Patent law, specifically software patent law, though not as flashy as "cyber-squatting" or cyberspace jurisdictional issues, has become a major part of the cyberspace focus.  Unfortunately, while patent law in general seems to be based on rather settled principles of law, the specialized area of software patents has been a source of difficulty for the United States Patent and Trademark Office (PTO) and the federal courts since the first application for a computer program-related invention.  This paper traces the history of software patent law from the initial per se rejections of all software patents up through the latest judicial decisions on the matter.  Specifically, the focus is an analysis of the relevant federal court cases covering software patent eligibility issues, and the ultimate goal is to provide some conclusions about the state of the law today.

Section 101 of Title 35 in the United States Code explains the fundamental requirements of a valid U.S. patent:

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.1

The language of this section sets forth three vital elements for a patentable invention.  First, the invention must fall into one of the four categories of eligible subject matter:  process, machine, manufacture, or composition.  In addition, the language requires that the invention be both useful and new (or inventive).  Other provisions of Title 35 further divide the inventiveness requirement into the necessary elements of novelty ( 102) and non-obviousness ( 103).2  Finally,  112 sets forth the requirement that the patent application contain an adequate description.3  Hence, the five main elements required for a patentable invention consist of 1) subject matter patentability, 2) usefulness, 3) novelty, 4) non-obviousness, and 5) adequate disclosure.4
 The main focus of the software patentability issue to this point has been the patentable subject matter requirement, although this paper will show that the focus may be shifting to the other patentability requirements.  The main purpose of the patentable subject matter requirement is to create an incentive for inventors to direct their creativity towards those inventions that fall within the description of "useful arts" found in the U.S. Constitution.5  If an invention does not fit within the parameters of one of the four patentable subject matter categories, the PTO will declare the invention unpatentable.
 A patent is a monopoly.  It can only be obtained by filing the requisite patent application, conforming to all requirements and regulations, with the United States Patent and Trademark Office (PTO) in Washington, D.C.  After consideration by the assigned patent examiner, the application may be accepted or rejected, based on whether the examiner believes that the invention meets the five statutory criteria set forth above.  If rejected, the applicant can appeal to the Board of Patent Appeals and Interferences (Board).  There are two avenues of appeal from the Board.  The patentee can file a lawsuit in the U.S. District Court for the District of Columbia against the Commissioner of the PTO under 35 U.S.C.  145.
The much more common course of action, however, is to appeal the Board's decision to the Court of Appeals for the Federal Circuit (CAFC).  The CAFC was specifically established to handle all appeals in patent-related cases, in addition to all customs and government contract disputes.  To that end, it is also the appellate court for all patent-related disputes that originate in federal district courts (which have exclusive subject matter jurisdiction over patent infringement suits).  From the CAFC, the case can be petitioned to the Supreme Court.  The CAFC was established in 1982 to replace the Court of Customs and Patent Appeals (CCPA), which, until then, had been the sole appellate court for patent-related disputes.

The hurdle for software patents:  subject matter patentability
The specific problem with software since the 1960's has been determining which category a software invention falls under.  Software inventions are extremely unique in the context of patentable inventions, and the singular characteristics create problems for the PTO in determining how to characterize these creations.6  A computer program-related invention might be a machine, defined as "a device that has relatively moveable parts and performs a useful operation."7  Or the software invention may fall into the catchall category, article of manufacture, which is "any tangible object, other than a machine or composition of matter, that is man-made and not found in substantially the same form in nature."8  Finally, the software invention might be a process, otherwise defined as "an act or series of acts which produces a desired result."9  Any or all of these three categories may apply to a computer program-related invention.
This problem is exacerbated by a long-standing legal doctrine that laws of nature, natural phenomena, and abstract ideas are unpatentable subject matter.  The statutory origin or justification for the doctrine is the phrase "whoever invents" in  101, which manifests the requirement that any invention must be made by man.10  In other words, anything found in nature that is discovered, not invented, cannot be patented.  The fundamental concern is the chilling effect on inventions if all the basic ideas, natural laws, and scientific principles known to man are monopolized by a few patent holders.  It would be impossible to obtain a patent for any new invention, because those few patentees claiming the rights to the underlying ideas of all new creations would own the rights to all existing and future inventions resulting from the ideas.    Hence, "an idea of itself is not patentable."11
The resulting legal doctrine protecting fundamental ideas and natural phenomena arose early in our legal history.12  Fundamental ideas were rejected in claims to such inventions as the telegraph in O'Reilly v. Morse13, and the telephone in The Telephone Cases.14  Ultimately, the unpatentability of basic ideas became the source of the "mental steps" doctrine created by the Court of Customs and Patent Appeals (CCPA) and applied to reject computer program-related patents.  The court described the doctrine in In re Abrams, stating that "if all the steps of a method claim are purely mental in character, the subject matter thereof is not patentable."15  The PTO applied this doctrine to all computer program-related inventions throughout the 1960's, and the CCPA followed its lead for most of the decade.16
In 1966, a report from the President's Commission On the Patent System recommended that all patent applications for computer program-related inventions be rejected due to a lack of facilities and personnel to process the applications and perform the necessary prior art searches.17  In response, the PTO's 1968 Guidelines (procedural guidelines set forth by the PTO to assist the patent examiners) explicitly stated that computer program-related inventions were to be rejected as unpatentable subject matter.18
As a result, patenting a computer program-related invention in the 1960's was essentially impossible.  The PTO, applying the "mental steps doctrine," declared all such inventions as non-statutory subject matter.  However, the CCPA, in what was to become a common trend in its relationship with PTO, did not agree with the PTO's application of the doctrine and seemed to take a much more software-friendly approach in several cases as the decade ended.  While agreeing that an invention based purely on mental steps was unpatentable, the court concluded in In re Prater (CCPA 1969) that a process capable of being performed without mental steps may be patentable subject matter, even if it is possible to perform it solely through the use of mental steps.19  While not explicitly stating so, the court in Prater placed into question the use of the "mental steps doctrine" for computer program-related applications.20  On a larger scale, the CCPA set the tone for the growing adversarial relationship between, on one side, the CCPA pushing for less stringent patentable subject matter restrictions regarding computer program-related inventions, and on the other the PTO, rejecting most, if not all, software inventions.21
Continuing the struggle against the PTO, the CCPA then adopted a much broader doctrine called the "technological arts" test in In re Musgrave.22  The new doctrine essentially stated that any software invention was patentable subject matter if it was based on process claims which were in the technological arts and not purely mental.23

The Supreme Court siding with the PTO:  Gottschalk v. Benson
Soon after, the United States Supreme Court finally stepped into the contest and decided its first software patent-related decision.  In a 1972 case with far-reaching implications that would influence the treatment and legal analysis of software inventions for years to come, the Court sided with the PTO in its struggle with the CCPA and seemed to adopt an approach strikingly similar to the "mental steps" doctrine.  The case, Gottschalk v. Benson, involved a process invention claiming a method of converting numbers from binary-coded decimal numbers into pure binary numbers (which are used to program a computer) with the use of a general-purpose digital computer.24  The two claims at issue, claims 8 and 13, described a method of converting signals into binary code and a method of converting numbers into binary code numbers.25  The patent examiner at the PTO rejected the two claims, concluding that it consisted of non-statutory subject matter.26  True to form, the CCPA reversed the PTO's rejection, holding that the claims did cover patentable subject matter.27
The Supreme Court affirmed the PTO's position in reversing the CCPA and holding the claims invalid.  In the course of its decision, the court seemed to set forth another new doctrine or test.  The Court first reiterated the previously described, fundamental doctrine that laws of nature, natural phenomena, and abstract ideas are unpatentable subject matter.28  In fact, it emphasized its point by describing several cases supporting this doctrine.29  The Court seemed to equate the invention's fundamental mathematical formula with an unpatentable law of nature, and emphasized that the claimed process could be performed on any computer or by hand.30  One key factor for patentable subject matter in a process claim not tied to a particular machine, according to the Court, was "[t]ransformation and reduction of an article 'to a different state or thing.'"31  In this case, the Court seemed to feel that the claimed invention did not transform anything and was, in fact, an unpatentable mathematical formula.  The claimed invention was "so abstract and sweeping as to cover both known and unknown uses" of the conversion.32  Thus, the Court concluded that the effect of allowing a patent for this invention would be the same as allowing a patent on the fundamental idea behind the invention; it "would be a patent on the algorithm itself."33
The decision seemed to ignore both the CCPA's "technological arts" test and the "mental steps" doctrine relied on by the PTO, even though the new Benson test did seem to rely on the same underlying rationale as the "mental steps" doctrine.  The new test seemed to be an examination of whether the patent claim containing a mathematical algorithm wholly preempted that algorithm.  If so, the invention was too abstract and thus unpatentable subject matter.  It was unclear when a claim would not be wholly preempting the algorithm, but the Court seemed to emphasize physical transformation and specificity as two characteristics of a patentable process claim containing an algorithm.
In general, the Supreme Court's decision in Benson created more questions than it answered and resulted in a state of confusion.  Fundamentally, it became questionable as a result of this decision whether any software patent was statutory subject matter under  101.  Some argued that the result of this case was that computer program-related inventions were per se unpatentable.34  And the term "mathematical algorithm" became the focus of much debate about its meaning and application in the new Benson test.  With all scrutiny aimed at the meaning and application of the term, the PTO and the courts were distracted from the real issue.  In essence, the use of the term as part of the test in Benson redirected the focal point of software patent examination from statutory subject matter to the "correct" definition of "mathematical algorithm."  Evidence of the negative distraction was exhibited in the new direction of the ongoing debate between the PTO and the CCPA, which began to center on determining the definition and scope of the new "mathematical algorithm" test.35
In fact, the debate surrounding the term "mathematical algorithm" went even further than that.  It was and still is argued that the term was used inaccurately by the Supreme Court and thus resulted in even more confusion about the holding of the case.36  In Benson, the Court defined "algorithm" as "[a] procedure for solving a given type of mathematical problem."37  Ultimately, the argument is that the general scientific definition of algorithm covers a lot more area than the Supreme Court's definition, including all software inventions.38  In fact, it is difficult to distinguish this scientific definition of "algorithm" from a patentable process.  Thus, the argument goes, the Supreme Court erred in using the term "algorithm" in a more narrow sense to mean only mathematical formulas.
The argument, however, seems somewhat extreme.  While technically accurate in concluding that the Supreme Court probably used the term "algorithm" in a misleading manner, it likely did not lead to any more confusion than that already produced by the substantive holding of the case.  In other words, even if the Supreme Court had used the term in exactly the same manner with exactly the same definition as that used by most academics and scientists, the case would likely have produced just as much confusion.  Moreover, the courts in later cases such as Parker v. Flook39 (discussed below) made a point to specifically define the term as it was to be used in that case, usually citing back to the definition in Benson.  In fact, it wasn't long before references to the test included the modifier "mathematical" in front of "algorithm," indicating an understanding that "algorithm" can go beyond mathematical formulas.
Adding to the confusion, the Supreme Court seemed to have difficulty remaining focused on the only patentability issue in question:  subject matter jurisdiction.  In the context of  101 as described above, there are several hurdles to patentability, with subject matter jurisdiction being simply the first.  The Court, referring to the process claim as "abstract and sweeping," was concerned with the breadth of the claimed invention in Benson.40  Yet, the breadth of the patent claims is the focus of a later patentability hurdle, after subject matter patentability has been determined.  By getting ahead of itself on this particular issue, the Court further compounded the confusion by erroneously including this irrelevant issue as an important factor of the subject matter patentability hurdle.

