Cite As: Seth A. Cohen,
To Innovate or Not to Innovate,
That is the Question: The Functions, Failures, and Foibles of the Reward
Function Theory of Patent Law in Relation to Computer Software Platforms,
5 Mich. Telecomm. Tech. L. Rev. 1 (1998),
available at <http://www.mttlr.org/volfive/cohen.html>.
Comments about this article should be sent to mttlr@umich.edu.
{1} The patent system has traditionally been
viewed as having two primary functions: the reward function and the prospect
function. Although these theories do explain some behavior which results
from the practical applications of the patent system, they also overlook
some behavior of the patent system which indicates a failure of these functions.
In order to properly prevent such failure, this paper proposes that the
patent system adopt an orientation that will lead to increased innovative
rivalry and competition. In Part I, using the computer operating system
software market as an example, I propose a framework for reconceptualizing
patent protection as it applies to software operating system platforms.
Part II briefly examines both the classical and neo-classical reward function
and prospect function theories.[2]
Part III defines the innovation market and describes the market dynamics
that create disincentives for innovation. These disincentives tend to limit
the number of competitors in innovation markets and create conditions which
reduce the effectiveness of the reward incentives to the extent that the
reward function fails in its entirety. Part IV examines the resulting harms
of this failure and identifies how reward function failure affects product
markets, which are dependent upon the reward function. Part V discusses
why the prospect function does not address the problems related to reward
function failure and the reasons that the promotion of innovative rivalry
would alleviate some of the problems. Part VI uses the relationship between
computer programs to illustrate the shortcomings of the patent system's
inability to prevent the problems created by non-competitive innovation.
This section also considers how the patent system might be better adjusted
to prevent reward function failure. Lastly, Part VI also proposes a series
of alternative frameworks for creating a competitively oriented approach
to the application of the patent system in the case of computer software
platforms.
{2} The genesis of the economic analysis of the
patent system is a result of the U.S. Constitution's initial economic assertion
related to the promotion of the "Progress of Science and useful Arts."[3]
Economists and legal scholars, when examining patent scope and innovative
improvements in light of this assertion, have long considered the patent
system as a framework serving two primary functions: the prospect function
and the reward function.[4]
Although theorists have used total innovative output as a gauge of the
benefits created by the patent system, much of the research in these areas
has used economics, specifically allocative efficiency, as a measuring
stick in determining the patent system's preference over alternative frameworks
that provide protection through trade secrecy.[5]
A. The Reward Theory
{3} The earliest and most widely accepted theory
related to the patent system is that it serves as a means of providing
rewards to those who develop and disclose ideas and products in the areas
of science and technology.[6]
Much of the basis for the patent system has been based on this theory.
The exclusive ownership rights are incentives for inventors to disclose
their inventions rather than withhold them as trade secrets.[7]
Both monopoly rights, as well as laws to protect the infringement of those
rights, allow the inventor to recoup the costs of research and development
as well as earn a fair amount of remuneration for the disclosure of the
idea.[8]
Ultimately, the reward function utilizes legal devices to create incentives
for technological innovation and disclosure by means of granting rewards
which are otherwise unrecognizable by the innovator. These rewards are
also essentially a quid pro quo exchange for the public benefit
based on both the innovation and disclosure of technology.
{4} Building on the general principle of the reward-oriented
function of patents, the reward theory has been split into two factions:
weak reward theory and strong reward theory.[9]
The key difference between the factions is that the supporters of the weak
theory view the reward function broadly and view its function as a general
inventive stimulant.[10]
In contrast, supporters of the strong theory support patent rights for
only those inventions which are "patent-induced.[11]
A third set of theorists have proposed that rent dissipation is the proper
theoretical framework for analyzing the way the reward function provides
incentives for innovation.[12]
B. Kitch's Prospect Theory{5} Edmund Kitch has made the seminal argument for the use of the prospect function in determining the patent system's development. Professor Kitch argues that a major function of the patent system is to maximize the efficiency of technological innovation.[13] Kitch identifies patents as "prospects," and argues that the general attributes of "prospecting" form the basis of inventive behavior.[14] Analogizing patents to mineral claims, Kitch reviews the policy implications of treating patents as prospects and presents numerous justifications for the supremacy of the prospect approach.[15]
{6} In his analogy, Kitch argues that the establishment
of a patent claim is similar to staking out a mineralization prospect.
Both mineralization claims and patents are limited based on breadth and
time and are based on a priority system that rewards the first to discover
and register the claim.[16]
Additionally, commercial significance is not required for either an enforceable
mineralization claim or for a valid patent. Although there are systemic
similarities, the key analogous characteristics of the mineralization claim
and patent systems are the incentives to stake and later exploit the prospects
and the exclusive rights which attach to valid claims. Once these exclusive
rights are granted, the holder can develop the prospect without fear that
other individuals will exploit the claim. Others are on notice to claim
other mineralizations or patents which do not overlap or infringe upon
the original stakeholder. Furthermore, both systems create a uniform framework
for the identification and exploitation of property rights, as well as
restrictions on abusive use of such rights.
{7} Analogizing the mineralization claim system
to the patent system led Kitch to establish several conclusions. Most significantly,
identifying a patent as a prospect establishes a future right to develop
the technological prospect. Kitch argues that unification of control over
research and development of the particular patent prospect provides the
most efficient means for developing the technology.[17]
Multiple and redundant inventive efforts may result in wasteful resource
usage or infringing developments.[18]
The disclosure elements of the patent system also serve to prevent wasteful
resource allocation. Under Kitch's model, the establishment of patent claims
does not merely provide information to the public; it also creates a mechanism
to direct the allocation of research and development resources away from
patented technology and towards pioneer innovations.
