cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <stef...@apache.org>
Subject [maybe OT] LZW patent [was Re: MathML package structure?]
Date Fri, 18 Jan 2002 11:30:04 GMT
"Andrew C. Oliver" wrote:
> 
> Just FYI
> 
> http://lpf.ai.mit.edu/Patents/Gif/Gif.html
> 
> It does not expire until next year.  Not sure what the ramifications of
> the "variants" and new patent "applications" are.

The patent is about the LZW algorithm that was used by Compuserve to
create the GIF image format, but they didn't know that Mr. Welch (the
'W' in the acronym) already patented it for high-speed disk controllers
at Unisys.

> I wonder if I could file a patent for a method for finding the composite
> whole value of two operands and charge 1.5%.

Well, careful here: Lempel and Ziv (the first two letters of the
acronym) did years of research on estimation of information entropy of
an ordered stream and the algorithms they devised, IMO, fully deserved
patentability because they were 'innovative' in any possible sense.

The *problem* here is not the concept of software patent, but the fact
that the US patent office does not have enough
power/knowledge/capabilities to distinguish between an 'innovation' and
an 'obvious change' of current art.

What Welch did was applying an extremely brilliant algorithm (the first
loss-less compression argorithm that didn't transmit explicitly its
dictionary, unlike good old Huffman coding) with an 'obvious'
modification to suit his needs, to an hardware device.

And patented *both*! the device (a disk controller with on-fly
compression) *and* the algorithm. which *he* didn't invent!

And in my book, the LZ algorithms are *NOT* comparable to "a method for
finding the composite whole value of two operands".

I think there are just a few software patents that really deserve their
status (the RSA encryption algorithm, for example) but there are *TONS*
that don't. Or, at least, I would *personally* have came out with if I
had to build a system of that sort. Almost all speech recognition
patents are dumb, but they use very complex math so they scare the crap
out of the patent office! like using convolutions to match signals! wow,
that's non-obvious! NOT! 

good for Cauchy and Cantor that aren't here to witness the result of
their geniality :(

Patents aren't the problem: the incredible ignorance of the US patent
office is!

-- 
Stefano Mazzocchi      One must still have chaos in oneself to be
                          able to give birth to a dancing star.
<stefano@apache.org>                             Friedrich Nietzsche
--------------------------------------------------------------------



---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org


Mime
View raw message