|
Authored by: macliam on Saturday, March 30 2013 @ 04:32 PM EDT |
Glad you found the State Street screed informative.
I might have gone
on to make some general points, but in the past hour, thanks to Google
Scholar I have come across some cases that were new to me.
I don't
know whether you have come across
Dann v. Johnston 425 U.S. 219.
Searching for Groklaw articles mentioning this case, I find three. The
first is in an interesti
ng article by Prof. Michael Risch which I must re-read.
The second is as a
cited case mentioned in footnote 19 of the
Government brief in Bilski v. Kappos. And the case was mentioned in
the
oral argument in Bilski v. Kappos. On my reading of Dann v. Johnston, it must surely be
relevant if not ruling precedent for CLS v. Alice.
Here is the opinion of
a unanimous Supreme Court:
Respondent has applied for a patent
on what is described in his patent application as a “machine system for
automatic record-keeping of bank checks and deposits.” The system permits
a bank to furnish a customer with subtotals of various categories of
transactions completed in connection with the customer's single account, thus
saving the customer the time and/or expense of conducting this bookkeeping
himself. As respondent has noted, the “invention is being sold as a
computer program to banks and to other data processing companies so that they
can perform these data processing services for depositors.” Brief for
Respondent 19A; Application of Johnston, 502 F. 2d 765 (CCPA
1974).
Petitioner and respondent, as well as various amici, have
presented lengthy arguments addressed to the question of the general
patentability of computer programs. Cf. Gottschalk v. Benson, 409 U. S. 63
(1972). We find no need to treat that question in this case, however, because we
conclude that in any event respondent's system is unpatentable on grounds of
obviousness. 35 U. S. C. ยง 103. Since the United States Court of Customs and
Patent Appeals (CCPA) found respondent's system to be patentable, Application of
Johnston, supra, the decision of that court is accordingly reversed.
Dann v. Johnston,
425 U.S. 220
(1976).
Whilst I did not spot any neat quotes in the
‘obviousness’ argument under § 103, the broad thrust is that
the accounting practices were well-understood, also computerized financial
information systems were well-understood in the art back in 1976 and
before, so the PHOSITA should have had no difficulty in automating a routine
accounting system.
______
The other case seems to me really
weird. This is a case,
Application of Chatfield before the
Court of Customs and Patent Appeals. 545 F.2d 152 (1976).
I arrived at it
through following a cryptic reference in footnote 11 of Benson v. Flook.
Apparently the PTO did not want to allow patents for ‘software’ and
rejected the application. The matter came before the CCPA. Three judges, Chief
Judge Markey and Judges Baldwin and Miller reversed the decision of the PTO.
They found (sensibly in my opinion) that there was nothing in Benson to
preclude patentability in the case before them. Two judges, Judges Rich and
Lane dissented.
After the CCPA ruled that the invention was patentable,
the director of the PTO applied to the Supreme Court for certiorari.
Certiorari was denied. And the patent, 4,183,083 duly issued. As
patents go (I am no great fan of patents), this seems a perfectly respectable
patent whose subject matter is an improvement to scheduling procedures for
multiprogrammed operating systems. It is full of technical details about
priorities, peripheral access queues, thruput, resource priorities, some
relationships expressed through simple mathematical equations, constraints
expressed through inequalities, use of the simplex algorithm
etc.
In a bizarre dissent, Judge Rich, joined by Judge Lane wrote a long
screed which offered no reason to deny patentablity other than his apparently
incomprehension of the rationale of Gottschalk v. Benson. He included
the following questions:
That brief observation adds
considerably to the already existing ambiguity inherent in the Benson opinion.
Does it say Benson's two claims were held non-statutory only because they were
not limited to any particular technology, apparatus, or end use?
Does
the Supreme Court not regard data processing as a technology, data-processing
equipment as apparatus, or the necessary conversion in such apparatus of
binary-coded-decimal to pure binary signals as an end use?
The answers
to these questions are very important because in the instant case, and in In re
Noll, the invention relates only to the operation of data-processing equipment
(‘computers’) without reference to any other
‘technology,’ ‘apparatus,’ or ‘end use.’
How, then, can this case be distinguished from Benson even when the narrowest
possible view of that decision is taken? Just how ‘limited’ was the
holding in Benson supposed to be? Benson's claims were in fact limited to a
method carried out in data-processing apparatus (including a ‘reentrant
shift register,’ claim 8) and to ‘A data processing method’
(claim 13). Benson did not claim a ‘formula.’
Application of Chatfield before the
Court of Customs and Patent Appeals. 545 F.2d 161
(1976).
Judge Giles S. Rich may have been distinguished as a
patent lawyer, and probably rightly so. But I am getting the impression that he
as a man who lacked a basic understanding of mathematics, and who could not
understand issues relating to computer programming, operating systems etc. And
this basic lack of understanding shows through in his judicial
opinions.