{8} Kitch concludes that when an inventor presents
a broad scale invention, the patent scope should be granted broadly in
order to allow for the coordination of research and development. This coordination
will prevent inefficient innovation and redirect inventive resources away
from competitive uses. While some scholars have criticized this conclusion,[19]
it has been cited as a significant argument against the development of
a competitive approach to technological innovation.[20]
{9} The prospect function analysis is consistent
with other allocative efficiency-based theories of the law.[21]
Nonetheless, the prospect function analysis fails to account for other
efficiency models based on non-monetizable qualities such as social efficiencies.[22]
Whereas competitive innovation may not be efficient in one economic model,
it may be maximally efficient in other models.[23]
Therefore, modeling the patent system's functionality on a particular efficiency
model incorporates certain value judgements related to the choice of that
model. Kitch, by modeling the patent system on monetizable types of allocative
efficiencies, expressly rejects social efficiencies and other alternative
welfare models. Thus, his analysis fails to account for other economic
approaches and, consequently, may deter the creation of beneficial efficiencies
which are created by these alternative models.[24]
{10} Despite the concerns related to Kitch's economic
modeling, Kitch's theory has significant implications related to the dynamics
of the inventive pool. The inventive pool, for the purposes of this paper,
is the group of innovators who are primarily focused on pioneering technological
innovation as a result of their endeavors in a particular technological
area.[25]
Innovation in this context is considered the search, discovery, development,
and implementation of new or adapted technologies.[26]
Despite Kitch's assertions regarding the benefits created by efficiencies
in the innovation process, it has been argued that Kitch's efficiency-based
approach may be appropriate if the inventive pool is significant enough
to withstand a decrease in the protection of incentive-based doctrines.[27]
If the inventive pool is small, Kitch's efficiency based doctrines may
further decrease the pool by concentrating the rewards on those who hold
the original broad or pioneer patents.[28]
{11} The reward and prospect function theories
do not explain all behavior in the area of technological innovation. Both
theories fail to recognize that other innovation market dynamics create
disincentives for pioneer inventive activity in markets which already have
a protected invention. These disincentives stem from the interaction of
commercial competitive advantages with intellectual property rights. When
the competitive advantages of the initial inventor are greater than the
rewards which are granted to a subsequent pioneer inventor (second-comer)[29]
by the patent system, the potential for optimal innovation is diminished.
I term the existence of these innovative disincentives reward function
failure.
{12} In reward function failure, potential second-comers
are discouraged from allocating inventive and financial resources towards
a particular innovative goal, even though competitive development may create
various market and consumer benefits. Such failure occurs in two contexts:
1) the limitation of the second-comer's ability to attain intellectual
property monopoly rights by legal doctrines, and 2) the second-comer's
limitation in gaining competitive advantages in the innovation market.
A. Ability to Attain Intellectual Property Rights{13} In apparent contradiction to the original intent, factors which discourage innovation are fundamental elements of the intellectual property rights granted by the government. Rights granted by the patent system provide both incentives and rewards for inventive efforts. However, in order to provide such incentives and rewards, the legal rights must provide protection from second-comers who are not deserving of their own legal protection. Nonetheless, both the length of the rights and the breadth of the rights may have unintended consequences which are different from their protective intent.[30] Rather than provide incentives for inventors to research and develop inventions diligently in one specific area so as to make inventive improvements or alternative uses, contemporary patent law serves to make potential inventors wary of innovation in an area in which patent rights have already been assigned. Although these second-comers may be able to develop and innovate products or improvements which satisfy the requirements of the Patent Act, some patent law doctrines may raise enough uncertainty to retard the inventors' efforts.
{14} The issue of patent scope is central to the
problem of inventive retardation because the determination of scope inherently
affects the number of inventions which may be found to be infringing.[31]
Depending on the breadth of the patent grant, inventors cannot sufficiently
rely on the certainty that their innovation will be recognized in its entirety.[32]
Additionally, the doctrine of equivalents adds uncertainty to patent scope
by expanding infringement beyond the literal meaning.[33]
Although the doctrine of equivalents prevents second-comers from usurping
rights of the patent holder, it is equally possible that second-comers
are excessively restricted in their attempts to attain certain legal rights
in new technologies.[34]
Therefore, both the fundamentals of patent scope and the definitions of
infringement may discourage the inventor's incentive for continuing development
and research of a discrete invention in a discrete technological area in
which pioneer patents have already been granted.[35]
B. Competitive Advantages of First Movers{15} Perhaps the most significant deterrent to second-comer innovation is the competitive benefits which accrue to first-mover innovators. These benefits, which are not without redeeming importance, nonetheless create advantages that are not easily duplicated for second-comers. In fact, certain competitive benefits grant advantages which even the patent system may not grant, especially in terms of time-related benefits and mechanisms for market control. The two most significant competitive benefits are first-mover advantages and compatibility/network externalities.
{16} First-mover advantages are established, not
by being first to invent, but by being first to successfully introduce
the product to market.[36]
Therefore, first mover-advantages do not accrue based purely on innovative
efficiency; they accumulate as the result of a combination of research
and development advantages and historical market reactions.[37]
{17} The initial advantages that result from early
exploitation involve learning benefits.[38]
These learning effects, which have the impact of establishing cost advantages,
are primarily relevant to creating research and development advantages
for future innovations.[39]
The sooner an inventor enters the learning curve, the less pronounced the
curve will be for future innovative developments.[40]
Furthermore, although the learning advantages may vary in degree, the first-mover
gains a sizable advantage when the learning curve is moderate.[41]
{18} Although the learning benefits may provide
first-movers with some competitive advantage, these advantages are more
damaging to innovation incentives when they are supplemented by early market
introduction and exploitation. Early exploitation of a product with no
marketable substitute and the subsequent growth of an installed base creates
competitive advantages which shape later development.[42]
Because of the lack of substitutability, the market is likely to adopt
the first invention more readily than it would if there were competitive
products. This market adoption creates switching costs which may, if too
high, effectively lock-in consumers to the installed base.[43]
{19} The problem with the installed base from the
second-comer's perspective relates to establishing market adoption of the
substitute goods. Foremost, if the installed base is significant in size,
the lesser likelihood of a successful market entry by a second innovation
might deter a second-comer from focusing on developing a rivalrous product.