Restricting software patentability even more:  Parker v. Flook
The Supreme Court addressed the subject of software patentability for a second time in Parker v. Flook in 1978.41  Flook's patent application claimed an improvement in the method of monitoring the catalytic conversion process.42  Variables from the conversion process, including temperature and pressure, were monitored closely, and inefficiency or danger could be prevented by regulating the process when any variable exceeded a predetermined "alarm limit."43  The invention allowed for the continual alteration or "updating" of the alarm limits through the use of a mathematical algorithm and a computer.44  Applying the Benson test, Flook argued that by tying the claims to the updating of the "alarm limits," the invention did not preempt the mathematical algorithm.45  The patent examiner rejected Flook's argument and the application as non-statutory subject matter, and the Board affirmed the rejection.46  Maintaining its traditional stance, the CCPA overturned the rejection, agreeing with Flook that the invention satisfied the Benson test, because the patent claims avoided preempting the mathematical formula by including post-solution applications of the formula (the updating of the "alarm limits").47
In a decision that seemingly strengthened even more the strict approach to software patentability set forth in Benson, the Supreme Court reversed the CCPA's decision and held that the invention was non-statutory subject matter.  While agreeing with Flook that the invention did not wholly preempt the mathematical formula, the Court focused on the fact that the only new element of the invention was the mathematical formula.48  Creating what came to be known as the "point of novelty" approach, the Court held that the novelty of the claimed mathematical formula is irrelevant and that it is the process itself that must be new and useful.49  Since the mathematical formula was the only new element of the process, the Court declared it unpatentable subject matter.  In addition, the Court rejected the applicant's argument regarding the post-solution activity, concluding that an other otherwise unpatentable invention does not become patentable as a result of some post-solution activity.50
Flook expanded and modified the Benson test by adding the "point of novelty" approach and dismissing post-solution activity as a patentability factor.  After Flook, an invention that did not preempt the mathematical algorithm it claimed might still have been non-statutory subject matter if the formula was the only novel element of the invention and despite the presence of post-solution activity.  Unfortunately, the Supreme Court's decision did little to clear the air in this area, leaving many questions from Benson unanswered and creating even more with the new "point of novelty" approach.  The confusion surrounding the meaning of "algorithm" continued, and the PTO and the CCPA persisted in fundamentally disagreeing about the patentability of software inventions.  And the decision provided a stronger case for those arguing that the Supreme Court considered computer program-related inventions to be per se unpatentable subject matter.
Furthermore, the Supreme Court extended the application of later patentability standards to the initial  101 statutory subject matter hurdle, as it had in Benson.  In Flook, however, the misapplication was more pronounced, because the "point of novelty" approach was included as a fundamental element in the new standard.  In essence, the approach erroneously mixed the initial patentable subject matter requirement with parts of the later inventiveness requirement. However, finding an invention to be patentable subject matter, in the correct statutory sense, is completely unrelated to whether the invention is inventive.  As a result, combining the statutory requirements in the new standard set forth by Flook made the law even more ambiguous.
Hence, as the Supreme Court saw it after Flook, the patentability of a software invention seemed to hinge on whether the mathematical algorithm in the invention was wholly preempted by the application.  If so, the invention was unpatentable.  And even if the algorithm was not wholly preempted, the process was still unpatentable unless it was 'new and useful.'  Further, some physical post-solution activity was not enough to make the process patentable.
The CCPA's response:  In re Freeman (the beginning of the Freeman-Walter-Abele test)
In re Freeman was actually decided by the CCPA before Flook came down from the Supreme Court, and it marked the next major step in software patentability.51  With this decision, the CCPA made an attempt to reconcile and solidify the law created by the Supreme Court in Benson, while setting the stage for a new test to be further refined by two more future cases (the Freeman-Walter-Abele test, described below).  In a predictably repetitive procedural sequence, the PTO rejected Freeman's application claiming a system for typesetting alphanumeric information using a computer, and the CCPA reversed the rejection.52
While the Supreme Court would soon compound the confusion of software invention subject matter patentability in Flook, Freeman attempted to eliminate some of the ambiguity surrounding Benson.  The CCPA's first move was to address the misapplication of novelty and obviousness doctrines to the relevant issue of patentable subject matter under  101.  In affirming the examiner's rejection of Freeman's application, the Board had concluded that the invention did not satisfy the "point of novelty" approach, because the only novelty in the invention was the computer program used to load the known display typesetting device.53  The CCPA firmly stated that "[c]onsiderations of novelty or obviousness are of no effect whatever in determining whether particular claims define statutory subject matter under 35 U.S.C.  101."54  The court continued by setting out a 2-step test that did little more than break the Benson test out into a simpler, more concrete approach.  The first step (the step that the CCPA felt was being overlooked by the courts and the PTO) was to determine whether "the claim directly or indirectly recite[d] an 'algorithm.'"55  The second step was to determine whether the claim, as a whole, wholly preempted the algorithm.56
 Next, the CCPA pointed out its recognition of the discrepancy between the broader scientific definition of "algorithm" as something similar to a process and the Supreme Court's narrower definition of "algorithm" limited to mathematical algorithms.  Accepting the Supreme Court's version, the CCPA concluded that to assume that the high court actually meant the broader version of the term would mean that any "process" claim would be unpatentable.57  Concluding that Freeman's methods claims (claims 8 - 10) were intended to include an algorithm in its broad, scientific sense, the CCPA found that those claims did not satisfy the first step of the new test because they did not claim a mathematical formula, and thus the court did not have to consider the second step of the test.58  Similarly, the court found no mathematical formula claimed in any of the seven apparatus claims (1 through 7), so step two was ignored.59  Hence, the invention was found to be patentable subject matter.
 Hence, it appeared that the CCPA made an attempt to restate the Benson analysis in a more "software-friendly" fashion.  While maintaining the core of the Supreme Court's analysis with the application of the algorithm analysis, the lower court clearly seemed to be analyzing the software patents with a less restrictive approach.
 Of equal importance, with this decision the CCPA actually took a small step toward making the law of software subject matter patentability a little clearer.  The court was able to 1) point out the irrelevance of the utility and obviousness doctrines to the subject matter patentability issue (in seeming contradiction with the Supreme Court's subsequent use of the "point of novelty" test in Flook), 2) clarify the Benson doctrine to some extent, and 3) help explain the Benson interpretation of "algorithm" in light of the broader version of the term.  One question, however, that the CCPA specifically avoided was the applicability of the Benson doctrine to apparatus claims.  Freeman had made the argument before the patent examiner that since claims 1 through 7 were not process claims, but rather apparatus claims, the Benson doctrine didn't apply.  The patent examiner, recognizing the problem, rejected the seven apparatus claims for reasons other than unpatentable subject matter.60  The CCPA, however, explicitly stated that it did not reach the question because none of the claims included a mathematical algorithm, so it refused to discuss the issue.61  Yet, the court did state that it was "anomalous" to grant an apparatus claim utilizing a method that would be unpatentable as a process claim.62

