are algorithms patentable?

[uunet!cwi.nl!Dik.Winter]

I take exception to this message, it misrepresents quite a few things.
Although I am in favour of the LPF, this is not the way to get at their
goals; it is unfair to Unisys.

1.  I do not know whether LZW has been published in CACM, but if it has
    this is not contradictory to it being patented.
2.  It has been known a long time that LZW was patented.  Actually there
    have been a lot of messages about this some 10 years ago on the net
    because the well-known Unix utility compress uses LZW.
3.  The current Unisys position is not different from its position some
    10 years ago.  Commercial use should pay a license, non-commercial
    use would be free (so compress would be off-the-hook as it came
    from BSD Unix).  This was posted on the net by people from Unisys
    back at that time (although I did not save the message at that time
    I think) making explicit that compress would be free.  Perhaps
    somebody from AT&T did save the message.  The stance is that
    developers of commercial software using LZW should license.
    Users never should pay.
4.  Laying the burden on Unisys is putting it at the wrong hands.  Look
    for the patent office.
5.  If you have talked with Unisys you should also talk with IBM who hold
    a patent for exactly the same algorithm (although that one is from a
    later date).

 >                                                              My understanding
 > of the position of Unisys and Compuserve in this matter is that any
 > program that encodes or decodes graphics in GIF format is an infringement
 > of the Unisys patent on the compression algorithm, 

This is completely wrong.  GIF in and off itself is no problem.  The use
of LZW might, but only if the program is commercial.

 > Mathematical software people, who have long relied on building upon each
 > other's work placed in the public domain, may well find this an alarming
 > turn of events.    Nor should this be considered a strictly American
 > phenomenon; one of the arguments advanced against ratifying GATT in the
 > American technical community is that it requires conforming American
 > intellectual property laws with those of the other treaty members,
 > which in many cases are said to be even worse-conceived.   

The problem is not so much with the laws as well with the outcome.  Actually
the problem is the US patent office.

 > numeric-interest is not the place for an extended discussion, but we all need
 > to be aware of this issue.

Actually, it is.  With LZW and RSA patented we might see patents in any
field.  Fast quad precision floating-point?

Now some actual misrepresentations:
 > > Until now, most computer professionals and companies have ignored the
 > > problem of software patents.  The GIF format for graphical images was
 > > adopted widely on the net, despite the Unisys patent covering the LZW
 > > data compression algorithm.  The patent dates to 1985, but its
 > > enforcement has been carried out with private threats; most victims
 > > are afraid to talk about it.  Now the patent has shown its teeth.
Not so.  It was well known that compressing-modem manufacturers licensed
from both Unisys and IBM for LZW.  This was no secret at all.
 > > 
 > > For a few days, the Internet community was shaking with anger at the
 > > surprise demand to pay license fees for the use of GIF format.
There is no license fee demand for the use of GIF format, see above.

I leave the remainder.

dik
--
dik t. winter, cwi, kruislaan 413, 1098 sj  amsterdam, nederland, +31205924098
home: bovenover 215, 1025 jn  amsterdam, nederland; e-mail: dikacwi.nl