Although a second-comer may be potentially successful in introducing a
discrete invention which is substitutable to that of the first-mover, it
is only in rare cases where a second product successfully overthrows a
large installed base.[44]
Therefore, second-comers are likely to be deterred from innovating on the
scale of the original developer. Follow-up research and supplemental improvement
become the inventive goal, rather than the development of pioneer research
and discrete substitute product innovation.
{20} The second competitive advantage which deters
second-comer rivalrous innovation is the system of network externalities
and compatibility effects which attend to a large installed base, especially
in computer and communications technology.[45]Network
externalities, as defined by Michael Katz and Carl Shapiro, are "the utility
that a given user derives from the good depend[ent] upon the number of
other users who are in the same 'network.'"[46]
These benefits are usually supplemental to installed base advantages, but
the two often forge a symbiotic relationship. The greater the installed
base, the greater likelihood that the network externalities will attract
more users to the installed base. Network externalities are not only limited
to positive effects, but may also generate negative or pecuniary effects.[47]
For a second-comer innovator, the introduction of a rival discrete innovation
(which is not an improvement, but rather a substitutable innovation) will
likely not produce sufficient network externalities to attract new users.
Thus, the first innovation to market has significant advantages in expanding
the externalities past the point of competitive vulnerability.
{21} While network externalities do not discourage
innovation entirely, their existence does affect the way second-comers
identify their paths of innovation. The relevancy of path dependence theory
in the technical innovation is based on the principle that small-scale
advantages of a particular technology can create substantial influences
on the market dynamics of that technology.[48]
The theory holds, in the case of technological innovation, that a substitutable
rival does not easily replace a technology with a large installed base.[49]
{22} In order for network benefits to flow through
to second-comer innovations, there must be a degree of compatibility which
allows the second-comer's innovation to complement, rather than supplant,
the innovation upon which the installed base is predicated.[50]
As a consequence of this necessary compatibility, a second-comer will likely
be deterred from developing a non-compatible equivalent product.[51]
This discouragement from innovation may restrain the introduction of innovations
which may, on their own merits, be as equally marketable as the first-mover's
product was at the time of its introduction or even after improvement.[52]
Therefore, it is arguable that the first-mover's mastery of the product
market's relevant externalities may deter an innovator from choosing such
a product to innovate.
{23} Although I have identified why certain
legal, economic and market factors deter entrants from competitively innovating
similar products, it is important to understand how these disincentives
are injurious to the development of primary and secondary product markets.[53]
The injury manifests in three different ways: 1) products are introduced
at a stage of innovation which creates a sub-optimal product, 2) primary
product markets which are dependent on intellectual property rights have
few large competitors who tend to have significant market power in their
product market, and 3) consumer welfare is harmed by limited choice in
the primary product markets.
A. Sub-Optimal Innovation{24} In markets in which there is little or no competition, there is a tendency to innovate sub-optimally.[54] Consequently, the product development may occur inefficiently once there is no threat of improved substitutable products.[55] Similar concerns exist in relation to product innovation prior to market introduction.[56] Nonetheless, unlike the improvement market in which there is already a baseline product, the members of the inventive pool do not have a baseline product upon which to base their innovation.[57] As mentioned above, it is possible, because of the first-mover advantages, that an inventor will attempt to establish legal rights to his product and introduce his product to market prior to the optimal technological innovation, or optimally commercial innovation. Although Professor Kitch has argued that this prospect function is the most efficient way to analyze the patent system, many notable scholars disagree.[58]
{25} A negative consequence of the prospect analysis
is that a product which is sub-optimal usually gains enough competitive
advantages to deter the innovation of superior substitute products.[59]
Therefore, the prospects which may have the strongest market presence may
also be the products which bring the least-optimal effectiveness or use.
A greater number of competing innovators would arguably force the most
optimal innovation of the baseline product, upon which successive competitive
efforts would create optimal improvements. Nonetheless, without such rivalry
in the innovative process, inventors are likely to attempt to commercialize
innovations as early as possible, believing that the first-mover advantages
will compensate for any disadvantages that a sub-optimal product may create.[60]
B. Reward Function Failure: IP -Dependent Markets Have Fewer Competitors{26} The lack of competitors in certain IP-dependent markets results from the deterrence of innovators to develop substitute primary products.[61] Therefore, instead of developing rival primary products, innovators choose to develop secondary products that are compatible to the unrivaled primary products.[62] The secondary market of compatible products may be competitive, while leaving a relatively non-competitive primary product market. The increase in size of the compatible secondary product market has the effect of reinforcing the primary product's dominance and the primary product's dominance has the effect of redirecting inventive resources to the development of compatible secondary products.[63] I term this redirection of inventive resources "inventive tipping."
{27} Inventive tipping is similar to consumer tipping,
which occurs as a result of network externalities. Whereas network externality
based tipping occurs in the dynamics of the user/consumer market, inventive
tipping occurs in the innovation market. Tipping occurs generally when
a baseline product (or in this context, a primary product) has a large
installed base which creates significant network externalities.[64]
These externalities/benefits become so great that the use of the product
is widely accepted based on the product's inherent use, as well as the
sum of its externalities. Essentially, the market has tipped over
to the side of the product with the greatest externalities.[65]
Inventive tipping occurs in a similar manner, with the exception that acceptance
of the product and its network benefits are examined from the inventor/supplier
side, as opposed to the user/consumer side.[66]
For example, there comes a point where a primary product is sufficiently
adopted by the market to reasonably indicate that it will continue to have
a significant market presence.[67]
Once inventors or developers determine that the baseline product is the
more accepted product (and the one most likely to accrue network benefits),
they will tailor their products to be compatible with that product, essentially
adopting the product as a standard.[68]
An example of such tipping can be found in the production of videotapes.
