[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Open Software License v2.1

Nathanael Nerode <neroden@twcny.rr.com> writes:

> Brian Thomas Sniffen wrote:
> <snip>
>>>Stop right there.  You didn't invent the software I wrote, regardless of
>>>what the overloaded US Patent Office might think.
>> Sure I did.  Well, if you're writing some software to do
>> Diffie-Hellman key exchange, that Diffie and Hellman most certainly
>> *did* invent that, and have a (now-expired) patent on it.  If you're
>> using the RSA cryptosystem to provide secrecy and authentication for
>> data, then R, S, and A really did invent it, and you're just following
>> in their footsteps.
> What if I invented it independently?  Assume for the sake of argument
> that I am a genius.  ;-)

Then wait until their patent expires.  That's no different from the
situation with traditional hardware patents.

>>  You didn't come up with the idea of using
>> discrete log as a trapdoor function, or the idea of using trapdoor
>> functions for key exchange.  You're just writing in accord with the
>> algorithm they published.
> Or did I?
> Patents are odd that way; they bite independent inventions.

But only for a very limited period of time.  And they require full
disclosure of the patented technique, so it guarantees that you won't
*need* an independent invention a while later.

>> Rivest et al. could have kept their algorithm secret, and only sold
>> sealed boxes for performing in-line encryption.  Instead, they
>> accepted a bargain with society in which they published their results,
>> allowing others to build on them, and received a monopoly on those
>> results for a period of time.
>> If you reject that bargain for software, that's fine -- but it means
>> you should reject both halves, and isolate yourself from hearing about
>> patented techniques like LZW compression, public key cryptosystems,
>> the tab key, etc. until after their patents have expired.
> Hey, I *never* knowingly read about patented techniques.  That doesn't
> mean I won't accidentally implement one.

IF the patent system implemented the ban on obvious patents the way it
was originally meant, then you would never accidentally implement
one.  The part that's broken is not the idea of patents on systems
incorporating software, but the part where a company can get a
patent's monopoly protection on an obvious technique.

To be specific, what I mean by obvious -- and what I think was
originally meant by obvious, if you read the notes from the drafters
of early patent law -- was that it couldn't be kept a secret in
practice.  So you (well, Eli Whitney) can get a patent on a method for
producing interchangeable rifle parts, because even if I'm staring at
the things I can't tell how you made them so precisely.

Similarly, with the patent on the technique of using LZW to compress
files, the technique of using RSA to protect secrecy, etc: even
looking at the inputs and the outputs, it's not obvious what you're
doing.  So society benefits by trading RSA, Inc. a short monopoly on
exploiting RSA as a cryptosystem in exchange for publication of how
they did it.

One-click shopping, on the other hand, is obvious.  As soon as I see
a description of the result, it's clear to me how to build something
like that.  XOR cursors, similarly, are obvious -- five seconds of
playing with such a machine is all it takes to understand what it's
doing and how to make one.

If that criterion in the already-written law is strictly implemented,
then patents on systems incidentally including software won't be a problem.


Brian Sniffen                                       bts@alum.mit.edu

Reply to: