2967187 / 2635571 / 2775674 / 2553922
Rovi Guides, Inc v BCE Inc 2022 FC 979 Brown J
2753243 / 2631957 / 2952467 / 2691719
These companion decisions raise the difficult issue of divided infringement, which arises
when a patented invention is jointly put into effect by two different entities, so it is
difficult to identify a single infringer. The problem is particularly important in method
claims, including personalized medicine and internet-related inventions, where one
party practises some element of the method and a different party practises the
remaining elements. An example is the invention at issue in the leading US decision,
Akamai v Limelight 797 F3d 1020 (Fed Cir 2015) (en banc)*. Akamai involved a method
claim relating to methods for delivering content over the internet. The defendant,
Limelight, operated a content delivery network. Limelight carried out several steps of
the claimed method, but Limelight’s customers, not Limelight itself, performed the
remaining steps of the claimed method, namely ‘tagging’ and ‘serving’ (1024). Between
them, Limelight and its customers put the claimed invention into practice, but neither
individually practised all the steps. The question was whether Limelight is liable for
infringement.
These companion decisions appear to raise similar issues. The patented inventions
relate to “interactive television program guides” [IPG]—the familiar interactive menus
used to select current and upcoming programs. The nature of the technology is not
discussed in detail, since the motion to strike turned on the law, but it is clear enough
that it raised problems of infringement involving multiple parties. Brown J, affirming
the decision of Aalto CMJ, refused to strike allegations of infringement by common
design and infringement by attribution, This does not, of course, mean that these are
necessarily good causes of action, but the decision does confirm that the law of
infringement involving multiple parties requires further development. Because the facts
of this case are not clear, for the purposes of the following discussion I’ll treat Akamai as
a paradigmatic example. The reasons are somewhat better developed in the Videotron
decision and in this post paragraph numbers will refer to that decision.
In the simplest type of infringement, a single natural person practices all elements of the
invention. But practising the invention may, and commonly does, involve multiple
parties. This does not necessarily raise doctrinal difficulties. For example, if the alleged
infringer subcontracts with another party to make a part for a patented invention, and
then takes delivery of the part and incorporates it into the completed invention, the
alleged infringer will be directly liable by virtue of having itself made the invention,
without any need to attribute the acts of the subcontractor to the primary party. But
other contexts involving multiple parties may raise more difficult problems; for
example, where one party encourages another to infringe (inducement), or in the
context of method or process patents, where one party performs some of the steps of the
claimed method or process and another performs the remaining steps (divided
infringement).
These scenarios create a tension between two principles. On the purposive definition of
infringement, the patent is infringed when the patentee is deprived of the benefit of the
invention: Monsanto v Schmeiser 2004 SCC 34 [44]. If the invention is in fact put into
effect by someone else, the patentee is deprived of that benefit whether it was put into
effect by one person or two people acting together. On the other hand, patent
infringement is a strict liability tort—a party will be liable if it performs the acts
constituting infringement, whether or not it intended to infringe. Since the patent is a
public document, a potential infringer can in principle determine whether it is infringing
by comparing its acts to the claimed invention to determine if it is practising the
invention. We shouldn’t exaggerate the efficacy of a freedom to operate search; even in
simple cases of direct infringement by a single party, the practical difficulty of
identifying relevant patents can create uncertainty. But the problem is evidently
substantially worse if a party can be made liable even though it is not putting all the
elements of the invention into effect. Consequently, the main objection to imposing
liability for divided infringement is that making a party liable for performing only one
element of an invention introduces uncertainty and a consequent chilling effect: see
Limelight v Akamai 572 US 915 (2014) 922; Akamai v Limelight 786 F.3d 899 (Fed Cir
2015) 905. Dealing with the chilling effect is a key issue in infringement involving
multiple parties. I’ll suggest that a common thread in the doctrines that extend liability
beyond the case in which a single entity which itself directly practices the invention is
that the defendant who is made liable for infringement has the same opportunity for
knowledge, or at least control, as a natural party who practices all elements of the
invention itself.
Inducement
In broad terms, there are three approaches to the problem of multiple parties.