After writing the above, I just found the CCPA decision on
Flook overruled by SCOTUS here: Application of Flook
(found on website www.leagle.com. Interesting! According to
the patent judges on the CCPA, calculating a value (say) 40% of the way between
x and y and then setting an alarm limit to that value is not
pre-emptive, and therefore is allowable under Benson, because you are not
precluded from calculating that value and not updating the alarm
limit.
The message I am getting from the patent judges is this:
scientists, mathematicians and computer programmers can play around with their
laws of nature, natural phenomena, abstract ideas, equations, theorems, abstract
and the like, so long as they are not doing anything useful. But every
form of useful human activity is patent-eligible under the 1952 Act. [ Reply to This | Parent | # ]
|
|
Authored by: macliam on Saturday, March 30 2013 @ 09:57 PM EDT |
Applic
ation of Bernhart (aka In Re Bernhart) is one of the cases that
you looked into, I believe, from the ‘programmed computer is a special
machine’ point of view.
Did you ever get round to looking at the
issued patent,
US3,519,997?
If you get round to looking at it, in the more readable
parts in the pages before the claims start, and on to the claims which
are in this case reasonably readable, do you see anything looking
remotely like computer programming, or technical detail? Do you see anything
that looks like anything other than pre-calculus analytic
geometry?
Admittedly it is rather odd to read a piece of otherwise
straightforward mathematical text that talks of signals
(xi, yi,
zi) rather than coordinates
(xi, yi,
zi).
But, in my view, In Re Bernhart is
a story, not about dicta concerning programmed computers being particular
machines, but a completely different story involving the weird rules governing
the determination of ‘nonobviousness’ in the days before KSR v.
Teleflex.
In the case In Re Bernhart under consideration, which
involved using a computer to calculate a projection from three dimensions to two
using basic pre-calculus coordinate geometry then plotting it, there was one
prior art reference, Tripp that taught the combination of the computer
and the plotter, but the main prior art was Taylor which also taught
using the computer to compute the same projection from three dimensions onto a
plane. But because the mathematics needed to transform from one two-dimensional
representation to another was considered beyond the skill of the PHOSITA, the
invention was held to be nonobvious under section 103: “There is nothing
to suggest that, within the context of automated drawing, one of ordinary
mathematical skill armed with the Taylor reference would be able to discover the
simpler equations which are the basis of the claimed programming.”
Applic
ation of Bernhart,
417 F.2d 1402.
Of course, the prior art would
of course be more likely to be found, not in the PTO databases, but in dusty
19th century geometry textbooks for schools and universities that people would
not normally consider taking down off the shelves, since that sort of
mathematics would only be of interest to a handful of specialists in the history
of mathematics with a particular interest in 19th century pedagogy and
geometry.
[ Reply to This | Parent | # ]
|
|
Authored by: macliam on Sunday, March 31 2013 @ 10:16 AM EDT |
This is the problem. A patent US 3,519,997 issued in July
7, 1970. The application was filed in 1961, following a successful appeal at
the Court of Customs and Patent Appeals Application of Bernhart (also known as
In Re Bernhart).
The Patent, entitled Planar Illustration
Method and Apparatus, concerns a projection and plotting
program.
Three-dimensional coordinate data (x,y,z) is
projected into two dimensions (v, w) in a fashion to be described
below in more detail. The co-inventors were Walter D. Bernhart and William A.
Fetter. (The patent was issued to Fetter.) Given a sequence of points in
three-dimensions, specified by their Cartesian coordinates
(xi, yi, zi), and
information to determine whether to draw a line segment between succeeding
vertices, the projection onto two-dimensions is calculated by the computer, and
the result is plotted using a standard plotting device. There was one piece of
prior art (US patent
3153224, issued to M. Taylor) that taught calculating a projection from
three dimensions to two with the same specification as that in Bernhart and
Fetter's application. There was another piece of prior art, the Tripp
reference, that taught using a plotter in conjunction with a computer. Thus the
claimed invention would have been ‘obvious’ under § 103 of the
Patent Statute over the combination of Taylor and Tripp, but for one missing
piece of the jigsaw:
The equations specifying the
two-dimensional co-ordinates (v,w) in terms of the three
dimensional
coordinates (x, y, z) set out in the claims of
Bernhart and Fetter were not to be found verbatim in the Taylor
reference.
Therefore the learned judges of the Court of
Customs and Patent Appeals ruled that the invention was nonobvious and therefore
patentable. The opinion was unanimous
To summarize: the claimed
invention would have been obvious over a combination of Tripp, Taylor and a
coordinate geometry problem. The judges ruled that the solution of the problem
would not have been obvious to the PHOSITA (Person Having Ordinary Skill In The
Art), though a similar transformation was present in the Taylor
reference.
It should also be noted that the equations in Bernhart and
Fetter seem to have been plucked out of thin air. The inventors just seem to
describe how to break down the task of calculating using the equations into
stages with intermediate quantities, so that the PHOSITA could program the steps
into the computer. They do not teach how their equations were arrived at. They
might have been (and I suspect probably were) just plucked from some
textbook.
