Re: How to spoil ideas for potential patentees?

[telford]

Lasse Kliemann <stu33404@mail.uni-kiel.de> wrote:
> Greetings,

> according to [1], which was in response to a question I asked[2] some
> time ago in this group, publishing an idea can prevent others from
> getting patents on it. However, derived ideas might still be in
> danger[3].

I've been asking along similar lines for some time and haven't yet 
found a reliable method of forcing an idea to become unpatentable.
The real problem is that the patent offices don't check prior art
very well and you don't know where they will or won't bother looking.

In Australia (and other places) you can buy a provisional patent for
$100 or so and it expires in one year. I believe that a deliberately
lapsed provisional patent will establish a priority date and prevent
the same idea being patented (has this ever been tested? someone
correct me if this is wrong). At least this is something that the
patent office will probably find since they usually search earlier
patents.

There are also websites getting established in order to build databases
of prior art for defensive purposes. These are all pretty new and I
doubt anyone knows what legal standing they have but in the long run
such databases are probably the best thing for the industry:

    http://www.ip.com/priorArtDatabase.jsp
    http://www.priorartdatabase.com/
    http://www.spi.org/freedbasics.htm
    http://www.nongnu.org/padb/

Also, some commercial offerings include defensive disclosure databases
but you have to subscribe to the service to be able to search or
publish (e.g. http://stneasy.fiz-karlsruhe.de) -- this still will be
cheaper than fighting someone who patents your ideas out from under you.

This is a commercial disclosure publishing service with a one-off
publication fee that guarantees a priority date:

    http://www.researchdisclosure.com/overview/index.html

> - Many people should read the idea as soon as possible, so they later
>   can testify that the idea has been published at or before a certain
>   date.
>
> - The idea should be unmistakeably linked with the author. Maybe use
>   some cryptographic signatures for this.

Many technical articles get published here:

    http://citeseer.ist.psu.edu/

And it keeps datestamps plus a local cache of the article.
It's a totally cool place for all types of publication and searching.
Citeseer knows how to read PDF files and PS.GZ files and for most
common article formats it figures out where the abstract is and who the
author is (just make sure your layout is something similar to a common
journal layout).

> - The published document should be signed by another party (or several
>   other parties) as well. These parties thereby would testify the
>   publications's submission to the public on the given date.

This one is difficult unless the database is deliberately designed
for the purpose of defensive publication.

> - Best, it should be some place which patent lawyers come across
>   naturally when doing their investigations. 

This is the big stumbling block because prior art searches are
notoriously bad and the courts can't be bothered cracking down on
patents that make little or no effort to list prior art. If courts
were a little less one-sided and showed a willingness to force the
patent application to really include a prior-art search then patents
would be a lot less dangerous. All it would take is for a few high
profile patents to be deemed unenforcable because important prior art
had been omitted from the application. At the moment, courts don't
bother enforcing prior art unless the evidence is huge (and even then
they may not).

> - Others from the free software community should read the publications
>   and make suggestions on how a derived idea might look like. Then,
>   they publish the derivation as well.

Which is a major job to do for not much return, most of us want to
produce real work, not pointless speculative articles. As has been pointed
out before, the barrier for "inventive" is set ridiculously low making it
very difficult to cover all the minor variations on a theme. A prior
art database can only do so much, we still have to push for law reform
at least enough to demand that new patents really have an inventive step.

> In the moment, I publish most of my ideas on my website[4], in the
> hope that many people come across it and that Google caches the site
> somewhere.

Don't forget that http://www.archive.org/ will keep a copy and a datestamp
every time you change your page. Every bit of evidence helps.

> If I write some software as a result of an idea, I publish
> a pointer on the Freshmeat application index. (The software is
> published under the terms of the GPL of course.)

Freshmeat is handy because it sends out updates as bulletins to a large
number of places, each of which get cached. However, these bulletins
don't contain the full content, only a summary so probably including a
package md5sum in your summary is a good idea.

> But I am afraid, this is not enough. I still hope that Europe will
> stay free of software patents. But if not, I would like to know some
> place where I can publish, and where I can point to in case someone
> claims a patent on my idea at a later time. I also would like to
> encourage others to publish their ideas and ideas derived from mine,
> in order to make more and more ideas unusable by potential patentees.

OK, here's a few more sites where you can publish articles (these are
not specifically for patent disclosure but intended as a general
publication system). Some of them you have to be a registered user to
be able to publish, getting registered usually isn't too difficult:

    http://cogprints.ecs.soton.ac.uk/
    http://dlc.dlib.indiana.edu/
    http://www.cstc.org/
    http://www.publiclibraryofscience.org/  (http://www.plos.org/)

There's also a registry of EPrints users, many of which will allow
publication from Joe Public:

    http://archives.eprints.org/eprints.php

I'll also point out that USENIX publishes lots of comp-sci papers and
provides open-to-the-public access to anything more than a year old:

    http://www.usenix.org/publications/library/

Naturally, you have to join to be able to publish (and publication is slow
because of peer review) but still much cheaper than fighting a court case.


        - Tel