When tape providers had readily determined that VHS recorders had sufficiently
overtaken the market from rival Betamax recorders, producers sold videos
in VHS format.[69]
This tipping, and the consequent reduction in sales of Betamax videos,
helped to further increase sales of VHS recorders and strengthen the VHS
installed base.[70]
{28} Although the inventors/producers are the ones
who effectuate the supply-side dynamics of the inventive tipping, it is
closely related to demand-side causation.[71]
Compatibility is a characteristic of technology products which affects
consumer choice and helps define the development of markets. Consumers
are less likely to demand non-compatible goods when there is a substitution
compatible good.[72]
To respond to this demand, inventors and producers allocate more of their
resources to satisfy this consumer demand. Although non-compatible secondary
goods may be optimally developed and designed, their incompatibility may
effectively negate any consumer desirability.[73]
{29} The sum result of this inventive tipping and
compatibility demand is that first-mover innovations can narrow product
development simply by establishing an installed base of their initial primary
product innovation. Even if second-comers attempt to innovate around a
first-mover's product, the second-comer may find that the ancillary compatible
goods markets which developed around the first-movers baseline good have
the effect of suffocating the competitive ability of the second-comer's
product. Secondary market developers further strengthen the installed base
of the baseline product and establish a new set of compatibility-based
network externalities.
C. Consumer Welfare Implications{30} A third problem related to non-rivalrous innovation is the impact on consumer welfare. Apart from the economic efficiencies, the issues pertaining to the monopoly powers of intellectual property rights holders and the attendant concerns of optimal patent life, consumers may suffer as a result of 1) sub-optimal development of the useful sciences, and 2) monopolistic control of the development of the useful sciences.
{31} Although it has been argued that intellectual
property rights should be granted only for those innovations which appear
to be optimal, the Patent Act only demands novelty, utility and nonobviousness.[74]
The Constitution only grants power to Congress to promote the "useful Arts,"
not only the most useful.[75]
Nonetheless, the underlying goal of the clause is to maximize the public
welfare by allowing Congress to establish a system that will encourage
innovations to be shared with the public.[76]
For example, the Patent Act already sets a bar that the innovation must
be useful.[77]
Although this requirement has been severely weakened, it is evidence of
one of the core values of the intellectual property system; useful developments
are favored over frivolous and non-useful developments.
{32} An intellectual property system that encourages
sub-optimal innovation and discourages individuals from innovating is inconsistent
with the goal of promoting useful sciences generally. Although the patent
system does reward innovation, it rewards innovation only up to a point,
and then, as discussed in the above sections, discourages further innovation
that may in fact be optimal. Consequently, the systemic disincentives are
contrary to both the letter and the spirit of the Constitution and fail
to protect the public welfare as it was originally construed.
{33} Another implication for consumer welfare is
the lack of competitive choice that results from the innovation disincentives
which limit the size of the primary product market.[78]
Assuming that a first-mover's installed base is significant enough to establish
inventive tipping, a consumer may find that the differential between the
externalities of the products gives her no real choice in determining which
technology she will use.[79]
Rather than choosing among competing substitutable products within the
same field of technology, she may find the need to choose between very
different technologies because no innovator developed an intermediate substitute
product.[80]
Furthermore, the same consumer may find that she has no substantial choice
because the secondary compatibility markets make the choice of the non-dominant
technology too costly.[81]
A system which fails to encourage competitive innovation may result in
protecting innovators while passing along limited and restrained choice
in the use of such useful sciences, thus harming consumer welfare.[82]
{34} As the above assessment of the problems which
result from reward function failure indicates, the disincentives to innovate
competing primary products has the effect of limiting technology and market
development and harms consumer welfare. Although these problems have a
significant impact on the dynamics of the inventive pool, and further exacerbate
reward function failure, both the disincentives and resulting problems
are ignored by the prospect function theorists.
A. Inadequacies of the Prospect Function{35} The prospect theory has already been criticized as understating the benefits of competitive rivalry. Professors Merges and Nelson have criticized Kitch's prospect theory, arguing empirical evidence suggests that competitive development of technological prospects is preferable to coordinated development.[83] They address the limitations of Kitch's mining model analogy and argue that Kitch's model ignores the tendency of prospectors to exploit their prospects inadequately, as well as the likelihood that rightholders will sub-optimally develop potential improvements.[84] Additionally, they suggest that both the legal and theoretical limitations which are created by property rights also undermine the effectiveness of the prospect theory. In this argument, Merges and Nelson refute Kitch's expansive view of property rights.[85] Doubting that unrivaled rightholders would exploit their rights as actively and efficiently as they would in a competitive environment, they are wary of expanding the property rights and endorse rivalry in the innovation process. Additionally, unlike Kitch, they provide historically-based empirical evidence which supports their theory.[86]
{36} In addition to the limitations of the Kitch
model stated by Professors Merges and Nelson, Professor Kitch's theory
is also limited in that it does not anticipate the interrelation of several
discrete prospects.[87]
Kitch's model anticipates that each prospect, like a mineraliztion, is
a singular and discrete claim that can be exploited without any interrelation
to a separate discrete claim.[88]
This analysis breaks down when one prospect is relied upon by other prospects
for purposes of compatibility and interoperability. Whereas Kitch anticipates
that the prospects are primarily exploited in a manner which improves the
original broad claim, he fails to foresee the impacts the exploitation
of that prospect will have on the behavior of other prospectors. The behavioral
impacts occur in two ways: 1) increased innovative activity in areas of
technology which are compatible to the original claim, and 2) increased
activity in the development of compatible technologies.
