Terry Harris wrote:

> On Thu, 11 Jun 2009 15:10:57 -0400, you wrote:
>
> >Let me give one example (out of many) of a particularly bad patent that was issued that affects some embedded systems.  US Patent 6233595 (issued in 2001) Fast multiplication of floating point values and integer powers of two. It covers the very narrow
> >case of multiplying a floating point number by an integer which is a power of two. Floating point numbers are defined as ssssssssss * 2 ^eee if you multiply this by 4 for example the result will be ssssssssss * 2 ^(eee+2). The claims (all 16) are based
> >around the significant not being altered in the process and only integer math is needed on the exponent. The case of multiplying a float by 4.0 is not claimed although it has the same solution.
>
> Disregarding the prior art what kind of claimed value could be justified
> for this invention?

I think it was a pre-emptive patent focused on a specific application area.
The patent is on the web in multiple places. The value is with a limited
set of numbers it is computationally less intensive than a full float mult.
(8 bit add vs a 24*24 bit multiply)

Not surprising this is the kind of optimization that compilers do in
support for float on small processors. We have been doing this for
years and internally check the floating point number for
+/- . . . . 0.125,0.25,0.5,0.0,1.0,2.0,4.0,8.0,16.0 . . .
as well as some other magic numbers that have unique solutions. The
patent covers a subset of those values in integer representation. I am
not sure why it was so specifically a float * integer that  was covered

> How many hours (minutes) could someone spend inventing this?
>
> The claimed value wouldn't cover the cost of filing and even if it was
> filed you or anyone else could afford to buy full rights and recover some
> money by licensing or selling them again (which would effectively force the
> original holder to license or sell restricted rights for much less).

I know of many examples where valuation might not be a good measure
and valuation might be troublesome. Crafting the rules for a financial
incentive may be very difficult. If for any other reason it might  force
patents to not be filed and never be publicly disclosed. The biggest
patent complaint I have seen is the failure to identify prior or obvious art.

A second reason cost of development has little relationship to value. Many
unique eureka moments opening up whole new technology and approaches
are worth a lot and they too should be rewarded. Karatsuba multiplies (which
I don't think is patented) and Morvin Gentlemens FFT which is, are examples
of break through ways of thinking about problems.


-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist