"Jerome M. BERGER" wrote:

>         I may be wrong, but I believe the chronology is as follows:
> 1. the LZ algorythm was published

In point of fact, several LZ algorithms were published, but yes.

> 2. Unisys developped some changes to this algorithm

One guy at a company that is now part of Unisys made some improvements.

> 3. Unisys patented those changes

His employer applied for patents, with an emphasis on use in hardware.

> 4. Unisys published those changees (*after* having patented them) under
> the name LZW

I'm not entirely sure that 3 came before 4.  Also, in terms of print
publications, it may be (I'm speculating here) that it was submitted to
some journal for review at the same time that the patent paperwork was
submitted, making the two effectively simultaneous.  I don't think the
publication made mention of any patent application.

> 5. the LZW method became very succesful (espescially in gif) and Unisys
> let it develop

CompuServe invented GIF, and they weren't aware of the patent on LZW at
the time.  GIF was out and well known for some time before Unisys spoke
up.  Note that TIFF also uses LZW; I'm not sure of the chronology there.

> 6. now Unisys want to enforce their patent
> 
>         Point 5 is the reason why I think such a patent becomes unenforceable
> in France (besides the problem with patenting an algorithm the exact
> status of which I don't know)

It's similar to the legal notion of need to defend trademarks, then?
 
> >   If I invent a new and revolutionary algorithm for raytracing 10 times
> > faster than any current method and write an article here explaining it,
> > I couldn't patent it here in Finland for two reasons:
> >   1) Algorithms can't be patented, only industrially profitable inventions
> > (mostly real devices) can be patented.

The argument may be that software that implements (reads or writes) LWZ
compression is a software simulation of the hardware it was originally
designed for.  Whether that holds water in France, I have no idea.

>         That's a good thing IMO, but I don't know how international la works in
> that respect...

It probably depends on what treaties the nation in question has signed. 
IANAL, and I haven't seen the treaties.
 
> >   2) Even if they could, I couldn't patent this because I published it
> > in a public forum (even if I deleted it an hour after sending it).
>         Yes but Unisys patented it *before* publishing it...

IMHO, if they wanted patent protection on the algorithm, they should
have noted at least the appication in the original publication.  I don't
know that they did that.
 
> >
> >   There is a case here in Finland where a patent was denied because the
> > applicant had put the specifications of his invention for 1 day in the WWW.
> >
>         That's stupid I think, but it's better than being able to patent
> anything and everything as is the case in some countries :)

Maybe it will all end if some enlightened individual patents
patenting...

-Mark Gordon

Post a reply to this message