{37} To borrow Kitch's mineral analogy, when a
prospector located mineralizations in a certain area, the usual result
was increased prospect activity in that particular area. This increased
activity did not affect the prospector's original claim, so long as he
had filed the proper claims. Nonetheless, although there may have been
other areas with greater amounts of mineralization, the tendency was to
prospect the region which had already proved to be mineralized.[89]
The greater the permitted claim sizes, the fewer the prospectors could
benefit from the find. Although it was understood that a prospector could
develop his own prospect, it was uncommon to find one claim holder organizing
the exploitation of the entire region, even if he was the first to locate
the mineralized area.[90]
Each prospector could exploit his particular claim in any manner, creating
rivalry in the development of claims. Consequently, Kitch's theory overlooks
the fact that this rivalry in exploiting the prospect is deterred when
the permissible claim size is too big.[91]
{38} The patent system works much the same way.
When patent rights are claimed in a particular technology, and that technology
is adopted by consumers, there is increased prospecting activity in that
technological field.[92]
Innovators prospect compatible technological innovations and claim patent
rights. The rivalry in developing these complementary or compatible products
helps create optimal innovation and consumer choice.[93]
If the original patent holder was granted too expansive a claim, there
would be no ability to establish any technological prospects, nor would
rivalrous innovation occur among compatible products because the patent
holder's rights would foreclose the development of all compatible technologies.[94]
{39} Secondary innovations that evolve from the
primary prospect are also foreclosed in Kitch's theory. This foreclosure
can be exemplified by the mineralization theory as well. Rather than using
mineralization generally, let us use a high-nitrate mineralization such
as limestone, which is a type of mineralization that enhances the soil
for cultivation. A prospector searching for mineralizations of limestone
could make a claim for the area in which the mineralization could be found.
Nonetheless, there are limitations to the property rights which he claimed.
He has no rights in determining how the surface land is used. Therefore,
if a farmer determines that the land is highly arable, the prospector has
no ability to prevent the farmer from using the land above the prospect
for farming. Significantly, the claim holder has the ability to extract
the minerals but not to control the information regarding the location
of the mineralization, which is useful in determining the most arable land.
{40} In patent theory, the information pertaining
to the claim is also released to the public by means of public disclosures.[95]
But unlike the limitations in the mineralization example, a patent holder
may be able to control the use of the information that is not expressly
granted by the patent. If the patented technology gains numerous competitive
advantages, the patent holder may be able to control the entire area of
technology, thus limiting the prospecting of the technology for alternative
or rivalrous purposes.[96]
Therefore, the distinction in the analogy, which Kitch overlooks, is that
patent law lacks the division that exists between mineral rights and real
estate rights. Rather, patent law creates a bundle of rights, which affects
the foreclosure of certain innovation. Whereas in the mineral prospect
analogy, a mineral rights holder cannot foreclose competition for the real
estate in the area of the claim, a patent holder does have the ability
to foreclose competition in the relevant innovation market.
{41} Nonetheless, the foreclosure of competition
in the innovation market is not likely to occur in every technological
field.[97]
It is more likely to occur when there are two innovation markets which
are closely related and contain interconnected products.[98]
These markets, which contain the primary and secondary products, are those
which are most susceptible to the harms of innovation disincentives. Because
of the symbiotic relationship of the primary and secondary products, either
a decrease in innovation or an occurrence of inventive tipping in one market
has significant effects in the development of the other market. Therefore,
one way to address some of the problems resulting from reward function
failure is to spur innovative competition. The following section addresses
the theoretical framework of why encouraging innovational rivalry can be
an important and functional element remedying reward function failure.
B. Solving Reward Function Failure: Rivalry in the Innovative Process{42} Numerous theorists have offered alternative means to moderate the problems which develop as a result of non-competitive innovation. Adjusting both patent life and scope has been suggested, [99]as has instituting compulsory licensing.[100] Therefore, much of the literature focuses on altering the elements which are generally considered part of the reward function. Rather than modifying the length and scope patent grants, I propose the patent system should be modified in a manner which remedies reward function failure by modifying the nature of the patent grants and technologies for which they are granted. Whereas the modification of patent scope affects the development of follow-on innovations and improvements, it does not address the disincentives to innovate primary products. Overbroad patent scope may deter innovation, but modifying the scope will not remove all of the deterrence created by reward function failure.[101] Rather, the promotion of innovation rivalry, by modifying the application of patent grants, helps counter the systemic problems of reward function failure, and consequently promotes, rather than discourages, innovation.
{43} Competitive innovation also prevents efficiency
benefits in the innovative process and protects consumer welfare. In recent
years, the use of rivalry as an economic policy goal has been subordinated
by the use of efficiency-based standards of competition.[102]
Nonetheless, there have been some attempts to establish rivalry as the
dominant goal in economics-based law.[103]
This resurgence of rivalry, based largely on the arguments that rivalry
creates efficiencies in innovation, helps eliminate X-inefficiencies, and
increases consumer welfare.[104]
{44} X-inefficiencies occur when firms do not face
regular competition in their industry, and thus do not face the same production
pressures which occur in a fully competitive market.[105]
Although these inefficiencies are normally considered to exist primarily
in the production process, there is evidence that such inefficiencies may
also exist in the innovative process.[106]
In fact, there may be more to lose from X-inefficiencies in the innovative
process. Rather than higher unit costs, the harm may be sub-optimal innovation
that may have numerous multiplying effects in the innovation process.[107]
Although a thorough evaluation of the existence of X-inefficiencies is
difficult,[108]
some empirical evidence has illustrated that rivalry limits the existence
of innovative characteristics that are similar to X-inefficiencies.[109]
{45} Rivalry also enhances consumer welfare by
promoting innovative efficiency and choice. The specific determination
of what constitutes consumer welfare is varied.[110]
In the innovation context, optimal innovation at the lowest cost is beneficial
to consumer welfare.[111]
By increasing competitive tensions, rivalry forces innovators to maximize
their efforts so as to withstand the rigors of the market.[112]
Although the costs which are incurred by increased innovative efforts may
reduce some allocative efficiencies, innovative efficiencies increase the
aggregate consumer welfare benefits above those created by the allocative
efficiencies.[113]
Consumers benefit foremost from increased choice among innovations of variable
price and quality.[114]
Another benefit is that rivalry creates the incentive for numerous competitors
to explore innovative development in areas which they would not otherwise
explore, thus creating the potential for even greater innovative efficiencies.[115]
These new innovations create even more consumer choice. Therefore, the
promotion of rivalry and innovative efficiencies created by such rivalry
provides benefits exceeding pure industrial-based allocative efficiencies.[116]
{46} Although rivalry appears to provide reasonable
benefits that create both efficiencies and public welfare, introducing
rivalry to innovation markets raises significant questions related to the
nature and scope of intellectual property protection. Specifically, the
type of patent protection, if there is any to be granted at all, may depend
greatly on the type of technology.[117]
Additionally, the close relationship between primary and secondary technologies
and products may also affect the nature of the protection to be provided.
