On 09/27/2011 11:50 PM, Hsuan-Yeh Chang wrote:
In reply to Mark's points, I would note the following:
1) Algorithm by itself is not patentable. To get a patent from a new
algorithm, the new algorithm must be combined with or used in a
patentable method or product.
While technically true in a pedantic sense, it is largely not true in
practice. Many software patents describe an algorithm in the context of
an application or process, but outside of the application or process the
algorithm has little or no application. So, in essence, as lay people,
we are speaking of algorithms.
2) If A made an idea available to the public (by, e.g., journal
publication) before B submitted a patent application in the Patent
Office, under AIA (i.e., first-to-file), A's publication would readily
hunt down B's patent. Under the old law (first-to-invent), however, B
would be able to establish evidence (if there is indeed evidence) and
"swear behind" A's publication. That is, the old patent law allows B
to assert a much earlier priority (invention) date. This has caused a
lot of trouble and created a lot of jobs for B's lawyer, trying to
establish evidence to safe guard B's patent right.
This paragraph isn't very clear and hard to parse. I don't understand
the sentence "A's publication would readily hunt down B's patent." Do I
understand that you concur that publication prior to the patent process
would invalidate the patent under the AIA? As that is how I think I
understand it.
3) Information on Github or Sourceforge (or anything posted on the
internet) can be considered as publication and thus may be used as
"prior art" against an issued patent. The trouble is, however, that
the things posted on Github are lines and lines of source codes. It
is difficult, if at all possible, to extract the design concept and
idea from thousands even millions of lines of source codes. So, even
if Gibhub may be used as a source of prior art, it would take a huge
amount of effort to apply those "prior art" against the claims of
issued patents.
That, so to speak, should not be our problem. As it is, it is difficult
to find prior art because of all the publication sources already. You
have to do it anyway. I'm speaking of course as someone against software
patents. A few more sources don't matter.
The real problem is the lack of uniformity of describing patents
themselves. By definition, software patents are intentionally worded in
overly broad terms. Engineers are "encouraged" by the legal department
to overly state the range of the "invention" to give the patent the
widest scope.
4) Other web postings are also troublesome, because people cannot
confidently rely on the posting date, and thus hard to make sure
whether such postings constitute as "prior art."
The time stamp is very important.
5) I believe that the open source community should at least try to
have some better understanding of what patents really are, and what
public interests do the patent system serve. As I said previously,
patents and open source can coexist peacefully. Ultimately, this
business world is operated under certain sets of laws and rules, and
only those who knows the laws and rules well can play a better game...
What we need is a evidenciary quality priorart.org repository where open
source software is stored, made public, and has a legally verifiable
submission date.
HYC
------------------------------------------------------------------------
*From:* "[email protected]" <[email protected]>
*To:* Richard Pieri <[email protected]>
*Cc:* [email protected]
*Sent:* Tuesday, September 27, 2011 9:07 PM
*Subject:* Re: [Discuss] The America Invents Act
There are two things that are needed to improve patents for open source.
(1) A codification that publication prior to patent submission of
algorithms eliminate patentability. I think the AIA explicitly states this
as an exception of first to file.
(2) A recognition that "publication" means publicly available and not only
through noted journals. Sourceforge, github, etc should serve to provide
prior art.
> I favor a return to the model requirement. When one submits a
patent for
> an invention one must include a functioning model of the invention.
>
> --Rich P.
>
>
> _______________________________________________
> Discuss mailing list
> [email protected] <mailto:[email protected]>
> http://lists.blu.org/mailman/listinfo/discuss
>
_______________________________________________
Discuss mailing list
[email protected] <mailto:[email protected]>
http://lists.blu.org/mailman/listinfo/discuss
_______________________________________________
Discuss mailing list
[email protected]
http://lists.blu.org/mailman/listinfo/discuss
