|
Authored by: Anonymous on Thursday, June 14 2012 @ 12:10 AM EDT |
That is a terrible idea.
Instead of making up yet another nonsense
definition to distinguish
patent law from everything else, they should just
ask the people who
actually know what mathematics is (leading
mathematicians and
computer scientists, for example) and then adopt that
definition.
Same with everything else like "software" and "machine" and
"algorithm"
and "computer".
Part of the reason we're in this mess now is that
various courts were
unaware of important facts about how computers actually
work, what
capabilities they have, what software is, etc. They got garbled
ideas from
lawyers and then made holdings which range from nonsensically
incorrect
to subtly-and-dangerously incorrect.
Somehow we need to inject
better definitions and proper facts into the
legal system, to cure this malaise
before it spreads any further. We need
legislators and judges to ask the
people who actually know the
facts, because the harm that comes with
letting lawyers just make up
their own incorrect story of how computers or
other technology works, is
intolerable. [ Reply to This | Parent | # ]
|
|
Authored by: Tkilgore on Thursday, June 14 2012 @ 01:00 AM EDT |
And here are some responses from me:
> I would like to see congress step in with real patent reform. Part of that
reform needs to be a definition of math for patents. Do they mean all math
including computation theory, or the lay definition of just formulas. If
congress does not do it, the Supremes will.
Can they be trusted to get it right even if they honestly try? I wonder. Though
I would certainly agree with this hope.
> I posted my math education to show that I have the typical math knowledge
of a
programmer. I agree that most programmers do not care about understanding the
math they use.
That the level of mathematical knowledge is typical, I agree. Your second
statement is probably true, too, not only about programmers but about too many
engineers as well. However, these statements most definitely do not apply to the
people who are at the top of the two professions. Some of them in fact publish
good quality research in mathematics journals. Others would do well to emulate
them.
I will add something else to this. I have worked for most of my professional
career as a mathematician. I was not interested in programming. I learned how to
program in C because I wanted to write a camera driver -- when I was 62 years
old, in 2003. But I will definitely tell you that my background as a
mathematician made it easy to learn how to write code. To that extent, the two
fields of mathematics and computer programming really are closely related.
As to the patent on conversion of Bayer sensor data to an RGB image, I would say
the following:
The problem is a mathematical problem, in its essence. One needs to reconstruct
an image from incomplete data. There are lots of ways to try to do this. But the
problem is at bottom a mathematical problem, and it has been a valid problem for
academic research among mathematicians, engineers specializing in digital
imaging, and others, for years. I think it is very sad indeed if, in a situation
like that, areas of legitimate research are closed off by patents. Because when
a patent is issued for any technique whatsoever which is used for that, it
closes off an area of research and for that matter an area of teaching the
existing state of the art to the next generation. In other words, the patent
office has just stepped in and taken away something from the academic
researchers and the educators, and by doing so it has short-circuited the
alleged benefit of a patent which is, to "teach."
Luckily, the method of Bayer demosaicing which I used in my algorithm must
almost certainly not resemble yours (though I dare not go and read your
patent!). My algorithm was, in fact, a practical simplification of the Adaptive
Homogeneity Demosaicing (AHD) algorithm. Unlike what you said about your code,
it is an absolute necessity to make choices (if-then statements) at every step.
Namely, one does the interpolation vertically and horizontally, and then at each
pixel one tries to determine which way was better. OTOH, even in the simplified
"practical" version the AHD algorithm seems to work quite well and
produces sharp images even from low resolution cameras. It gets rid of the
"zipper effect" quite well. It probably tends to mitigate the moire
effect, too, though that was not an explicit objective. Or it could make it
worse because it produces in general a much sharper image. Who knows? I would
say, having checked up on Wikipedia's description of the moire effect, that it
too raises a mathematical problem and a very interesting one. The whole
explanation of what the problem is, is a mathematical explanation, in fact.
Again, I say that how unfortunate it is that nobody can give that as a problem
or project to an applied mathematics student or an engineering student because
nobody dares go near that patent.
> Yes, I think the solution in both example patents is obvious after finding
it.
I think that is true for almost any patent (ever see a drawing of variable
valve
timing?).
I think that by "variable valve timing" you mean with respect to the
valves on an internal combustion engine? Well, I think it must have been about
1956 when some of us high school friends were sitting around talking and one of
them mentioned the idea of what he called a "3 dimensional cam." When
asked what he meant, he explained that a car engine would run better if it could
open the valves wider at some speeds than others, and it would need to do this
differently under load or acceleration, too. He proposed that the camshaft
should have cams on it which were wide and whose shape would be changed from one
side of the cam to the other, and that the camshaft should be moved lengthwise
to adjust the behavior of the valves. The last time I bought a car, in 2008,
something like that was an advertised feature, touted as an improvement in
efficiency and economy. No doubt it is patented, but it isn't really new is it,
if a couple of high school students back in the 1950s came up with that while
shooting the bull?
> Once you know how to do something it is easy.
There is that about the situation, too, of course. And that is why we are
bedeviled with a patent system. It's the old story of Christopher Columbus and
the egg. Even though it is also true that Columbus made his big discovery
because he was relying on a calculation of the circumference of the earth which
had been written by an Arab astronomer in Baghdad, and someone had applied the
wrong conversion factor when translating that work into a European language. The
calculation was actually pretty good, but the erroneous conversion factor
reduced the earth's circumference to about 17,000 miles. Columbus was a very
lucky man; if there had been no land to hit then he would have started to sail
across the wide Pacific just about the time he would have run out of provisions.
The above is one of those amazing things you can learn on Groklaw, by the way,
and I threw it in free of charge.
[ Reply to This | Parent | # ]
|
|
|
|
|