Therefore, when determining the nature of patent protection in light of
the policy goal of innovative rivalry, the technologies which raise the
greatest number of problems are those which have 1) a high propensity for
product success based on the aggregate network externalities which the
product creates, and 2) a proximate relationship between primary and secondary
products. One technology which matches these criteria is computer software.
The following section reviews the dynamics of the computer software market
and proposes a framework for evaluating potential types of patent grants
which may best enhance innovative rivalry and prevent reward function failure
in this particular technology.
A. Technical Aspects of Computer Software{47} Computer software programs, at a very basic level, can be divided into two categories: 1) operating systems and 2) applications.[118] Each of these categories provides a different function in computer processing but are co-related. Although the division between what constitutes an application and what constitutes a function of an operating system has been blurred, for the initial purposes of this discussion it is important to understand the distinctions between these types of programs. [119]
{48} An operating system is the program which coordinates
all activity on a computer, essentially providing a basic roadmap for the
functionality of the hardware.[120]
Its key tasks are to recognize and control input and output functions,
manage file directories, and manage the behavior of the external peripheral
functions of the computer.[121]
Additionally, the operating system manages the processing schedule of the
microprocessor and controls resource usage.[122]
{49} One of the most essential attributes of the
operating system is the software platform.[123]
The platform is the basic framework which is used to run application programs.
For an application to run successfully on a particular operating system,
the program must be designed in a manner that makes it compatible with
the operating system's software platform.[124]
Normally this design is based on the use of a common set of application
programming interfaces (APIs) which allow the application to communicate
with the operating system in the same language and can create compatibility.[125]
Therefore, the development of a platform and its APIs is a necessary element
for the development of applications.
{50} Applications are programs that create end-user
functionality,[126]
such as word processors, spreadsheets, and design programs.[127]
Applications are essentially high level sets of commands that use the underlying
computer processing ability to serve specific functions. Unlike operating
systems which manage system resources, applications are the higher level
products of those resources. As stated above, all applications contain
Application Programming Interfaces (APIs) which communicate with the software
platform. Without APIs, applications will not initialize or function with
the software platform, will be unable to request system resources, and
will possess no independent use in regard to that particular computer platform.
{51} Therefore, in light of platforms and applications,
APIs are essentially the technical program specifications which need to
be present for applications (which in this case constitute secondary products)
to run properly on a platform (the primary product). The designer of the
operating system/platform determines the APIs. An application designer
can use any API of his choice, but if he intends for the application to
be run on a specific platform, he needs to use those platform-specific
APIs.[128]
B. The Special Role of APIs and Software Platforms in the Application Marke{52} When a software programmer designs an application program, he must make a choice as to the software platform on which the program will run.[129] This decision is likely to be affected by numerous factors. Foremost, the developer will want to insure that he chooses the operating system which will maximize the commercial success of his application. Therefore, the developer will most likely choose the operating system with the highest commercial success or is popular in the market in which the application is most likely to be used.[130] The network externalities of the most successful operating system provide powerful incentives for the application developer to choose that particular system over others both because of its strong market position and because of the consumer benefits stemming from the network externalities. From the application developer's viewpoint, the application's success is partially dependent on the continued success of the operating system. A commercially unpopular operating system, although technically superior, may neither have sufficient network benefits nor the potential network benefits to attract applications.[131] Therefore, the choice of operating system, from the developer's perspective, is more dependent on market dynamics rather than technical functionality.[132] Although other factors may affect the developer's decision, empirical evidence indicates that developers tend to design applications which are compatible with the most successful system on the market, even if that system is not the most technically superior alternative.[133]
{53} Once the choice of operating system has been
made, the developer must acquire the specific APIs which will create interoperability.
The proprietary nature of the APIs thus creates the transactional issues
which are raised in application development. In almost all cases, the programmer
will need to undertake measures to ensure that the application contains
all the elements necessary to run on the specific platform. These measures
usually consist of the licensing of patent or copyright rights as well
as the licensing of trademarks.
C. Innovation Market Implications of Technical Aspects of Computer Software{54} As noted above, a significant implication of the relationship between operating systems and platforms is that the development of applications depends largely on the software platforms of operating systems. The market success of operating systems consequently has an impact on the development of application programs. Those developers who seek to maximize the commercial success of their application must choose compatibility with the most commercially successful operating system. Alternatively, developers of alternative operating systems will find increased difficulty in attracting application developers to choose compatibility with rival systems. The circular result here is that the growth of the compatible application innovation market further expands the competitive advantages of the particular operating system, and the enhanced competitive advantage of the operating system encourages growth in the compatible application innovation market. The number of rival operating systems is limited by the innovation in the application market. Therefore, to encourage rivalry in the operating system innovation market, one of the roots of the dynamic must be addressed - the intellectual property protection of software platforms.
