[Fsfe-ie] Re: Conversation with Mrs. Doyle last week

[Justin Mason]

[trimmed cc list]

Niall Douglas writes:
> 2. All existing patents registered at the EPO should become null and 
> void though if their holders choose, they may reapply under the new 
> criteria and if successful, they get a new patent in preference to 
> any newer applications.

Oops!  *software* patents, not just patents in general ;)  I would clarify
that...

Apart from that oversight (I would think ;) it's a great letter BTW.
Thanks for the link to the gnu-friends.org article with all those quotes
-- I hadn't seen that before.

The point about most of the EPO's list of software patents belonging to US
multinationals, is a really key point IMO.

> 1. Member States shall ensure that whenever a patent claim names 
> features that imply the use of a computer program, a well-functioning 
> and well documented reference implementation of such a program shall 
> be published as a part of description without any restricting  
> licensing terms
> 
> Reason: You can filter out "patent land grabs" as commonly practised 
> by software multinationals who patent thousands of promising areas of 
> software development and yet further develop only a few. If however 
> they have to write a reference implementation, not only do you 
> strongly encourage the company to further develop the patent but you 
> also substantially lessen the patent being used to prevent further 
> development by competitors.

I'm not sure this is a solution.  Consider these patents -- the
"transforming long filenames" Sun one at the EPO, or the "drawing a pie
chart" MS one at the USPTO.  Both are easily demonstrated in a ref impl,
since they are both fundamental techniques easily written in 3 hours.

Since they are so fundamental they should not be patents -- but the
ref impl requirement doesn't place any impediment in the way.

--j.