The CCPA's second step in the Freeman-Walter-Abele test:  In re Walter
 In re Walter followed two years after Freeman (and Flook) and further modified the two-step test set forth in Freeman.63  In a rare decision upholding the PTO's rejection of an invention as unpatentable subject matter, the CCPA further explained the second step of the test.  According to the court, the second prong requirement that the algorithm be "wholly preempted" needed to be restated for clarification (and to reconcile the test with Flook).64  Thus, the second step of the test, after determining that there is a mathematical algorithm in the claim, is to examine the claim in its entirety.65  "If it appears that the mathematical algorithm is implemented in a specific manner to define structural relationships between the physical elements of the claim (in apparatus claims) or to refine or limit claim steps (in process claims)", then the claim is statutory subject matter.66  However, "if . . . the mathematical algorithm is merely presented and solved by the claimed invention . . . and is not applied in any manner to physical elements or process steps" then the claim is non-statutory, regardless of any post-solution activity or any recitation of the algorithm's intended use that might be present.67  The CCPA further explained that while an invention producing a pure number was non-statutory, an invention producing a physical thing represented by a number was statutory.68
 Speaking to the problem first introduced in Freeman of apparatus claims and where they fall in this analysis, the court created a new procedural approach to handle the question.  Whenever the computer-related invention solves formulas or makes calculations, the patent applicant must prove that the invention claims an apparatus that is distinct from any other apparatus that could do exactly the same mathematics.69  If the applicant is unable to make this proof, then the invention will be treated like a process and analyzed as such.
 Thus, by refining the two-pronged test that came to be called the Freeman-Walter test as a result of this decision, the court further reconciled the Supreme Court's Flook decision with Freeman.  Flook had held that an algorithm need not be wholly preempted to be deemed unpatentable, so the Walter decision moved away from the use of the "preemption" concept.  The CCPA was also able to help clarify the apparatus vs. process claim distinction.  As an aside, the court also discussed the "point of novelty" test, concluding that the Supreme Court had never actually adopted a "point of novelty" test for the subject matter patentability issue in Flook, because the high court would never destroy the patent statutes in such a manner as that kind of test would.70  This appears to be a very weak and unbelievable position for the court, considering that the Supreme Court in Flook described in detail the approach of dividing an invention into its old and new elements and then examining only the new element for patentability.71  It appears that the Walter court may have been going out of its way to read the prior cases selectively in order to bring them in line with the Freeman-Walter test.

The Supreme Court siding with the CCPA?  Diamond v. Diehr
 In 1981, the Supreme Court got another opportunity to speak on the topic of computer program-related subject matter patentability in the case of Diamond v. Diehr.72  The inventors, Diehr and Lutton, claimed an invention that controlled the operation of molding presses instrumental to the production of rubber products.73  The relevant claim, claim 1, was a methods claim with several steps.  Most of the steps in the claim described the monitoring of the molding process and the constant calculations of time remaining.74  However, the last step in the claim consisted of the automatic opening of the press when the elapsed time and the time required by calculation were equal.75  Again, the examiner rejected the methods claims as non-statutory subject matter, and again, the Board affirmed the examiner's rejection.76  And once again, the CCPA reversed, after which the Commissioner of the PTO, Diamond, sought review before the Supreme Court.77
 In a landmark decision for the computer program industry, the Supreme Court affirmed the decision of the Court of Customs and Patent Appeals, but avoided any discussion or application of the new Freeman-Walter test.  This was the first case by the Supreme Court to find a method claim containing a mathematical algorithm to be patentable subject matter.78  The Court first defined "process," then determined that the claimed invention, as "a physical and chemical process," fell within that definition of process in the context of  101.79  The Court then seemed to apply the original Benson test by acknowledging the existence of a mathematical algorithm in the process, but concluding that the invention does not preempt the algorithm. "Arrhenius' equation is not patentable in isolation, but when a process for curing rubber is devised which incorporates in it a more efficient solution of the equation, that process is at the very least not barred at the threshold by  101."80  In reaching the decision, the Court took its cue from the CCPA and renounced any use of novelty analysis in the initial subject matter patentability issue.81  Plus, the Court, like the Walter court, stated that an unpatentable mathematical formula cannot be made patentable simply by limiting the field of use in the patent or including "insignificant post-solution activity."82
 For the first time, despite utilizing the same doctrine applied in a more restrictive fashion in earlier cases (Benson and Flook), the Supreme Court seemed to be taking a broader, CCPA-like view of software patentability in Diehr.  The Court in Flook had rejected an invention which, like the Diehr invention, utilized continually updated calculations to monitor a physical process (the catalytic conversion)83, and arguably provided for the computer to activate some physical response if the calculation reached some value (the "alarm limits"), although the invention did not claim the actual step of activating an alarm.84  It would appear that the Flook court would have held the Diehr invention to be unpatentable.  However, the Supreme Court did not see it that way in Diehr.  In the decision, the Court distinguished Flook by pointing out that the Flook application contained no disclosure regarding chemical processes or the physical triggering of an alarm.85  In other words, the Supreme Court's position is that the Diehr invention has more connection with the physical process due not only to the actual opening of the press at the end of the curing period, but also to the description of that physical activity in the patent application.  Whether that is evidence of a consistent application of the Benson test is open to debate.  There is a strong argument that the Diehr interpretation of the Benson test requires less connection to the physical element of the invention.
 Diehr was also an important case for attempting to eliminate, once and for all, any use of inapplicable later patentability doctrines in determining software subject matter patentability.  After being the initial proponent of the misguided doctrine, the Supreme Court seemed to finally concede to the CCPA's position without expressly overruling its previous decisions on the issue or acknowledging that the high court was following the CCPA's lead.  In addition, Diehr appeared to set forth a new significance standard regarding post-solution activity.  By stating that "insignificant post-solution" activity would not make an otherwise unpatentable invention patentable86, the Court seemed to be inferring that significant activity tied to the solution of the mathematical algorithm could make the process patentable.
 Unfortunately, Diehr did create some level of uncertainty by finally siding with the CCPA but apparently applying the Benson test instead of the Freeman-Walter test perfected the previous year by the special jurisdiction appeals court.  Perhaps the Supreme Court was concerned that by adopting the test created by the lower court, it would not only be potentially admitting that its previous decisions were poorly reasoned but also conceding to a lower court that the lower court's decisions contained better legal reasoning.  On the other hand, it could be argued with equal force that the Supreme Court simply felt that the CCPA test was a poor approach to the issue in comparison to the Benson test.  Either way, there is also evidence that the case did not create confusion in the area of patentable subject matter, but rather had the opposite effect.  Cases following Diehr seemed to exhibit a sense of comfort and stability in the available legal doctrine that might be attributed to the decision in Diehr.87

The CCPA's third step in the Freeman-Walter-Abele test:  In re Abele
 In re Abele, decided by the CCPA in 1982 (after Diehr), was the final installment to the Freeman-Walter test, consequently making it the Freeman-Walter-Abele (FWA) test88.  In Abele, the court reversed the PTO's rejection of an invention claiming a method of displaying data from a computerized axial tomography (CAT) scan.89  The court's relevant holding was another broadening of the test's second prong.  The second prong became a simple requirement that "the algorithm be 'applied in any manner to physical elements or process steps,' provided that" the algorithm is restricted by more than just a field of use limitation or insignificant post-solution activity.90  In short, the second prong should ask whether, even if inoperable or less useful, the claim would be "otherwise statutory" without the algorithm.91  If so, then the claim is patentable subject matter.
 The establishment of the FWA test seemed to be a major step for the CCPA in settling subject matter patentability issues involving computer program-related inventions.  And as mentioned above, the test in some respects also seemed to align CCPA's analysis with that set forth by the Supreme Court.  Yet, the CCPA appeared to apply its test in a less restrictive manner.  Section 101 rejections of software patents had now moved, in just over a decade, from the per se rejection of all computer program-related applications to the rejection of only those claims involving "a mathematical algorithm and no external physical connections or effects."92  Furthermore, the test created stability in the area.  The PTO and the courts applied the FWA doctrine throughout most of the 1980's with few changes and little controversy.  Even the constant feud between the PTO and the CCPA died down.  In fact, the CCPA was replaced by the Court of Appeals for the Federal Circuit (CAFC) soon after the Abele decision.  The CAFC was established with the same jurisdiction as its predecessor, and effectively adopted the earlier court's precedents by reaffirming the authority of the CCPA's decisions.93

