
Indeed. I will fight to the death to ensure that the process of curing
mental rubber is not patentable. If you have any rubber in your head
please feel free to cure it. :-)
> Or you could
> simulate the process on computer.
Good point. My position also implies that the process of simulating
rubber curing on a computer is not patentable.
None of this has anything to do with Diamond v. Diehr, which dealt with
curing *real* rubber. That's a *physical* process.
> Again, the rule would undermine all processes patent,
No, it wouldn't. I am rapidly tiring of this argument. You cannot, no
matter how hard you try, cure real rubber in your head. (If you can I'm
sure the National Enquirer will run an article on it.) Curing rubber is
therefore not a mental process. Why do you persist in saying that it is?
My proposal only has to do with mental processes. If it's essential for
the completion of a process that it be applied to physical elements,
then my proposal doesn't say anything about the process. I mentioned in
a previous article a patent on converting yeast into glue. To carry out
this process it is essential that you begin by heating the yeast. That's
an essential physical application. The process would thus remain
patentable with or without my rules. Your claim is thus false.
---Dan
Path: sparky!uunet!munnari.oz.au!metro!cluster!riv...@theory.lcs.mit.edu
From: riv...@theory.lcs.mit.edu (Ron Rivest)
Newsgroups: comp.patents
Subject: Re: Patents v. Trade Secrets
Message-ID: <4261@cluster.cs.su.oz.au>
Date: 27 Feb 92 02:25:23 GMT
References: <45295@athertn.Atherton.COM> <1992Feb17.150629.28895@vicom.com>
<4232@cluster.cs.su.oz.au>
Sender: n...@cluster.cs.su.oz.au
Organization: MIT Lab for Computer Science
Lines: 28
Approved: pate...@cs.su.oz.au
Dan Berstein says that ``no-one in his right mind would use this service''
(a service that solved instances of NP-complete problems). He argues that
there are good algorithms for many problems of interest (linear programming,
for example).
Dan's argument is a non-sequiter, and the conclusion is false. An algorithm
for solving NP-complete problems would also enable one to solve many problems
for which we do not have good algorithms now. Breaking cryptosystems,
optimizing circuit layouts, or proving mathematical theorems (at least ones
with short enough proofs) are all things some would pay good money for.
Dan also misunderstood the assumption implicit in the example: that
the inventor would make the most money from licensing his patent, and
the second-most amount of money by keeping it a trade secret and
selling his services. Thus, the patent system would motivate the inventor
to disclose (and patent) his invention. If you disagree with the assumption
in this example, then that merely means the example may not be the best one.
I'm sure it's easy to find other examples---ones that even Dan would agree
to---where licensing is obviously more profitable than selling the sevice.
The other assumption that some have responded with--that the patent shouldn't
have been granted because others would invent it anyway---is merely an
argument against patents in general...
Ron Rivest
MIT Lab for Computer Science
Path: sparky!uunet!munnari.oz.au!metro!cluster!0001811...@mcimail.com
From: 0001811...@mcimail.com (Carl Oppedahl)
Newsgroups: comp.patents
Subject: Re: Patents v. Trade Secrets
Message-ID: <4262@cluster.cs.su.oz.au>
Date: 27 Feb 92 03:05:00 GMT
Sender: n...@cluster.cs.su.oz.au
Lines: 92
Approved: pate...@cs.su.oz.au
Dan Bernstein (brns...@kramden.acf.nyu.edu) writes:
>Ron Rivest writes:
>> (A) if his algorithm was not patentable, he would set up shop solving
>> other people's problems, keeping his algorithm a trade secret.
>> In other words, you bring your traveling salesman, knapsack
>> or integer programming problem to him, and he solves it for
>> $5,000 a problem.
>
>Nobody in his right mind would use this service. Simulated annealing
>works wonders for the NP-complete problems people care about. The
>simplex method is adequate even for very large instances of linear
>programming problems; Khachian's method and its subsequent improvements
>provide guaranteed bounds if you need solutions in real time. These
>advances have all come out of universities or very large companies,
>which have always published their results as a matter of course. Your
>hypothetical inventor wouldn't earn a penny --- unless you give him a
>patent, which lets him extort fees for 17 years.
I don't think Mr. Bernstein is being fair to Mr. Rivest. Is Mr. Bernstein
saying Mr. Rivest chose an unfortunate hypothetical -- that as a factual
matter no one who has found a way to solve NP problems fast will be able to
make money renting out the engine that does it? If so, he is wasting
everyone's time. Mr. Rivest was giving an example to illustrate his basic
point -- that in the absence or unavailability of patent protection, many
people will choose trade secret protection instead. And Mr. Rivest is right
when he says this. Let me offer other examples:
1. Michelin making better radial tires. Nowadays, it is impossible to buy
a non-radial tire. But as recently as the 1970's, most tires were "bias-ply"
tires, and radials were rare and expensive. Only one company, Michelin, was
able consistently to make good radial-ply tires. They could have patented
all their manufacturing methods, but instead chose to keep them secret. The
resulting tires were not "self-revealing" -- looking at the tire did not
reveal how Michelin made them. I believe Michelin chose the trade-secret
route at least in part because they figured they'd make more money that way
(boy, did I go out on a limb in saying that!). Instead of selling tire-
making machines (read: selling linear programming software) they chose to
keep the machine secret and sell its output (read: sell solutions to linear
programming problems).
2. Formula for Coca-cola. The Coca-cola company, had it patented its
formula, would have had an "exclusive" on Coke for 17 years. Instead, it has
kept its exclusive for much, much longer.
3. Some clients of mine, whom I can't name, do the very sort of thing Mr.
Bernstein says no one in their right mind would pay for. For one reason or
another (e.g. concern that they might try to get a patent and then fail)
they have chosen to keep their methodology, which is part software, secret
and sell the only computational results.
