Re: Sort method patents and standard

From: Tony Harminc (
Date: Mon Apr 22 1996 - 16:15:15 EDT

Posted on 21 Apr 1996 at 12:00:44 by Alain LaBont/e'/

Ah - I was hoping for a response from you, Alain.

>in fact for this sorting/searching feature, I have seen one US patent of IBM
>and it has been embarrassing for some IBM people as my name is not mentioned
>but I am the inventor of what they claim... they acknowledged it (they could

I have found the following US patents assigned to IBM:

5070456 Method for Facilitating the Sorting of National Language Keys
         in a Data Processing System

5077669 Method for Quasi-Key Search within a National Language Support
         (NLS) Data Processing System

5072386 Method for Culturally Predictable Keysort within a National
         Language Support (NLS) Data Processing System

5231581 Method for Culturally Predictable Keysort within a National
         Language Support (NLS) Data Processing System

5195034 Method for Quasi-Key Search within a National Language Support
         (NLS) Data Processing System

All of these list the "inventors" as Denis Garneau [formerly of the
NLTC in Toronto] and Wen-Hsiu Sears. In my technical opinion, all
of them cover essentially the same ground, with a few bells & whistles
to differentiate them. Further, the ground covered is exactly that
published in the National Language Design Guide, Volume 2 (SE09-8002)
and the "Redbook" Keys to Sort and Search for Culturally Expected Results
GG24-3516. Both these publications are dated *after* all the above
patents. As you know, Alain, the NLDG does acknowledge your contribution
and that of Denis Garneau. (Do you know the whereabouts of Denis these
days ? I haven't seen him since just after he left IBM several years ago.)

>not do otherwise because I had a copyright on that; however because the
>patents laws are not the same in Canada and in the USA, a patent could not
>be obtained in Canada [once you have talked about it, it's finished here,
>while in the United States you have 1 year to get your patent even after you
>made the idea public]) and told me they patented the thing to avoid another
>company doing it and charging them royalties (this already happened to them,
>apparently, as one company patented something that had come out of IBM
>research labs!)...
>They also told me their intent was to give free access to that patent to
>anybody... they did it for their own protection only...

If that is the case, it would save much time and effort all round if
IBM would simply publish a statement to this effect. Somehow I don't
think it is the case, though I would like to be proved wrong. IBM holds
hundreds of software patents, and browsing through any of them is enough
to make one want to join the League for Programming Freedom immediately.
Many of them cover material that is blindingly obvious (e.g. US4615002
patents the concept of calling a message issuing routine and passing it
both a message ID and a language ID, and having the routine issue the
message in the right language! Hard to believe, but the US patent
office evidently thought this was a non-obvious idea in 1986.

>If it is the case, then fine, I agree that this work be in the public
>domain, and that is what my copyright said as early as 1986 (final version
>of my method in 1988, followed by a reduction technique in 1989)...
>I also know for sure that a patent was registered by Kulinek and Lee
>(although I have not seen the actual patent), from Bell Northern Research,
>around 1988 also (they did their work in parallel with me, we had not met
>nor were we mutually aware of our respective works then; around 1988 we
>began to work in common on a Canadian standard about the issue);

I did not find any US references to such a patent, but I'm just an
amateur net searcher, not a patent professional.

Alain, do you believe that your method covers essentially *all* of what is
mentioned in the NLDG Volume 2 ? If so, and you are making it public
domain, then it would seem that anyone can use these methods, at least
everywhere outside the US. And it seems to me that the US is the one
country where no one cares about this stuff anyway. (Please - no insult
intended to the USers on this list - what I mean is that there's essentially
no commercial demand for this stuff in the US.) On the other hand, there is
mention of at least one European Patent on some of these techniques. But
if you published the material before the EP application, then it is surely
invalid, since only the US has the "date of invention" scheme for patents.

Questions, questions... Perhaps I should just write to IBM's Director
of Licensing and ask them for permission to use the patents. Or maybe
get the lawyer to write... :-(

Tony Harminc

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:31 EDT