The CAFC's restrictive first major attempt at a software patent:  In re Grams
 In 1989, the Court of Appeals for the Federal Circuit sifted through the available analyses from the CCPA and the Supreme Court in deciding In re Grams.94  Grams and Lezotte claimed "a method of testing a complex system to determine whether the system condition is normal or abnormal and, if it is abnormal, to determine the cause of the abnormality."95  More specifically, the invention was essentially a set of steps to diagnose abnormal medical conditions in a patient, with the first step being to test the patient and the rest of the steps consisting of an analysis of the resulting data to determine the possible causes of any abnormality.96  The patent examiner rejected the application, and the Board of Patent Appeals affirmed the rejection on the grounds that the invention was essentially directed to a mathematical algorithm or business method.  Grams and Lezotte appealed to the CAFC.97
 Contrary to the long-standing tradition of the CCPA to reverse the PTO's rejections of computer program-related inventions, the CAFC affirmed the PTO's rejection.  Pointing out that Benson had not been overruled and was still good law, the court set forth the basic precept that an algorithm is not patentable subject matter, with the caveat that including physical steps might make the invention patentable.98  The court then cited Walter and Abele in summarizing the FWA test without ever acknowledging that designation.99  The CAFC continued by stating that this mathematical algorithm analysis was simplified when, as here, only one step was physical and consisted of the mere collection of data.100  Quoting a 1978 CCPA case, the court stated that "[i]f the steps of gathering and substituting values were alone sufficient" to attain subject matter patentability for an otherwise algorithmic claim, then any useful algorithm would be a patentable process.101  Abele was distinguished on the basis that the algorithm in Abele's invention refined "a process step in a process that is otherwise statutory" while the physical first step in Grams' invention was not refined by the subsequent algorithm steps.102
It is difficult to determine whether the CAFC's application of the FWA test harmonized with the CCPA's application of the test as it stood in the early 1980s.  The CAFC did make a strong argument that the algorithm at issue did not refine the initial physical step of the invention.  However, that conclusion may arise from a slight misinterpretation of the FWA test.  Applying the pure FWA test as it stood in 1982103 seems to result in the conclusion that Grams' invention was patentable subject matter.  As a reminder, the test (after identifying an algorithm) was to determine whether the algorithm was "applied in any manner to physical elements or process steps," assuming there was more than a simple "field of use limitation or insignificant post-solution activity."104  It could be argued that the algorithm steps (analyzing the data and making any possible diagnoses) were applied to the physical step of collecting the data through lab tests.  However, given the close nature of this question, Grams may not stray far from the CCPA analysis.

The CAFC's confusing second major attempt:  In re Iwahashi
The CAFC had the opportunity to address the issue of software invention patentability four days later with In re Iwahashi.105  The patent application specification generally described an invention relating "to an auto-correlation unit for use in pattern recognition to obtain auto-correlation coefficients as for stored signal samples," and the main embodiment of the invention in question was related to voice recognition.106  Essentially, the invention calculated auto-correlation coefficients at high speed for use in pattern recognition without the need for the complicated circuitry or the expensive multiplier required by the prior art.107  All of the claims were in the "means plus function" form except for the apparatus claim directed to a ROM ("read only memory"), which was a permanent storage device used to store the squares of certain variables.108  Like the invention in Grams, Iwahashi's invention was rejected by the PTO as nonstatutory subject matter, "because it is merely a mathematical algorithm."109  And like Grams, Iwahashi appealed to the CAFC.
Unlike Gram, Iwahashi prevailed in the Court of Appeals for the Federal Circuit, which reversed the PTO's rejection, bringing to mind the long-standing disagreement between the PTO and the CCPA.  The CAFC first laid out the Freeman-Walter test, then cited Walter for the definitive test:

[I]f it appears that the mathematical algorithm is implemented in a specific manner to define structural relationships between the physical elements of the claim (in apparatus claims) or to refine or limit claim steps (in process claims), the claim being otherwise statutory, the claim passes muster under  101."110

Agreeing with the patentees that the "means plus function" claims at issue were apparatus claims, the court concluded that there was no reason not to find that these claims were directed to a patentable machine or manufacture.111
 The court's reasoning in Iwahashi seems a bit unusual in light of the Grams case of four days earlier.  While the CAFC relied on the FWA test in Grams (without actually giving it a name), the court ignored the existence of Abele as a part of the test in Iwahashi and applied the test as set out in Walter (actually calling it the Freeman-Walter test).  In other words, the two decisions, handed down four days apart by the same court, applied two different tests on the same issue.  It seems that the CAFC was having a difficult time with the application of the available analyses for software subject matter patentability.  In light of these two cases, it was almost impossible to predict what analysis would be applied in any similar case in the future.

The CAFC applying the FWA test or Benson?  Arrhythmia v. Corazonix
 Then Arrhythmia Research Technology, Inc. v. Corazonix Corp. came before the CAFC in 1992.112  The case was an infringement action involving an issued patent claiming an invention for analyzing electrocardiograph signals to detect a particular heart condition called ventricular tachycardia.113  The claims consisted of both apparatus and process claims, and the apparatus claims were in a form called "means plus function," which is a claim structure permitted by section 112, paragraph 6, stating:

An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material or acts described in the specification and equivalents thereof.114

As a result of the emphasis on function and not structure, determining the scope of the protection of a "means plus function" claim requires an examination of the patent specification.  The structure taught in the specification would be a starting point for the scope of the patent, and would include "equivalents thereof," as the text of  112 explains.115  In other words, the patent covers not only the structure described in the specification, but anything beyond that structure that could be considered equivalent.  The end result is a potentially very broad claim, especially in the area of computer program-related inventions.  "It appears unlikely that an expert could design a device that does not fit the means provided in the specification, and equivalents thereof, and thus infringe the patent."116
Arrhythmia originated as an infringement suit by the plaintiff against the defendant, Corazonix, who in turn asserted the defense that the plaintiff's patent was non-statutory subject matter under  101.117  The district court agreed with the defendant and dismissed the lawsuit.  The CAFC, however, reversed the lower court, finding that the patent in question did claim patentable subject matter.118
 After going through the line of Supreme Court cases, including Benson and Flook, the CAFC set forth the FWA test.119  As manifested in Arrhythmia, the first step was described as an examination of whether a mathematical algorithm was directly or indirectly set forth in the claim.120  The second step was to determine whether the algorithm in the claim was "applied to or limited by physical elements or process steps."121  If so, the claim was patentable subject matter.  Applying the first step to the patent in question, the court found that a mathematical algorithm was present in both the methods and the apparatus claims.122  Moving to the second step, the CAFC first pointed out that the methods claim was limited to identifying the presence or absence of a particular signal level.123   Then the court emphasized the physicality of the process input signals, which were produced by the patient's heart.124  Finally, the court described the physicality of the invention's output in the form of a signal produced from a digital waveform filtration process.125  Following from its pronouncement that electrical signals are physical, the CAFC concluded that the mathematical algorithm was applied to physical process steps and thus satisfied the second prong of the test.  Strangely, the court then seemed to drop back to a quasi-Benson analysis.  It described the invention as not wholly preempting the mathematical algorithm, and concluded that the patentee was not attempting to patent a mathematical formula.126
 Moving on to the apparatus claims, the CAFC again emphasized the physical aspects, describing the claimed means as converting one input signal into another output signal and emphasizing again the physicality of the end product.127  Interestingly, the court neither mentioned the FWA test (or any test for that matter) nor cited back to its earlier discussion of the doctrine.
 Arrhythmia sent some mixed and confusing signals.  The decision purportedly relied on the CCPA's FWA test as promulgated by the CAFC during the 1980's, and the emphasized physicality of electrical signals clearly exhibited the broader approach to the patentability of computer program-related inventions that had been prevalent in the courts for most of that decade.  However, the court also seemed to drop in a quick pre-FWA Benson analysis through its discussion that the invention did not "wholly preempt" the algorithm.  It is difficult to pinpoint exactly where the court was falling on the choice of legal doctrine in this case.  And the confusion was compounded by the presence of the  112 "means plus function" format for the apparatus claims.  It seems like the court omitted any legal analysis of the apparatus claims, and instead focused on the connection between the mathematical algorithm and the physical elements of the invention.
In a concurring opinion, Judge Rader felt that the court was erroneously relying on Benson.  Rader argued that the Supreme Court had severely limited the Benson approach in Diehr and that the CAFC should not be relying on its doctrine in this case.128  In fact, it appears that Rader's argument was intended to criticize the FWA test as part of the Benson approach.  Rader refers to the majority's analysis as a "permutation of the Benson algorithm rule" and the "two-step post-Benson test."129  Diehr not only limited Benson, according to Rader, but also introduced a new approach to software subject matter patentability based solely on the language of the statute in  101.130  Essentially, Rader wanted to eliminate any discussion of mathematical algorithms and simply focus on whether the invention was literally a process or not.131

