Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!mips!pacbell.com!decwrl!sgi!karsh@trifolium.esd.sgi.com
From: karsh@trifolium.esd.sgi.com (Bruce Karsh)
Newsgroups: comp.arch
Subject: Re: Patents and Architecture
Summary: Patents are good
Message-ID: <62959@sgi.sgi.com>
Date: 27 Jun 90 09:23:47 GMT
References: <62864@sgi.sgi.com> <1990Jun26.234134.2302@ico.isc.com>
Sender: news@sgi.sgi.com
Reply-To: karsh@trifolium.sgi.com (Bruce Karsh)
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 279

| In article <1990Jun26.234134.2302@ico.isc.com> rcd@ico.isc.com (Dick Dunn) writes:
>karsh@trifolium.esd.sgi.com (Bruce Karsh) writes:

> ...What I think is happening is that so much time and effort is spent
> re-creating old ideas that sw engineers are not adequately exposed to
> the process of innovation...

|Possibly so, but doesn't this follow directly from the fact that so much of
|what we do, at the macro level, is the same as what we've done before?

Yes, we redo the same tired old things too much.  This is very bad for
users.

|That is, solving essentially the same problem requires essentially the same
|pieces unless there's some reason to think there's a better alternative.

So patents say that you either have to a) find a better alternative, b) do
something else, or c) pay the price to the holder of the patent.  There's
nearly always a better alternative.  We haven't really gotten close to an
ideal software architecture.

|In other newsgroups, this would be recognized as the beginnings of "reuse"
|of software and would be lauded!

Patents don't prevent reuse.  If you license a patented software invention,
then you may re-use it.  If you don't license it, then the issue of re-use is
moot since you can't even use it.  Patents discourage software plagarism.

|Allow a different statement of the problem:  I resent the fact that I may
|come up with a way to solve a problem but be restrained from using my
|solution because someone else, somewhere else, thought of it and patented
|it.

Should people resent that they can't republish other's writings?  The
inventor has rights to his invention in much the same way that an author
has rights to his writing.  Why should you be allowed to make money off of
someone else's invention without paying the inventor.

|  It's not enough to come up with the right answer; you'll have to
|search to see whether you're allowed to use it.

The alternative is for people to keep their inventions a secret.  If they
do that, then you still won't be able to use them.  Patents may cause
software engineers to have to do more research.  That seems like a good
thing to me.

>     o	There is an economic incentive to invent.  This causes there
> 	to be more inventions.

|No.  There is an economic incentive to patent.  This causes there to be
|more patents.  There is greater benefit from having a patent on something
|ubiquitous and mundane than on something novel and creative.

Try to get a patent on something that's not an invention.  It's very
difficult.  Occasionally the patent examiners may make a mistake, but
they are pretty sharp guys and they put a great deal of effort into
making sure that the patent application really represents an invention.

>     o	There is an economic disincentive to repeatedly re-implement the
> 	same thing.  Hence inventors are constantly forced to find new
> 	intellectual areas to invent in.

|There are many tasks which do not require invention.

Perhaps there are, but I don't know of any which are economically important.

|Invention is a business risk--it can't be planned and scheduled.

That's one reason why patents are so important.  If a company invests a lot
of money in creating an invention, the patent gives them a fighting chance
to make enough money to recoup their R&D expendetures.  Without a patent,
someone else will steal the idea and sell it for less because they aren't
burdened by the cost of inventing.

|  Forcing invention where
|it is not needed simply makes it more difficult to navigate the minefields
|of patents to solve a perfectly ordinary problem.

Practically speaking, is this really a problem?  Normally, it's only a
problem if you are re-solving a problem the same way that someone else
already has.  Other fields have had to contend with this problem and it
has only helped these fields.

If you can find even one place where the invention was published, then you
are OK.  The problems have not occured when people stumbled into new ideas.
The problems occur when people try to make money off of someone else's
invention.

|There is no prohibition against full disclosure of a non-patented idea!
|It is of even greater use and provides more of a building block for others
|by the fact that there is no restriction on its use.

Right.  If you invent something and publish it, then it can't be patented.
Of course, if you then try to sell you invention or try to go into business
selling it, then you'll have a hard time getting anyone to invest.  If you
try, you'd better wear a very white shirt.

> Since patents require full disclosure of the invention, the invention
> becomes "obvious" after it has been disclosed.  That is one of the
> primary purposes of a patent.  So it doesn't do much good to argue that
> a patented invention is obvious.  If you want to invalidate a patent,
> you have to find prior art.

|There are several patently (sic) false statements there.  Inventions do not
|become "obvious" by disclosure.

Very often they do.  Certain window system ideas and XOR cursors are examples
of this.  Once you've seen these ideas, they are obvious.  If they were so
obvious before and if they are so necessary now, why didn't anyone do them
before the patent was filed?

There are other ways, of course, to invalidate a patent other than prior
art.  For instance, you can try to show that the patent application is
fraudulent.  The point is that arguing that a patent is obvious is a
difficult case to make.  If you find prior art, you've got a good chance
to break the patent.

|A clever idea is still clever even after it's been explained to you.

I think this is so.  However, many people deny that a new idea is
clever.  They say that the idea was obvious.  They usually say that
after they've already seen someone else making money off the idea.

|Nor, conversely, does the granting of a patent
|make something an "invention".  Moreover, it *does* do some good to argue
|that a patented invention is obvious--patents are *not* supposed to be
|granted for "inventions" which are reasonably evident to practitioners in
|the field of the invention.  That's the way the law written; that's the way
|it's supposed to work.  The US Patent Office has, unfortunately, been
|bamboozled on a few...but still, it is *not* always necessary to show prior
|art.  (In other words, the existence of "prior art" is a sufficient, but
|not necessary, condition to invalidate a patent.)

