Re: [dvd-discuss]Does software really satisfy the requriments for Copyright?

["Michael A Rolenz" <Michael.A.Rolenz(at)aero.org>]

"Stephen L. Johnson" <sjohnson@monsters.org>
Sent by: owner-dvd-discuss@eon.law.harvard.edu
05/15/2002 01:11 PM
Please respond to dvd-discuss

 
        To:     dvd-discuss@eon.law.harvard.edu
        cc: 
        Subject:        Re: [dvd-discuss]Does software really satisfy the requriments for 
Copyright?


>
>Richard H. Hartman was heard to have written:
>> The biggest problem is attempting to shoehorn a New Thing (tm)
>> into laws written for other things entirely.  There is much debate
>> on whether software is expression or device (to be protected by
>> copyright or patent).  Well ... it has aspects of both, and potentially
>> in greater or lesser degree depending upon the software in question.
>> Whoever said "there is nothing new under the sun" was quite wrong,
>> sometimes there _is_ something new, and it deserves new and unique
>> treatment instead of attempting deal with it as if it were something
>> else.
>> 
>> btw: on the copyright vs. patent argument, I tend to believe
>> that software -- in general -- is "most like" a device (albiet
>> one that has a non-physical state) and deserves patent protection
>> rather than copyright.  That said, the patent process both 
>> a) moves too slowly and b) protects too long to be the appropriate
>> protection mechanism.
>
>In short and speaking strictly theoretically, I think that copyright and 
>patents are both appropriate for software. Copyright for protecting the 
>implementation of an idea, and patents for protecting truly unique and 
novel 
>algorithm. But we've got do deal with the real world and not a 
theoretical one.
>

Actually, claiming both copyright AND patent for the same thing is 
theoretical rather than pragmatically rooted in the "real world". There 
was good reason to make the terms different originally. Furthermore given 
the extreme term of copyright today, there is even more pragmatic reason 
to not grant ANY copyright to software. Do you want to administer it? NO. 
Does it benefit anyone in 120yrs, 100yrs, 50yrs,25yrs,10yrs? NO.BTW ? 
Algorithms are NOT patentable. The USSC ruled that they were more like a 
thing of nature and that was wise.

Pragmatically there is no reason to patent algorithms either. They are 
merely clever ways to manipulate bits. Given the speed of computers, there 
is even less reason today. If I don't need to do a 1M point FFT, you can 
brute force anything in the thousand point range during a windows hiccup. 
Pragmatically do you want to open up a Pandora's box of litigation over 
families of algorithms. Go to the library and look at a book on FFTs. 
Cooley Tukey is only ONE algorithm. You can't patent a family of 
algorithms because it's too general. SO you are patenting specifics. Every 
minor modification to an algorithm that gives it better performance in one 
circumstance would qualify for a patent (e.g., "FFT algorithm that 
performs XX times faster than Cooley Tukey on Cray 1 computer" ? that's 
not made up. I saw a paper presented years ago). Do you allow someone to 
patent generalizations of specific algorithms? Does that mean that the 
first one patented infringes upon the generalization? How much of a 
generalization does it have to be? (e.g., Generalization of binary 
algorithm to general Galois Field?)



>Based on modem copyright doctrine (I'm not including that abomination the 

>DMCA) and the way the PTO is allowing patents for everything under the 
sun, 
>copyrights are the best protection for software. And it would be perfect 
if 
>copyright terms where limited to say around 20-30 years, and the old 
>requirement to having to register copies for full copyright protection.
>

NO argument there. I'd accept Macauley's 42yrs as the "You have had your 
bribe not cease and begone. Never discuss this matter again least you get 
NOTHING in return"

>Programmer/companies have the protection to allow them to (try) to make 
money 
>off selling/licensing/etc. Other would have access to the sources to 
examine, 
>learn from and using fair usage for using small bits of code in other 
>programs. If companies want to protect any other "Intellectual Property" 
by 
>not release the source code, let them go with Trade Secrets and NDA 
>agreements. But you don't get copyright protection also.
>

I contend that Trade Secrets and NDA are the proper form of protection for 
source code. As for object code, the what is the purpose of copyrighting 
it? What does one wish to accomplish WRT to object code? What is the end 
goal? Is full copyright really necessary to further that goal?

>And I say no to patents under the current PTO policies. Computer 
programming 
>is appling a subset of a finite number of algorithms. Most, if not all, 
of the 
>basic algorithms have been know for decades with plenty of prior art. 
Nothing 
>much is new under the sun.
>

Then there certainly is nothing patentable in an algorithm. Perhaps only 
certain small classes of very useful algorithms. The problem with the PTO 
is that they haven't revised the things that can be patented but increased 
the scope. Take compression algorithms. One can argue that patents may 
have encouraged the development of compression algorithms <D/A here. I 
don't believe this has been the case> but do we need a zillion variations 
on LZW? The patent monopoly worked. We have some great, close to optimal 
algorithms <again I am being D/A. compression is class of algorithms along 
with encryption that is patentable>. Do we need to keep patenting nearly 
suboptimal ones? Or ones that have utility in only a few circumstances? 
It's a lot of work for little pay off or progress.

>Right now the Patent Office is awarding patents for trivial things that 
don't 
>deserve one. Patents for software that any competent programmer would 
come up 
>if they were required to develop it. Given the same task to a large 
number of 
>programmer, a significant number of them come up with a similar solution. 

>There goes the patent requirement of being non-obvious.
>
>The only softare patent I come close to agreeing with is the RSA patent 
on 
>encrypting with trap door functions. That was clearly non-obvious to 
somebody 
>competent in the field. It took a PHD Math Professor to discover it. And 
it 
>was truly novel and unique.
>

Encryption is patentable. I agree it was non obvious and I think it 
deserved it. Interestingly enough, allowing patenting of encryption also 
forces public disclosure and allows peer review-which is exactly what 
DVDCCA etc are trying to prevent.

>Sorry for the meander comments. I'm been rolling these though around in 
my 
>head for some time. They just cam spilling out.
>
>Stephen L Johnson <sjohnson@monsters.org>
>
>
>
>>