The CAFC moving beyond algorithms with a new test and new confusion:  In re Alappat
Hence, after a period of relative stability, software subject matter patentability was again in a state of confusion. Out of this sea of ambiguity arose a landmark decision, In re Alappat, from the Court of Appeals for the Federal Circuit in 1994.132  And once again, the old feud between the PTO and the patent appeals court (now in the form of CAFC) was renewed.  Like the claims in Arrhythmia, the patent application included apparatus claims with the  112 "means plus function" structure.  These claims were directed to a method of manipulating pixel intensity in a way that smooths the appearance of images on a digital oscilloscope screen.133  The PTO, after an initial rejection reversal and then an unusual reconsideration process by the Board, ultimately rejected Alappat's application.134
The CAFC reversed the rejection of claim 15, holding that it was a patentable invention directed to the "machine" category provided in  101.  After stating that the invention was a machine, the CAFC pointed out that the "mathematical algorithm" analysis may not apply to machine claims, and then applied the analysis anyway.135  Instead of relying on the Benson analysis or the more refined FWA test, the court introduced a new test.  First, the court read the "mathematical algorithm" analyses of Benson, Flook, and Diehr to simply stand for specialized versions of the more fundamental, underlying doctrine that abstract ideas are not patentable.136  The court's next step was to cite Diehr in pointing out that any claim must be analyzed as a whole, and that just because one element of the claim is unpatentable does not automatically make the entire claim unpatentable.137  Given that background, the test was "whether the claimed subject matter as a whole is a disembodied mathematical concept . . . which in essence represents nothing more than a 'law of nature,' 'natural phenomenon,' or 'abstract idea.'"138
Applying this test to Alappat's invention, the court determined that the invention was not a "disembodied mathematical concept, . . . but rather a specific machine to produce a useful, concrete, and tangible result."139  To this end, the CAFC pointed out that 1) claim 15 was limited to a specific combination of events that resulted in the transformation of digitized waveforms into smooth waveforms,140 2) the preamble specifically described the intended purpose of the machine and the claims recited back to the preamble141 (distinguishing this from a mere field of use limitation), and 3) a special purpose computer was created from a general purpose computer once the computer was programmed to perform certain functions.142
The court's conclusion that computer programming created a specialized computer related to the use of the "means plus function" structure for claim 15.143  As described above, the focus on the function element of a "means plus function" claim can result in a rather broad scope for the structure element.  In this case, the court interpreted claim 15 broadly in deciding that a programmed (specialized) general purpose computer was an equivalent of the claimed means.144
Alappat seemed to be a step back from the "mathematical algorithm" analysis that had been prevalent in the caselaw for over 20 years (back to Benson in 1972).  While the CAFC purported to simply interpret the analysis of Benson, Flook, and Diehr, in reality the court was doing much more than interpretation.  The court was actually adopting what it called the "core" of each of the cases, the "abstract ideas" doctrine, and that "core" was a far cry from the "mathematical algorithm" doctrines applied in each of those cases.145  By downplaying the "mathematical algorithm" focus of the software subject matter patentability inquiry, the court seemed to settling on a more "pure" doctrine of subject matter patentability based on the language of  101.  In fact, the majority's approach here seemed very similar to Judge Rader's approach in his concurrence to Arrhythmia.  As discussed above, Rader advocated elimination of the "mathematical algorithm" element of any subject matter patentability test in favor of a more textual doctrine.146
Yet, there is a question whether the doctrine was substantially changed.  Despite creating a new test with less focus on the presence of a "mathematical algorithm," the test still looked for a "disembodied mathematical concept."147  Plus, the court still concerned itself with whether the claim wholly preempted any other apparatus using the same mathematical calculations.  Furthermore, the specific limitations placed on the claim in the preamble (a "field of use" limitation) is an important factor for the court, despite the fact that the "field of use" limitation was an element of the "mathematical algorithm" analysis.  Therefore, it is unclear that the application of this new test will actually result in a different outcome than if the "mathematical algorithm" doctrines were applied instead.
Judge Archer's dissent criticized the majority's decision for a different reason.  The judge felt that this was a charade, a sly attempt to make an unpatentable invention patentable by claiming a machine instead of a process.148  The concern was that the CAFC ultimately determined that the invention in Alappat was patentable because the "means plus function" structure increased the scope of the claims to include a programmed general purpose computer.149  In other words, any potential patentee concerned that her process claims to a software invention would likely be rejected as unpatentable subject matter needed only to restructure those claims as "means plus function" claims in order to get the benefit of the broad scope for the function element and ultimately obtain a patent.  The dissent argued that the result would be a flood of software patents flowing into the PTO, because all those claiming a programmed general purpose computer will be statutory subject matter.  A likely response might be that the CAFC, despite stating that there was no requirement to do so, still applied the new Alappat test to the machine claim in this case.  The argument would continue by claiming that this move exhibited an intent on the part of the CAFC to apply the same test to both process and machine claims, and that this equal application of the test would prevent the charade that Judge Archer was concerned about.
It is almost impossible to say which side of this argument has the advantage, because the CAFC's opinion in Alappat provided no clear answer on the issue.  While declaring Alappat's invention to be patentable subject matter, the court did not expressly indicate how the inclusion of a programmed general purpose computer affected that patentability.  Due to this ambiguity, the court may have been effectively creating a bright-line rule that all such claims attached to a machine are patentable, or the court may have simply been saying that a machine claim must be subjected to the same analysis provided in Alappat as any similar process claim.
Other ambiguities abound in the opinion, including the status of the FWA test.  The court gave no indication of the disposition of the old CCPA test after Alappat.  The CAFC seemed to delve beyond that test to the underlying "core" principle of the "abstract idea," creating the impression that the FWA test was being thrown out.  Yet, considering that the opinion did not expressly overrule the test, the two-step analysis still seemed to be available.  Thus, the only real conclusion that can be drawn from the Alappat opinion is that the CAFC created a doctrine with a broader approach to computer program-related subject matter patentability than the FWA test.
By this point, the courts had made it clear that there were two  101 subject matter categories available for a computer program-related invention.  The PTO would give serious consideration to any software invention claimed as a "process" or a "machine," assuming that the application met all other requirements.  One other category available under  101 that seemed to be a possible avenue for software inventions was the "article of manufacture," a kind of "catch-all" of statutory subject matter.  The courts had not spoken on the subject in the past, but that began to change in 1994 when the CAFC heard In re Warmerdam.150

The first "article of manufacture" case:  In re Warmerdam
The invention involved in Warmerdam was a collision avoidance system that operated by controlling the movement of an object to avoid collision with another object.151  The application consisted of both methods and apparatus claims.  The PTO rejected the claims as unpatentable.152
The CAFC affirmed the rejection for 5 of the 6 claims, but reversed as to claim 5.  After criticizing the "mathematical algorithm" approach to computer-related subject matter patentability, the court applied the general "abstract idea" test to find that the first four claims (the methods claims) were directed to nothing more than "the manipulation of basic mathematical constructions."153  The CAFC then moved on to find claim 5 to be a patentable claim for a machine.154  Finally, the court turned to claim 6, which was directed to any data structure created by the processes of claims 1 through 4.  Noting that the application did not define "data structure," the court cited a computer dictionary defining it as a "'physical or logical relationship among data elements, designed to support specific data manipulation functions.'"155  Emphasizing that the claim did not contain any physical elements, and distinguishing it in that respect from a data structure found to be patentable subject matter in In re Bradley, the court held the data structure to be unpatentable for the same reasons as claims 1 through 4.156
Although Warmerdam did not directly deal with the issue of the article of manufacture claim as a possibility for software inventions, the CAFC did provide an interesting discussion of data structures that would later be helpful in that area.  Specifically, the court's emphasis on the hardware, or physical, element of the patentable claims pointed to the use of such hardware in relation to software in order to achieve patentability.  Note that the CAFC found the data structure unpatentable on the basis that it had no ties to any physical structure (hardware), in contrast to the patentable data structure in Bradley, which "was a physical, interconnected arrangement of hardware."157  This seemed to open the door to the possibility that a claim directed to a software invention may be patentable subject matter when tied to some sort of hardware, such as a diskette or a hard drive.