There are a whole bunch of ways to try to invalidate a patent.  Trying to
show that an idea is obvious without any prior art is going to be a pretty
hard battle.  If the invention is so obvious and so necessary, then why
didn't anyone do it before?

|...and the reason people are upset about some patents is that they *force*
|different architectures (or implementations).  Innovation is good sometimes
|but not always.

The architectures we have now need lots of improvement.  Now is not the
time to stop innovating.  I want to see new and better architectures.  A
lot of people have spent so much time dealing with the old architectures that
they've lost sight of the importance of making new ones.

>...If you invent something really new in your
> architecture, then the law gives your architecture an advantage.

|OK, but a lot of us agree with that only because you said "really new."  If
|you just come up with some trivial tweak, you don't need to pretend that
|you've just invented sliced bread and go around trying to get a bundle of
|money for it.

If you've come up with a "trivial tweek", then who's going to give you
money for it?  Only someone who needs the "trivial tweek" really
badly?  If it's needed really badly, then it isn't trivial - its
basic.

|Perspective:  There's a little faction that thinks patents are an
|incredibly bad idea and should be completely eliminated.  There's another
|little faction that thinks patents are an incredibly good idea and should
|be created whenever possible.

I'm in the later category.  The continuous reinventing and plagarizing of
computer inventions has really hurt the progress of the computer field.  I
say that computer engineers should either patent or publish their ideas.

Further, the secrecy about computer algorithms has hurt.  There are a lot
of inventions in modern software which are being kept secret.  Patents can
provide protection without requiring secrecy.

|  The rest of us are in between:  We think
|patents can do some good, but there are some patents that should never have
|been granted.

Perhaps, but have these people looked at the research expenditures which went
into creating the patents and the search for prior art that occured before
the patents were granted?  Or are these people just grumbling because they
aren't allowed to make money off of somebody elses work?

I'm sure that the patent office makes some mistakes.  But they generally do
an exceptional job.

|Every patent creates a potential for litigation, a need for paperwork, and
|a bit of an impediment to progress.  The patent ought to be significant
|enough to make the hassle worthwhile.

Patents are not cheap.  Few organizations or people will invest in getting
one unless they believe that the invention is significant.

|Imagine if I had to deal with a patent every time I created a hash table,
|ran a loop counter down to zero instead of upward, used a doubly-linked
|list, passed a procedure as a parameter to another procedure, exchanged two
|values without a temporary (wonder if that one's covered by the xor-cursor
|patent?!:-), etc.  We could easily create a world (or rather, a country;
|not everyone will put up with this) in which every programmer has to work
|accompanied by an attorney.

All of the software technologies that you have mentioned have been previously
published.  You may use these techniques to your heart's content.  What
techniques are you prevented from using which are not inventions?

How about if we create a world in which intellectual property rights are
respected?  If these rights are respected, then the lawyers won't be
required in order to program.  Right now we don't adequately respect
other's rights and inventors have to resort to lawyers for protection.

> Therefore, instead of complaining about patents, I'd encourage software
> engineers to go out and invent new things and patent them.  It'll be
> good for you (or your company), and it'll be good for the field.

|So the purpose of invention is to create patents???  This will quickly
|become circular.

No the purpose of patents is to promote invention.  I urge software engineers
to create inventions and to patent them in order to enhance their invention's
economic value.

> Have people noticed that certain new window systems look almost exactly
> like the Macintosh?  They don't look that much like NeWS or like UWM.
> They look like the Macintosh.  If I were one of the designer's of the
> Macintosh UI, I'd be upset about having my design stolen...

|If one were less inclined to jealousy and possessiveness, one might be
|flattered to have found a good way to do something, and to have one's
|ideas corroborated by others.

Of course, a musician is flattered when somebody plagerizes their music.  But
you can't pay your band with flattery.  I suppose the window system designers
were flattered.  But perhaps they were just a little upset about being
ripped off too?  What about the people who funded the R&D to make the window
system?  It must have crossed their mind that someone was trying to make
money off of their investment without paying for it.

| And if one were truly inventive, one would
|be off inventing the next step instead of litigating the previous one.

Invention costs money.  If your invention is stolen, then you're not too
likely to go off inventing the next step.  Who'll pay the bills?  Patents
allow you to continue inventing because you can make some money from it.

>...Why didn't
> software designers create a new look instead of rote-copying someone
> else's?  Could it be for the same reasons that unscrupulous students
> sometimes plagarize other's work?  Is it because they don't think they
> are capable of doing better themselves?  Or is it because they don't
> want to invest the same amout that the original inventor invested?

|Flame time.  If you're too close-minded to see that there's an advantage
|to having user interfaces look similar across various systems, you've got
|no business preaching about invention.  You're advocating invention and
|change for their own sake, users be damned.  If you'd paid any attention at
|all to the concerns of user interface design, you'd know there's some value
|in consistency, so that users can move from one system to another without
|having to start over from scratch--or worse, to have their usage habits and
|patterns on one system turn out to be destructive on another.

You assume that the users really like the UI's which are out there
now.  They don't.  They barely tolerate them.  Are we going to serve
our users "last year's model" forever?  We need to make new UI's that
serve users better.

If a user interface is in the public domain, then it can be copied and
copied forever.  Feel free to do this.  I don't know whether users will
prefer the new ones with innovative features or the standard ones.  But
if I had to bet, I'd bet on the new ones.

|You've just preached that consistency is lazy, incompetent, and perhaps
|even unscrupulous...and that NIH is not only noble but should be encouraged
|by law.

Not at all.  Existing software designs are not good enough for users.  We
have to continue to make better ones.  The inventors of the new ones should
be paid for their inventions.  It's not OK to plagarize them.

			Bruce Karsh
			karsh@sgi.com