>Another argument: If your inventor (think of Karmarkar if you want)
>really thought he could make money selling his invention, WHY WOULD HE
>USE THE PATENT SYSTEM ANYWAY? Sorry for shouting, but your argument
>seems to be "If we give people algorithm patents, they'll make less
>money than they would with trade secrets, so that's good for everyone."
>If this is true, why wouldn't your inventor keep his algorithm secret?
>Do you think he's stupid?
Mr. Bernstein misunderstands the argument. The argument is that in a world
where systems containing software cannot be patented, some inventors will
choose to keep the system a trade secret and sell its output, rather than
try to make money selling the systems. The reason some inventors will do this
is that selling the systems (which are self-revealing, given the availability
of reverse compilers and such) essentially begs others to rip off the idea
and sell derivative systems that do the same thing, all at a lower price, or
for free. Depending on the facts of a particular software system, the
inventor may realistically have no choice but the trade-secret route. And
depending on the facts of a particular software system, both the inventor
and the public may be worse off than if system sales had been possible at
a reasonable price.
The argument is that in a world where systems containing software can be
patented, some inventors will go the patent route. In many cases this is
a route that leaves both the inventors and the public better off. The public
is better off because the patent document spills all the beans -- scholars
and programmers alike enjoy the benefit of being able to read what is
analogous to "commented code" - a detailed description of the preferred
embodiment. The public is better off because they can have the engine itself,
and use it whenever they wish from then on, rather than waiting for free
moments in a trade-secreted machine that is only rented out. The public is
better off because the patent will expire and then everybody gets to use the
invention free of charge. The inventor must, by definition, be better off
because the inventor's conduct (which is assumed to be rational) was to select
the sales route rather than the (still available but forgone) trade-secret
route. No one (except, apparently, Mr. Bernstein) is assuming the inventor
is stupid.
Carl Oppedahl
30 Rockefeller Plaza
New York, NY 10112-0228
Path: sparky!uunet!think.com!mips!mips!munnari.oz.au!metro!cluster!
egg...@twinsun.com
From: egg...@twinsun.com (Paul Eggert)
Newsgroups: comp.patents
Subject: Re: Patents v. Trade Secrets
Message-ID: <4288@cluster.cs.su.oz.au>
Date: 1 Mar 92 20:16:02 GMT
References: <45295@athertn.Atherton.COM> <1992Feb17.150629.28895@vicom.com>
<4232@cluster.cs.su.oz.au> <4261@cluster.cs.su.oz.au>
Sender: n...@cluster.cs.su.oz.au
Organization: Twin Sun, Inc
Lines: 13
Approved: pate...@cs.su.oz.au
Nntp-Posting-Host: bi
Apparently-To: comp-pate...@uunet.uu.net
Ron Rivest asks us to conduct a gedanken experiment, and suggests that
in a world without software patents, people would keep their software
inventions a secret, and would instead sell services that solve
instances of problems. But we don't have to conduct a gedanken
experiment, because we already conducted a real experiment. Until the
mid-1980s software was generally exempt from the scope of patent laws.
If Rivest is right, then allowing software patents should have caused a
flood of patent applications for software methods that had previously
been kept secret for the reasons he described. But we observe no such
patent applications -- on the contrary, most software patents are for
techniques that are obvious to anyone versed in the field. So Rivest's
conclusion is contradicted by the facts.
Path: sparky!uunet!munnari.oz.au!metro!cluster!0001811...@mcimail.com
From: 0001811...@mcimail.com (Carl Oppedahl)
Newsgroups: comp.patents
Subject: Patents v. Trade Secrets
Message-ID: <4319@cluster.cs.su.oz.au>
Date: 4 Mar 92 02:37:00 GMT
Sender: n...@cluster.cs.su.oz.au
Lines: 16
Approved: pate...@cs.su.oz.au
Paul Eggert (egg...@twinsun.com) says:
If Rivest is right, then allowing software patents should have
caused a flood of patent applications for software methods that
had previously been kept secret ...
But we observe no such patent applications ... so Rivest's
conclusion is contradicted by the facts.
It happens patent applications are kept secret by the Patent Office.
So no one (except those who work in the Patent Office) has observed such
patent applications, or any other patent applications.
What Mr. Eggert has not seen in the area of patent applications is thus not
support for the claim that Mr. Rivest is contradicted.
Path: sparky!uunet!munnari.oz.au!metro!cluster!brns...@KRAMDEN.ACF.NYU.EDU
From: brns...@KRAMDEN.ACF.NYU.EDU (Dan Bernstein)
Newsgroups: comp.patents
Subject: P v. NP Problems ( was Re: Patents v. Trade Secrets
Message-ID: <4348@cluster.cs.su.oz.au>
Date: 4 Mar 92 14:27:42 GMT
References: <45295@athertn.Atherton.COM> <1992Feb17.150629.28895@vicom.com>
<4232@cluster.cs.su.oz.au>
Sender: n...@cluster.cs.su.oz.au
Organization: IR
Lines: 19
Approved: pate...@cs.su.oz.au
Ron Rivest writes:
> Dan Berstein says that ``no-one in his right mind would use this service''
> (a service that solved instances of NP-complete problems).
[ ... ]
> An algorithm
> for solving NP-complete problems would also enable one to solve many problems
> for which we do not have good algorithms now.
That conclusion is completely illogical. There is no reason to believe
that a polynomial-time algorithm for NP-complete problems will ever be
useful in practice. (And the premise is poorly stated: we _do_ have an
algorithm for solving any bounded problem, even NP-complete problems.)
I strongly suggest that you avoid imaginary examples in illustrating
your point. Otherwise you're arguing about some la-la land which has
little relevance to the real world.
---Dan