The second "article of manufacture" case:  In re Lowry
If Warmerdam opened the door, then In re Lowry was the first step through the doorway.158  Decided by the Court of Appeals for the Federal Circuit in the same year as Warmerdam, Lowry involved an application for a patent on a way to store and use data in computer memory.  The invention claimed a data structure that was "accessible by many different application programs," and emphasized the organization of data on the basis of "attributes and relationships between attributes."159  In other words, this memory storage invention utilized an efficient data structure that could be utilized by many programs and had a complex, hierarchical organization.160  In contrast to Warmerdam's claimed data structure, Lowry's seemed to be more specific and limited, with claim 1 direct to "[a] memory for storing data for access by an application program being executed on a data processing system."161  While claims 1 through 5 were directed to this memory (including the data structure), claims 6 through 19 were to a data processing system in the form of "means plus function" claims, and claims 20 through 29 were to methods.162
The patent examiner rejected the memory claims (1 through 5) as non-statutory subject matter under  101, and continued on to reject the rest of the claims for other statutory reasons.163  The Board reversed the rejection as to claims 1 through 5, concluding that these claims were to an article of manufacture and were directed to patentable subject matter.164  However, the Board then moved beyond the subject matter patentability issue to affirm the overall rejection of all the claims on grounds of obviousness.  The Board concluded that the data structure was analogous to printed matter, which the court stated is indistinguishable from prior art when not functionally related to the substrate.165
The CAFC reversed the rejection, holding that the Board's statutory unpatentability analysis was misplaced and thus in error.  Specifically, the court held that the  102 "printed matter" doctrine was incorrectly applied to the Board's  103 rejection.166  While not directly discussing subject matter patentability, the court concluded that this was no abstract invention, stating that "the claims require specific electronic structural elements which impart a physical organization on the information stored in memory."167  In fact, the court, citing In re Bernhart, concluded that there is a physical change in a computer when it is programmed.168
The CAFC in Lowry seemed to take a generally more physical view of data structures than in Warmerdam, defining them as "the physical implementation of a data model's organization of . . . data."169  In contrast to the earlier definition, which did not contain any reference to any physical aspects, the inherent physicality of the Lowry definition seems to shorten the step between analysis of the data structure and the conclusion of patentability.  In fact, one could read Lowry to be concluding that the only characteristic needed to create a patentable article of manufacture was a connection to a physical medium or any physical change in the structure (which Warmerdam was lacking).170
In addition, the court seemed to accept the Board's designation of claims 1 through 5 as article of manufacture claims.  The CAFC provided no direct discussion of the "article of manufacture" designation when there was plenty of opportunity for such a discussion if the court had been concerned.  In fact, it has been argued that the court's reasoning was, in fact, completely consistent with the view that the invention was an article of manufacture.171

An anomaly in the "article of manufacture" cases?  In re Trovato
Before In re Trovato172 was decided by the CAFC in December of 1994, it appeared that the logical progression from Warmerdam to Lowry was a reflection of the court's increasingly broad view of the patentability of data structures and the increasing possibility of patenting a software invention as an article of manufacture.  However, Trovato seemed to be a wrench in the works. Trovato's two inventions literally found the best path between two points, whether that path was optimized by distance, cost, or some other variable.173  They operated by utilizing a "configuration space," to "model possible object movements," and this configuration space was stored in a data structure.174  The patent examiner rejected both applications, which included methods and apparatus claims, on the grounds that neither invention could survive the Freeman-Walter-Abele test.175  The Board sustained the rejection as to most claims, but reversed the rejection as to 3 claims which recited specific alarm systems and identified specific objects to be moved, such as emergency vehicles.176  Trovato appealed to the Court of Appeals for the Federal Circuit.
The CAFC affirmed the Board.  Considering the methods claims first, the court applied the FWA test and determined that the claims, despite not reciting a formula, were nonetheless directed to a mathematical algorithm and contained no underlying physical process.177  Next, the court noted the similarity between the methods claims in this case and those in Warmerdam.178
Moving to the apparatus claims, the court pointed out that they were in the same format as the methods claims except for the insertion of the words "apparatus" and "means for" in each one179.  Citing In re Bradley for the rule that even a machine claim must be analyzed for the presence of mathematical algorithms, the CAFC concluded that the additional terminology added to the apparatus claims were "of no patentable significance."180  The word apparatus in the preamble could be disregarded, because "[t]he effect preamble language should be given can be resolved only on review of the entirety of the patent to gain an understanding of what the inventors actually invented."181  Furthermore, the court found no structure in the specification to represent the possible means of the invention, as required by 35 U.S.C.  112,  6.182  Hence, the CAFC held that the apparatus claims were not directed to a machine, did "no more than solve a mathematical algorithm," and thus were non-statutory subject matter.183
In re Trovato appeared to ignore the most recent precedents of the Court of Appeals for the Federal Circuit.  First, the court seemed to be ignoring the more recent Alappat test, instead applying the FWA test to the methods claims.  Second, the dismissal of the purported apparatus claims as nothing more than process claims with a few extra words added seemed to contradict the Lowry decision.
In response to the second point, an argument could be made that the invention in Lowry had a stronger overall connection with physical elements than the Trovato invention, and thus could be distinguished in that manner.  A comparison of the inventions, however, seems to support the contrary position.  While Trovato's invention is directed to the seemingly physical endeavor of finding the shortest distance between two points and actually works within a "physical task space,"184 Lowry claims what appears to be a more abstract invention - an improved organization and storage of information in a memory.185
In this context, it was impossible to predict what the court was actually saying.  Possibly, it was backing away from its more liberal analysis of software inventions in the previous few years.  Or perhaps the court had found Lowry and Trovato to be distinguishable on some point that it simply did not mention.  Yet, the court in Trovato does not even acknowledge the existence of the Lowry decision.  A third possibility was that the court was experiencing some internal disagreement on the issue.  In that context, it may be of more than passing relevance that the three judges who decided Lowry (Rich, Skelton, and Rader) were not involved in the Trovato decision (decided by Nies, Michel, and Schall).  Whatever the reason, confusion reigned again.

Finally accepting software inventions as "articles of manufacture":  In re Beauregard
In re Beauregard186, handed down by the CAFC five months after the Trovato decision, never actually had to be decided by the court.  Beauregard's article of manufacture claims were directed toward an article "'composed of a computer usable medium in which a program code is embodied.'"187  The PTO had rejected the application based on the "printed matter" doctrine discussed above in the Lowry analysis.  However, after Beauregard appealed, the PTO moved to dismiss, because it concluded "'that computer programs embodied in a tangible medium, such as floppy diskettes, are patentable subject matter under 35 U.S.C.  101,'" and that the printed matter doctrine did not apply.188  Hence, the CAFC reversed the PTO's rejection and remanded.
 The Patent and Trademark Office took quite a leap with Beauregard, seemingly giving blanket patentability to any computer program embedded in an item capable of being read by a computer.  In other words, any computer program on a diskette or CD-rom could be patentable subject matter.  It is conceivable that the PTO's conception of software patentability jumped beyond the holdings of Warmerdam and Lowry by arguably creating a lower threshold than any interpretation of those two cases.189  And Trovato appeared to be even more of an anomaly than before the PTO's concession.  Regardless, computer software inventors now had a third avenue by which to apply for software patents without fear of a per se subject matter patentability rejection.

The CAFC's latest very broad yet very confusing test:  State Street v. Signature Financial
In 1998, the Court of Appeals for the Federal Circuit heard State Street Bank & Trust Co. v. Signature Financial Group, Inc.190  The patent at issue contained 6 "means plus function" apparatus claims directed to a system for organizing and managing a consolidated mutual fund.  Basically, the computerized accounting system made it possible to pool the funds from several smaller independent mutual funds (referred to as "Spokes") into one larger mutual fund in the form of a partnership (referred to as the "Hub"), while maintaining separate statistics for each Spoke and the Hub and allowing allocation to each Spoke of daily income, expenses, etc. based on each Spoke's percentage share of the total fund.191  The consolidation created administration cost savings and tax advantages for the partnership.192
The lawsuit arose after the plaintiff attempted to license the patented system from the defendant.  When the negotiations fell through, the plaintiff brought an action in the district court to invalidate the patent for failure to claim statutory subject matter.193  The district court applied the Freeman-Walter-Abele test and held the patent to be invalid because it merely solved a mathematical algorithm and because it was a business method (explained below).194
The CAFC first cited Alappat for the proposition that machine claims in the "means plus function" structure may only be interpreted as process claims when "there is no supporting structure in the written description that corresponds to the claimed 'means' elements."195  Under this rule, the court determined that claim 1 (the primary claim at issue in this dispute) was a machine claim.196  The court then analyzed the claim in the context of what the district court had called the "mathematical algorithm" exception.  The CAFC emphasized that the exception was nothing more than another manifestation of the concept that basic ideas and scientific theories are not patentable.197  Ultimately, the court concluded, an algorithm is only patentable when "applied in a 'useful' way."198  To that end, the court again seemed to set out a new test asking whether the invention containing the mathematical algorithm produced "a useful, concrete and tangible result."199  If so, then the invention is patentable subject matter.  The court concluded that Signature's patented system produced a "useful, concrete, and tangible result," and thus was patentable subject matter.200
But the CAFC did not stop there.  Cognizant of the fact that the state of the law in this area was in some disarray, the court went on to discuss the older FWA doctrine.  It stated that after Diehr and Chakrabarty (another similar case), the "test has little, if any, applicability to determining the presence of statutory subject matter."201  Continuing in that vein, the court argued that the focus of subject matter patentability should be on the essential characteristics of the invention, especially utility, rather than where the invention might fall among the four statutory categories.202  Finally, the CAFC also abolished an old exception to statutory subject matter that declared all business methods unpatentable.  Declaring that it had never invoked the business methods exception and that the exception was eliminated by the 1952 Patent Act,  the court found the exception contrary to established subject matter patentability law.203
Following from Alappat, one would assume that this case would have started out by applying the "disembodied mathematical concept" approach, but the CAFC once again created a new analysis.  The new "useful, concrete, and tangible result" test in State Street takes an even greater step away from the old "mathematical algorithm" approach than was taken in Alappat.  And the court took quite a leap in legal reasoning to make that step.  First, Judge Rich reasoned that mathematical algorithms are found to be unpatentable because they are abstract ideas that are not useful.204  Continuing from that logic, all patentable algorithms must be "applied in a 'useful' way," and hence the focus of the new test on practical utility.205  In this manner, State Street appeared to have finally eliminated, at least for this case, the "mathematical algorithm" focus of subject matter patentability.  The court finally succeeded in getting away from what some critics called an erroneous focus on a issue that is irrelevant to subject matter patentability.206
However, the victory may have come at great cost.  In the course of eliminating the old test, the CAFC may have introduced an erroneous doctrine similar to that which the courts had previously struggled with for several cases and recognized as inapplicable in Diehr.207  In short, the emphasis of the new test on utility has again created the possibility that the courts are applying later patentability doctrines to determine subject matter patentability.  As stated previously, the fundamental characteristics of a patentable invention include eligibility (invention must be eligible subject matter), inventiveness (the invention must be novel and non-obvious), and utility (the invention must be useful).  In order to reach the examination of the separate and distinct requirements of novelty and utility, the invention must first be determined to be eligible subject matter.  And since those further requirements are examined in full once the initial eligibility requirement has been met, there is no reason to consider the later requirements during the initial inquiry into subject matter patentability.  Yet, State Street seems to justify inquiries into utility before there is any reason to consider that aspect of the invention.
Additionally, the CAFC's emphasis on practical utility also seems to ease, or even eliminate, the physicality requirement.  The courts had emphasized the transformation of matter as an essential characteristic of a patentable software invention and focused on the abstract idea's tie to the physical element of the invention.208  Yet, the CAFC effectively concluded in State Street that the production of calculated dollar amounts was enough transformation to justify patentability.  In contrast to the waves produced in Alappat, which seemed to have some semblance of physicality, the numbers produced by the Signature patent did not.  Considering the lack of any physical element in the end-product, State Street may be interpreted to say that any computer program-related invention that calculates useful numbers is patentable subject matter.
While State Street also seemed to eliminate the Freeman-Walter-Abele test, the court's rationale for dumping the doctrine is deeply flawed.  As mentioned above, Judge Rich stated that the FWA test has little applicability after Diehr and Chakrabarty.209  Yet, Abele was decided after both of those cases.  In fact, some argue that Abele was written to reconcile the previously existing F-W test with the Diehr decision.210  The argument that Diehr or Chakrabarty had any effect on the FWA test can hold no water.  Thus, it is difficult to predict whether this seeming attempt to reject the FWA test will be followed, rectified, or ignored.