D. Preventing Failure: Reevaluating Protection for Software Platforms[134]{55} In order to promote rivalry in the development of operating systems, the special nature of software platforms should be reflected by the intellectual property protection which is granted. The key question is how platforms should be protected considering their special nature. Tailoring a type of protection for platforms raises a host of questions, including the determination of what constitutes a software platform and what, if any, proprietary protection should be granted. The answers to these questions raise the possibility that the best way to protect the market regulation function of the patent system in regard to operating systems is to design sui generis or hybrid protection for software platforms.
{56} The first question relating to the special case of computer platforms is how to define the platform itself. One way to perceive the software platform is to recognize it as the central component of the operating system. If this is the case, then the operating system itself may be the property which needs protection. Alternatively, if the software platform is merely one discrete element in the operating system, without which the operating system would still serve a majority of its functionality, perhaps only the code which makes up the routines involved in the software platform should be considered as the platform.1. Proprietary Protection: Determining the Scope
{57} The latter alternative raises another significant
question related to which portions of the actual code should be affected
by alternative treatment of platforms. Not all of the code which creates
the software platform is specifically assignable to the interoperability
function of the platform. Portions of the platform code relate to the functions
of the user interface, and may direct system resources toward these functions.[135]
Although applications usually manipulate the user interface, the platform
may synthesize the application interface commands with the operating system's
user interface commands. Therefore, when determining the relevant platform
code, a distinction may be drawn between the code which creates the interoperability
functions of software platforms and the code which creates or directs the
user interface.
{58} However, drawing such a distinction may raise
problems because interface compatibility may be a significantly desirable
commercial attribute of applications. For example, when using the Microsoft
Windows platform, a computer can seamlessly combine the Windows interface
with the interfaces of applications designed to be compatible with Windows.
Conversely, when a user attempts to run a MS-DOS application, the computer
must reset the user interface to DOS standards. This reset usually results
in the computer switching interfaces during the time that the non-Windows
application is being used. Although the change in interface may not have
significant functional implications, it does alter the seamless interface
created by Windows-compatible applications. Nonetheless, determining the
distinction between interoperability code and interface code may be a significant
issue in determining the components of a software platform.
{59} Although interface compatibility may not be
as technically significant as functional interoperability, dividing the
two may have significant commercial effects. By separating the platform
code in such a manner, it is possible that application developers could
develop functionally interoperable software. However, consumers would still
not respond to the applications because of the interface incompatibility.
Thus, in this case, excessively limiting the boundaries of the operating
system could have drastic commercial effects.
{60} The aforementioned problems all raise one
significant issue: the boundaries of the operating system are not easily
determined. Perhaps technological innovation will make this distinction
clearer. Recent events have highlighted the fact that operating systems
can perform much more functionality than basic organizational processing
decisions and incorporate new functions which are more commonly associated
with software applications. If this is the case, although it may be more
difficult to distinguish the applications from the operating system, it
may be easier to distinguish the software platform from the rest of the
application-type functions of the operating system.
{61} In addition to the questions regarding scope are the questions related to defining the proprietary nature of software platforms. As a result of their importance in the innovation markets of applications and operating systems, software platforms have several attributes which make it difficult to define their proprietary nature. In fact, software platforms have more similarities to non-proprietary concepts such as languages, algorithms, and industry know-how. Therefore, one possible alternative is to determine whether or not the software platform elements of an operating system are unprotectable because of their underlying importance to the functionality of all other software.2. Defining the Proprietary Nature of Software Platforms
{62} This alternative is rooted in the argument
that software platforms serve functions similar to those of algorithms,
scientific principles and languages. Although software platforms are not
inherently self-limiting in their expression, they are the principal building
blocks to software interoperability. As a result, the same justifications
which are used to deny patentability to the aforementioned concepts apply
to software platforms. For example, an algorithm is not proprietary because
it is a mathematical function and not a machine or process and therefore
is unpatentable under the Patent Act.[136]
The algorithm, as a mathematically created function, is inseparable from
the larger structure of mathematical science.[137]
By appropriating an algorithm, a patent holder could exclude others (who
do not license the algorithm) from performing mathematical functions which
may depend on the use or variation of the algorithm. This exclusion not
only limits the development of mathematical theory, but also limits the
development of technologies which could use the algorithm to perform processes
or mechanical functions.
{63} Similarly, a person appropriating a software
platform would have the ability to foreclose the development of other software
programs and functions. But substitute platforms, market dynamics and innovation
disincentives may make the development of such a substitute less likely.
By excluding others from the software platform's basic interoperability
function, a patent holder would be denying all software developers half
of the equation necessary for the development of functional software.[138]
Essentially, the interoperability function of a software platform transcends
a mere idea or process. It is the principal means of operating software,
just as algorithms are a principal means of mathematical analysis. Therefore,
rejecting patent protection for software platforms would be both theoretically
and legally consistent.
{64} The alternative that software platforms be
considered unprotectable does raise a difficult question of what platform
components can still be protected by other intellectual property protection.
For example, if the APIs are the components that are essential to interoperability,
it is possible that the remainder of the platform may still qualify for
other types of intellectual property protection such as copyright. A careful
distinction must be drawn between these components so that the lack of
protection of the platform will still provide incentives for platform innovation
while reducing the harms that result from the exclusivity of proprietary
rights.[139]
{65} The exclusion of the software platform's compatibility
components from patent protection removes two significant barriers to the
development of rivalry in the innovation of software. First, the relationship
between applications and platforms will cease to be a limiting factor in
the development of software platforms. Application developers will not
need to make a choice regarding platform compatibility because all of the
APIs and other compatibility information will be costless. Therefore, a
developer will be able to acquire all APIs for use in an application and
not be limited by compatibility costs. Since application developers will
not need to make such a choice, the potential for inventive tipping will
be mitigated. Consequently, platform developers will no longer be subject
to a significant disincentive for inventing a rival software platform.