Rays meeting at the eye are projected onto a plane P.
The origin has coordinates (0,0,0). I will denote the coordinates of the eye
position by (a,b,c) (departing from the notation of the
patent for ease of typing, so as not to have to bother with subscripts etc.).
Rays passing through the eye position are projected to points of the plane
P with equation ax + yb + cz = 0 so that the ray is
projected to the point of the plane P where it cuts the plane. Moreover
this plane is the unique plane passing through the origin that is perpendicular
to the line joining the origin to the eye position at (a, b,
c).
Now the points of the ray passing through (a,
b, c) and (x, y, z) have coordinates of the
form
( (1-t)a + tx,
(1-t)b + ty,
(1-t)c + tz
)
where t is a real variable. Therefore the point
(p, q, r) at which the line cuts the plane P is of
the above form, but with the value of t determined to satisfy the
equation
(1-t)a2 + tax
+
(1-t)b2 + tby
+
(1-t)c2 + tcz =
0.
Therefore
t (a2 +
b2
+ c2
- ax - by -
cz)
= a2 + b2
+
c2.
Thus
p
=(1-t)a + tx,
q =(1-t)b +
ty,
r =(1-t)c +
tz,
where
t = (a2 +
b2
+ c2)
(a2 +
b2
+ c2
- ax - by -
cz)-1.
Note that, once the problem is
specified as determining the coordinates (p, q, r) of the
point where the line passing through the points (a, b, c)
and (x, y, z) cuts the plane P, the solution to that
problem is uniquely determined. The solution is compelled by the standard
properties of plane geometry, and by the laws of logic that underpin
mathematical reasoning. It is then necessary to assign coordinates (v,
w) to points in that plane. Analysis of the formula in Claim 1 of the
Fetter patent shows that the ‘inventors’ used an orthonormal
coordinate system with two coordinate axes at right angles passing through the
origin. Moreover the coordinate axis corresponding to the coordinate v
was chosen to be horizontal. This ensures that the other coordinate w is
proportional to the vertical height of the point within the plane P. The
coordinates v and w are in fact constant multiples of the
orthonormal coordinates with constant of proportionality K. Thus the
formulae in the patent of Bernhart and Fetter are uniquely determined by the
requirement that rays passing through the eye at (a, b, c)
are projected to points of the plane P defined above, and the projected
points are then represented in Cartesian coordinates determined by coordinate
axes at right angles, with one axis horizontal, and where the resulting
orthonormal coordinates are multiplied by some given constant K. (I
don't write out the formulae here because they involve fractions, square roots
and plenty of subscripts, and are therefore not suitable for reproduction in
plain HTML.) And had we the requirement that one coordinate axis in the plane
be horizontal, specification of an angle would have determined the orientation
of the chosen axes in the plane. Thus all possible ways of determining
two-dimensional projected coordinates (v,w) from three-dimensional
Cartesian coordinates (a, b, c) consistent with the
requirements of the problem that Bernhart and Fetter were investigating are
determined by two real constants: a constant of proportionality K, and an
angle θ: it is possible to obtain other solutions from the given solutions
by replacing v and w by
v cos θ +
w sin θ
and
-v sin θ + w cos
θ.
where the angle θ is chosen
suitably.
Of course the patent was not in this mathematical method
itself. But patent judges, textbook writers and lawyers in the Giles Rich
School have regarded any application of a mathematical idea as being patentable,
provided that the requirements of sections 102 and 103 of the Statute are met.
And they have castigated Supreme Court Justices, accusing them of confusion and
incoherence, for venturing to depart from this basic principle in Benson
and Flook.
What the patent was on, of course, was the physical
system consisting of the computer programmed to perform the calculations
described, and plot the results on a standard plotter. The involvement of
(programmed) machines is more than enough (in the opinion of the judges
of the Giles Rich School) to satisfy the requirements of 101. And, in the case
of Bernhard and Fetter's projection and plotting application, the requirements
of sections 102 and 103 (novelty and nonobviousness) were also clearly
satisfied, because there was no prior art reference describing exactly what
Bernhart and Fetter did, and it would be unreasonable to expect a PHOSITA in the
art of automated drawing to solve coordinate geometry problems.
But what
about programming details. Surely there must be complex programming involved.
Well, at certain points the specification, there are statements to the effect
that the program contains subroutines to compute this or that intermediate
quantity. And there is a general remark;
It is obviously
within the skill of ordinary programmers to program the equations hereinbefore
given into flow charts or diagrams and to translate the latter into computer
subroutines for solution of such equations along with a compatible computer
language for processing input data and instructions to produce an output
directly (or on post processing) useful for controlling a planar
plotter.
Thus practical implementation details are essentially
irrelevant. the PHOSITA can handle those. And this was true back in 1961 when
the patent was filed.
[ Reply to This | Parent | # ]
|
|
|
|
|