Conclusion
As can be seen from this analysis, the latest installment in the judicial attempts to set forth a clear line of reasoning regarding software patent eligibility seems to be a major failure.  There is no way to know, based on the wildly erratic past decisions of the courts, the long-term status of this new "useful, concrete, and tangible result" test.  It may be eliminated with the CAFC's next software case, it may be drastically modified, it may be ignored, or it may be followed.  It is even difficult to know the state of the doctrine today, considering that none of the previous approaches have been explicitly overturned and may pop back up tomorrow.  In addition, the erroneous focus on utility as the focus for the new test further weakens the new approach.
In contrast, while the long-term impact of State Street is impossible to predict, it seems from the decision that any computer program is eligible subject matter.  Most commercial software products produce an arguably useful result, whatever the scope of that usefulness turns out to be.  Only time will tell if the courts will be able to clear the muddied water of software patent eligibility, or if they will be content to maintain the current broad approach to eligibility and focus on the later requirements of patentability for computer program-related inventions.
In summary, nothing concrete can be concluded about the current status or direction of the software subject matter patentability issue.  One definitive conclusion to be drawn at this point is that software inventors now have three avenues to patent their inventions, and this has proven to be major development for the software industry.  Assuming a particular invention meets the other statutory criteria, it may be patented as a method (process), an apparatus (machine), or an article of manufacture.  Unfortunately, this position was not achieved as a result of direct goals or clarity on the part of the PTO or the courts.  Rather, the resulting broad but ambiguous doctrines were produced in a chaos of continual political and doctrinal battles with no clear winner.  Although it is unlikely that the now broad approach to software subject matter patentability will be narrowed again, anything is a possibility as long as the caselaw surrounding the issue remains so scattered and cryptic.

1 35 U.S.C.  101 (1997).

2 35 U.S.C.  102, 103 (1997).

3 35 U.S.C.  112 (1997).

4 Vincent Chiappetta, Patentability of Computer Software Instruction as an "Article of Manufacture:" Software as Such as the Right Stuff, 17 J. Marshall J. Computer & Info. L. 89, 99 (1998).

5  Id. at 99.

6 Brian Richard Yoshida, Claiming Electronic and Software Technologies:  The Effect of the Federal Circuit Decisions in Alappat, Warmerdam, and Lowry on the Claiming of Mathematical Algorithms and Data Structures, 45 Buff. L. Rev. 457, 458 (1997).

7 Id. at 468.

8 Id. at 468.

9 Id. at 468.

10 Id. at 468.

11 See Rubber-Tip Pencil v. Howard,  87 U.S. 498 (1874).

12 Chiappetta, supra note 4, at 131.

13 O'Reilly v. Morse, 56 U.S. 62 (1854).

14 The Telephone Cases, 126 U.S. 1 (1887).

15 In re Abrams, 188 F.2d 165, 166 (C.C.P.A. 1951).

16 Yoshida, supra note 6, at 472.

17 Id. at 471.

18 Id. at 471 (note 69).

19 Id. at 472 (citing In re Prater, 415 F.2d 1393 (C.C.P.A. 1969)).

20 Prater, 415 F.2d at 1402.  The court explained:

Couched in terms of the present statute, it may well be that today purely mental steps are unpatentable because purely mental steps may, at present, be too vague and indefinite to comply with 35 U.S.C.  112.   . . . Whether or not a sequence of purely mental steps comes within the bounds of 'process' as used in 35 U.S.C.  100 and 101 is, we feel, an issue which has never been squarely decided.

21 See Chiappetta, supra note 4, at 107.

22 Yoshida, supra note 6, at 473.

23 Id. at 473 (citing In re Musgrave, 431 F.2d 882 (C.C.P.A. 1970)).

24 Gottschalk v. Benson, 409 U.S. 63, 64 (1972).

25 Id. at 73.

26 Id. at 64.

27 Id. at 64.

28 Id. at 67.  The court explained that "[p]henomena of nature, though just discovered, mental processes, and abstract intellectual concepts are not patentable, as they are the basic tools of scientific and technological work."

29 Id. at 67 (Citing Mackay Co. v. Radio Corp., 306 U.S. 86; Rubber-Tip Pencil Co. v. Howard, 87 U.S. 498; Le Roy v. Tatham, 55 U.S. 156; Funk Bros. Seed Co. v. Kalo Co., 333 U.S.
127).

30 Id. at 67.

31 Id. at 70.

32 Id. at 68.

33 Id. at 72.

34 See Chiappetta, supra note 4, at 107.

35 Id. at 107.

36 Yoshida, supra note 6, at 462.  The author concludes that the Supreme Court mistakenly defined the word "algorithm" to cover only mathematical algorithms.  In addition, the author contends that the method in Benson did not solve a mathematical problem.

37 Benson, 409 U.S. at 65.

38 Yoshida, supra note 6, at 463.

39 Parker v. Flook, 437 U.S. 584 (1978).

40 Benson, 409 U.S. at 68.

41 Flook, 437 U.S. 584.

42 Id. at 584.

43 Id. at 585.

44 Id. at 585.

45 Id. at 590.  Flook's contention is that the alarm limit adjustment is a physical post-solution activity, distinguishing the invention here from that in Benson.

46 Id. at 587.

47 Id. at 587.

48 Id. at 594.  Essentially, the court pointed out every element of the invention was well knonw, including the chemical processes, the monitoring of variables, the alarm limits, the recomputing of the limits, and use of computers for such monitoring.

49 Id. at 591.  The court emphasized that as part of the "point of novelty" approach, the algorithm must be assumed to be a "familiar part of the prior art."

50 Id. at 589.

51 In re Freeman, 573 F.2d 1237 (C.C.P.A. 1978).

52 Id. at 1238.

53 Id. at 1243.

54 Id. at 1243.

55 Id. at 1245.

56 Id. at 1245.

57 Id. at 1246.

58 Id. at 1246.

59 Id. at 1247.

60 Id. at 1242.

61 Id. at 1247.

62 Id. at 1247.

63 In re Walter, 618 F.2d 758 (C.C.P.A. 1980).

64 Id. at 767.  The court recognized that Flook created the possibility that an invention was nonstatutory despite not wholly preempting the algorithm, and felt the need to adjust the 2nd prong of the test accordingly.

65 Id. at 767.

66 Id. at 767.

67 Id. at 767.

68 Id. at 768.

69 Id. at 768.

70 Id. at 766.  Pointing out that the "point of novelty" approach would debilitate U.S. patent law, the CAFC was so brash as to actually say that it did "not believe the Supreme Court has acted in a manner so potentially destructive."