{66} The second barrier which is removed is that
the innovation market dynamics will be altered to recreate benefits which
would otherwise be created by the reward function. Although the lack of
rewards created by patentability may be dissipated, rival developers may
have other sufficient rewards for rivalrous development. If the non-compatibility
related components are still protected, a developer may have sufficient
justification for developing a new operating system based on an unprotected
software platform. Specifically, the benefits received from developing
a new operating system may outweigh the lost benefits of releasing a new
and improved non-proprietary platform component. Application developers,
no longer limited by inventive tipping, will be able to easily enhance
the marketability of any new software platform because they can add compatibility
easily and without cost. Therefore, operating systems will be a stronger
market competitor regardless of whether or not it is the first to market.
{67} Essentially, by leaving software platforms
unprotected, the first-mover advantages are significantly limited, and
the disincentives they create are removed. As a result of such a removal,
second comers are not subjected to the dynamics of reward function failure.
Rivalry among platforms will develop because path dependence and lock-in
will no longer limit the potential for second-comer success. Rather, the
open aspects of software platforms will create more choices for platforms
without sacrificing application compatibility.[140]
As a result, leaving software applications unprotected will create competition
in both the innovation market as well as the product market, and will provide
a remedy for the harms of reward function failure.
{68} The creation of sui generis protection for computer programs has been a subject of much academic discussion in recent years.[141] These arguments resonate in the particular situation of software platforms. As noted above, the current protections available for computer programs do not prevent reward function failure in the operating system innovation market. Consequently, alternative protections must be designed to better protect innovation market competition. These protections may be either modified proprietary protections, compulsory licensing requirements, or a limited general public license requirement. Although these three protections are diametrically different (one grants property rights, whereas the other two create a type of public good), they all serve the function of creating competition in the innovation market.3. Potential Sui Generis Protection for Software Platforms
{69} An example of a modified proprietary protection
for software platforms is a patent system which leaves certain components
unprotected, while designating other components to be patentable. Another
alternative is the creation of an altogether new protection which is a
hybrid of patent and copyright laws.[142]
For both alternatives, the proprietary rights that are created would need
to be sufficiently limited to prevent reward function failure in the innovative
market, but be sufficiently broad to promote innovative rivalry. Nonetheless,
the sui generis protection should be limited only to those software
platform components which create compatibility (e.g. APIs) and should not
necessarily be extended to other non-essential components.
{70} The second alternative, compulsory licensing
of the compatibility elements, is not novel, but it does demand a closer
examination in regard to software platforms. The compulsory license may
be limited only to those elements of the operating system which compose
the software platform. The remaining portion of the operating system code
will retain the attributes of a patented invention. Again the difficulty
in this hybrid may be in determining the make-up of the platform. This
problem, however, does not make it impracticable. The compulsory license
may be structured so that the actual design of the compatibility component
need not be licensed, but the code needed to make applications compatible
with the platform is subject to a license. Such a license would prevent
the duplication of the platform but would force the function of the platform
to be licensed. This license is similar to the "Automatic Anti-Cloning
Protection Followed by an Automatic Royalty-Bearing License" as discussed
by Professor Samuelson.[143]
{71} Compulsory licensing does not completely cure
reward function failure and the disincentives for rivalrous innovation.
In one respect, the compulsory licensing element will prevent reward function
failure by limiting the excess competitive advantages of a first-mover.
Competing innovators will be allowed to incorporate rival software platforms
into their operating system. The overall result of this practice will be
to create cross platform compatibility by default.
{72} Nonetheless, the compulsory licensing framework
does not solve the problems related to inventive tipping. Although application
developers would have the platform's compatibility codes available for
licensing, the developer would still need to make cost-based decisions
regarding the choice of platform. Assuming application developers cannot
license every platform code (unless the licensing fees are insignificant),
the application market will still tilt towards one particular platform.
An alternative would be to create a license which covers all the software
platform compatibility components of all the rival software platforms.
Such a universal license fee could then be split among all the software
platform owners covered by the license. This license is similar to a system
proposed for software innovations based on registration and automatic licensing.[144]
Furthermore, this alternative might solve the cost-based problems associated
with the compulsory license, thus preventing a significant innovation disincentive.
{73} The last type of alternative to a modified
proprietary system is a non-proprietary general public license (GPL) system
for software platforms similar to the one developed by the Free Software
Foundation.[145]
The GPL is a hybrid of contract and copyright law, creating proprietary
rights while also granting a license which allows the modification of the
copyrighted work. The hybrid's goal is to further development of the material
covered by the GPL while preventing its appropriability.[146]
Although arguments against such a private propertization of intellectual
property rights exist, these arguments are based on the claim that the
overall benefits to society are based on the underlying principal of property
rights. Contrary to this argument, in the context of software platforms,
the GPL may produce optimal and beneficial results.
{74} In such a system, the software platform (or
at least its compatibility components) would be subject to a GPL and freely
distributed. If developers improved the platform, they would be required
to disclose the improvements by means of documentation. Although the platform
developers would not be able to control the development of the platform,
they would be certain that no other developer could appropriate the proprietary
rights either. The mandatory disclosure would assure all developers that
trade secrets will not balkanize the platform.
{75} Furthermore, there would still be incentive
to innovate platforms because the benefits would accrue to the development
community as a whole. As a result of such a distribution of benefits, there
would be a community-wide effort to develop the platform to its most superior
form, thus preventing sub-optimal innovation.[147]
While the essential nature of the reward would be different, it would still
serve the same function, providing incentives to innovate.[148]
An operating system developer, possessing all the compatibility-related
information, could then focus on developing other aspects of the operating
system. Consequently, competition in the operating system market could
be enhanced by placing all the developers on the same level in regard to
compatibility. Additionally, inventive tipping would become meaningless
because all platforms would share the same components, making the applications
universally compatible.
{76} Although these alternatives are not flawless,
they serve as a starting point for examining what type of legal protection
should be tailored for software platforms. In order to prevent the reward
function failure in the operating system market, steps must be taken to
prevent the occurrence of disincentives to innovate, as well as a basis
for rivalry in development.