[linux-audio-dev] Gigasampler vs Halion PR war

Paul Davis paul at linuxaudiosystems.com
Mon Nov 11 08:16:01 UTC 2002

>Steve Harris wrote:
>>Prior art doesn't seem to be very useful in defending against patents for
>>some reason (IANAL of course). The recent British Telecom v's W3C fiasco
>>was won on the grounds that the BT patent's system operated in a different
>>way to the Web, rather than because of prior art (which there is plenty
>otoh, a patent needs to express a new idea afaik. the technique
>of buffering before playback is applied to 'slow' data providers
>on a regular basis -- all the .mp{1-4}, .ra, .mov, .avi, .swf etc 
>players do exactly the same thing when playing from a http host,
>and they've been doing this for a long time now.
>in this light, claiming the patented idea is either original or 
>new is nonsensical.

this is a misunderstanding of how the patent system works in all
countries that i know anything about. you don't get a patent on an
idea. you get a patent on the application of an idea to a domain. if
you figure out how to take an idea from one domain and use it in
another you can often patent what you've done. the only issue for the
patent office is that there is clear separation of the domains, which
they will usually look for implicitly in terms of the language used.

pre-buffering data from any "slow" source could be patented if you
phrased it that way, and that would be a powerful patent. but the
nemesys/rockwell patent doesn't cover that. it covers specifically
prebuffering data from a disk source with the intent of providing
essentially instantaneous response to commands to produce audio output
(IIRC). it therefore has nothing to do with prebuffering data from a
network source. prior art in the domain of handling data from the
network would not be seen to invalidate a patent that handles data
from a disk.

now, as programmers, we know what the problem with this is: its all
just code, and the mechanisms used to handle one domain ("disk data")
can be identical to those used to handle another ("network data"). the
US patent office doesn't get this at the moment, and they possibly
never will. we know that an OS like linux contains dozens of
algorithms, methods, designs all of which can be applied to many
different area of application software, and that each time this is
done, nothing particularly original is going on. the patent office
doesn't agree with this perspective. they think that
read-ahead-for-instant-start-for-audio is fundamentally different from
read-ahead-for-rapid-throughput. i think they are utterly and
completely wrong about this, and that their attitude demonstrates a
complete misunderstanding of what software is.


More information about the Linux-audio-dev mailing list