71 Flook at 591.

72 Diamond v. Diehr, 450 U.S. 175 (1981).

73 Id. at 175.

74 Id. at 181.

75 Id. at 181.

76 Id. at 181.

77 Id. at 181.

78 Yoshida, supra note 6, at 478.

79 Diehr, 450 U.S. at 184.

80 Id. at 188.

81 Id. at 188.  The court flatly stated that novelty was irrelevant to subject matter patentability.

82 Id. at 191.  According to the court, the unpatentability of mathematical algorithms cannot be "circumvented by attempting to limit the use of the formula to a particular technological environment.  . . . Similarly, insignificant post-solution activity will not transform an unpatentable principle into a patentable process."

83 Flook, 437 U.S. at 598.

84 Id. at 585.

85 Diehr, 450 U.S. at 186.

86 Id. at 191.

87 Yoshida, supra note 6, at 480.

88 In re Abele, 684 F.2d 902 (C.C.P.A. 1982).

89 Id. at 904.

90 Id. at 907.

91 Id. at 907.

92 Chiappetta, supra note 4, at 109.

93 Yoshida, supra note 6, at 480.

94 In re Grams, 888 F.2d 835 (Fed. Cir. 1989).

95 Id. at 836.

96 Id. at 837.

97 Id. at 836.

98 Id. at 838.

99 Id. at 838, 839.

100 Id. at 839.

101 Id. at 839 (citing In re Sarkar, 588 F.2d 1330, 1335 (C.C.P.A. 1978)).

102 Id. at 840.

103 See Abele, 684 F.2d 902.

104 Id. at 907.

105 In re Iwahashi, 888 F.2d 1370 (Fed. Cir. 1989).

106 Id. at 1371.

107 Id. at 1371.

108 Id. at 1373.

109 Id. at 1371.

110 Id. at 1375.

111 Id. at 1375.

112 Arrhythmia Research Technology, Inc. v. Corazonix Corp., 958 F.2d 1053 (Fed. Cir. 1992).

113 Id. at 1054.

114 35 U.S.C.  112 (1997).

115 35 U.S.C.  112.

116 Martin J. Adelman et al., CASES AND MATERIALS ON PATENT LAW 133 (1998).

117 Arrhythmia, 958 at 1054.

118 Id. at 1054.

119 Id. at 1058.

120 Id. at 1058.

121 Id. at 1058.

122 Id. at 1058, 1060.

123 Id. at 1058.

124 Id. at 1058.

125 Id. at 1059.

126 Id. at 1059.

127 Id. at 1060.  The court cited In re Sherwood for the statement that "'[t]he claimed invention ... converts one physical thing into another physical thing just as any other electrical circuitry would do'." 613 F.2d 809, 819 (CCPA 1980).

128 Id. at 1061.

129 Id. at 1061.

130 Id. at 1064.

131 Id. at 1064.  Rader argued that if no limits exist in the text of the statute, then none should be imposed.

132 In re Alappat, 33 F.3d 1526 (Fed. Cir. 1994).

133 Id. at 1537.

134 Id. at 1531.  The Board first reversed the patent examiner's rejection.  The examiner requested that the Board reconsider the decision (pursuant to a rarely invoked PTO provision), and further made the highly unusual request that the Board hear the issue as an expanded panel.  That expanded panel affirmed the examiner's rejection.

135 Id. at 1542.  The court recognized that, despite being provided with no substantial reason by the PTO for applying the "mathematical algorithm" analysis to a machine, the court's own precedent pointed toward that approach.

136 Id. at 1543.  The court emphasized that "at the core of the Court's analysis in each of these cases lies an attempt by the Court to explain a rather straightforward concept, namely, that certain types of mathematical subject matter, standing alone, represent nothing more than abstract ideas until reduced to some type of practical application."

137 Id. at 1543.

138 Id. at 1544.

139 Id. at 1544.

140 Id. at 1544.

141 Id. at 1544.

142 Id. at 1545.

143 Id. at 1540.  The court held that  112  6 should have been applied in making the subject matter determination.

144 Id. at 1545.

145 Id. at 1543.

146 Arrhythmia, 958 at 1064 (see note 125 above).

147 Alappat, 33 F.3d at 1544.

148 Id. at 1552.  Archer posited that the decision was "illogical, inconsistent with precedent and with sound principles of patent law, and will have untold consequences."

149 Id. at 1557.

150 In re Warmerdam, 33 F.3d 1354 (Fed. Cir. 1994).

151 Id. at 1355.

152 Id. at 1355.

153 Id. at 1360.

154 Id. at 1360.

155 Id. at 1362.

156 Id. at 1362 (citing In re Bradley, 600 F.2d 807 (C.C.P.A. 1979)).

157 Id. at 1362.

158 In re Lowry, 32 F.3d 1579 (Fed. Cir. 1994).

159 Id. at 1580.

160 Id. at 1580.

161 Id. at 1581.

162 Id. at 1581, 1582.

163 Id. at 1582.

164 Id. at 1582.

165 Id. at 1582.

166 Id. at 1583.  The court stated that the printed matter doctrine did not extend to this field.

166 Id. at 1557 ("[T]he mere association of digital electronics or a general purpose digital computer with a newly discovered mathematic operation does not per se bring that mathematic operation within the patent law.").

166 33 F.3d 1354 (Fed. Cir. 1994).

166 Id. at 1355 ("[The] application is directed to a method and apparatus for controlling the motion of objects and machines, such as robotic machines, to avoid collision with other moving or fixed objects.").

166 Id. at 1355.

166 Id. at 1360.

166 Id. at 1360.

166 Id. at 1362.

166 Id. at 1362 (citing In re Bradley, 600 F.2d 807 (C.C.P.A. 1979)).

166 Id. at 1362.

166 32 F.3d 1579 (Fed. Cir. 1994).

166 Id. at 1580.

166 Id. at 1580.

166 Id. at 1581.

166 Id. at 1581, 1582.

166 Id. at 1582.

166 Id. at 1582.

166 Id. at 1582.

166 Id. at 1583 ("[T]he Board erroneously extended a printed matter rejection under sections 102 and 103 to a new field in this case, which involves information stored in a memory. This case, moreover, is distinguishable from the printed matter cases.").

167 Id. at 1583.

168 Id. at 1583.  Arguing that the a computer is physically different after being programmed in a new way, the court concluded that "its memory elements are differently arranged. The fact that these physical changes are invisible to the eye should not tempt us to conclude that the machine has not been changed."  In re Bernhart, 417 F.2d 1395, 1400 (C.C.P.A. 1969).

169 Id. at 1580.

170 Yoshida, supra note 6, at 494.

171 Id. at 492.

172 In re Trovato, 42 F.3d 1376 (Fed. Cir. 1994).

173 Id. at 1377.

174 Id. at 1377.

175 Id. at 1378.

176 Id. at 1378.

177 Id. at 1379, 1380.

178 Id. at 1381.

179 Id. at 1382.

180 Id. at 1382 (citing In re Bradley, 600 F.2d 807 (C.C.P.A. 1979)).

181 Id. at 1382.

182 Id. at 1382.

183 Id. at 1383.

184 Id. at 1377.

185 Lowry, 32 F.3d at 1580.

186 In re Beauregard, 53 F.3d 1583 (C.A.F.C. 1995).

187 Chiappetta, supra note 4, at 120.

188 Beauregard, 53 F.3d at 1584.

189 Chiappetta, supra note 4, at 145.

190 State Street Bank & Trust Co. v. Signature Financial Group, Inc., 47 U.S.P.Q. 2d 1596 (C.A.F.C. 1998).

191 Id. at 1598.

192 Id. at 1598.

193 Scott M. Alter, Federal Circuit Broadens Scope for Software Patents, 15 NO. 10 Computer Law. 24, 25 (1998).

194 Alter, supra at note 192, 25.  The court further explained that the invention did not transform or convert any physical element (the algorithm was not tied to anything physical).

195 State Street, 47 U.S.P.Q. 2d at 1599.

196 Id. at 1599.

197 Id. at 1601.

198 Id. at 1601.

199 Id. at 1601.  The court derived the new test from the text of Alappat, in which the court had concluded that the rasterizer produced a "useful, concrete and tangible result."  33 F.3d at 1544.

200 Id. at 1601.

201 Id. at 1601.

202 Id. at 1602.

203 Id. at 1602, 1603.

204 Id. at 1601.

205 Id. at 1601.

206 Arrhythmia, 958 F.2d at 1062.  The concurrence in Arrhythmia emphasized the difficulty of implementing a rule that included numerous vague terms.

207 See Diehr, 450 U.S. 175.

208 See generally, Arrhythmia, 958 F.2d 1053, and Diehr, 450 U.S. 175.

209 State Street, 47 U.S.P.Q. 2d at 1601.

210 E. Robert Yoches, Recent Developments Concerning Software Patents, 532 PLI, Patents, Copyrights, Trademarks, and Literary Property Course Handbook Series 299, 306 (1998).  The author argues that "[i]n Abele, the C.C.P.A. explicitly analyzed the Diehr case and relied on it to form the 'Abele' modification of the former Freeman-Walter test: whether 'the algorithm be applied in any manner to physical elements or process steps.'"