In indirect infringement, the question is as to the liability of a party, A, who in some
manner facilitates infringement by a direct infringer, B, typically by supply of a product
which is itself non-infringing, but which is used for an infringing purpose by the direct
infringer. In the paradigmatic case, there is a single direct infringer which practices all
elements of the invention. Inducement is a well-established form of indirect
infringement in Canadian law. (It was pleaded in this case, but was not subject of the
motion to strike: [6]–[7].) It is well established that the indirect party must know that
its influence would result in the acts constituting infringement, thus addressing the
chilling effect problem; for example, a party who supplies a commercial staple will not
be held liable even if the purchaser used it for an infringing purpose, since there are
many non-infringing uses for a staple product and the vendor would not normally be
aware of what the purchaser will do with the product.
Another form of indirect infringement (albeit not formally recognized in Canadian law)
is contributory infringement, in which liability is imposed for supply of a product
especially adapted to infringe. Because the product only has one use, the vendor must be
taken to know what it will be used for, and consequently is in as good a position as the
direct infringer to verify whether the use is indeed infringing.
Attribution
Second, some doctrines attribute the acts of one party, B, to another party, A, so that A
may be held to have directly infringed even though neither A nor B carried out all the
elements of the invention. This type of infringement was pleaded in this case, specifically
“direct infringement as a result of attribution of [third party] actions” [6]. Aalto CMJ
remarked that the concept of attribution is novel in patent law [62], [65], but he and
Brown J accepted that the principle is well-accepted in tort law [62], [68] and given that
patent infringement is generally considered to be a form of tort, it would be an
“incremental development in the law” to import the concept into patent law [66].
Consequently, Brown J, affirming Aalto CMJ, allowed it to proceed, while expressly
noting that he was not saying that it was a good cause of action, but only that it should
be allowed to proceed [68].
The main authority relied on by Brown J and Aalto CMJ was the 2015 decision of the en
banc US Fed Cir decision in Akamai v Limelight 797 F3d 1020 (Fed Cir 2015) (en banc);
as noted above, the defendant, Limelight, performed some steps of the claimed method
and its customers performed the remaining steps, namely ‘tagging’ and ‘serving’ (1024).
The Fed Cir had previously recognized that infringement involving multiple parties
could be attributed to one of the parties, so as to constitute direct infringement, solely in
three circumstances: principal-agent relationships, contractual arrangements, and joint
enterprises (1023). In Akamai the Fed Cir held that this was unduly restrictive. On the
facts, Limelight required its customers to perform the tagging and serving steps, both by
contract and technically (1024–25). Consequently, there was substantial evidence that
Limelight “directed or controlled its customers’ performance of each remaining method
step” (1025). The Fed Cir held that this allowed the customers’ acts to be attributed to
Limelight so as to make Limelight a direct infringer.
So far as I can tell, there isn’t a unified doctrine of attribution as such in US law; rather,
the Fed Cir used the term as a label for a variety of circumstances where the acts of one
legal entity are imputed to another. The court also noted that the categories are not
closed: “[i]n the future, other factual scenarios may arise which warrant attributing
others’ performance of method steps to a single actor” (1023).
To my mind, the facts in Akamai illustrate why it was sound to allow the plea of
attribution to proceed. It is clear on the facts in Akamai that the problem that the
defendant might not know that it was infringing does not arise. Limelight specifically
instructed all the steps of the claimed method, even though it did not carry them out
itself. The knowledge requirement of the well-established inducement test would clearly
be satisfied. The encouragement and instructions would also satisfy the second step of
the inducement test. Thus the chilling effect problem is avoided: a party who encourages
another to carry out certain known acts is in as good a position as the other to determine
whether those acts constitute infringement. The only reason this might not constitute
inducement relates to the first step, the requirement of direct infringement. In Limelight
v Akamai 572 US 915 (2014) the US Supreme Court held that there could be no
inducement unless a single party was directly liable for the entire act of direct
infringement. Whether this is true in Canadian law is a separate question, which I’ll turn
to next; for now it suffices to say that if a trip to the US Supreme Court was warranted
on this issue in the US, a trip to trial is warranted in Canada.
Apart from relying on Akamai,
Brown J stated that the concept of attribution is well-accepted in tort
law. I must admit that I’m not familiar with it in that context, at
least
not by that name, and no cases were cited. Perhaps he had in mind tort
doctrines of
vicarious liability, under which an employer is liable for tort of an employee
carried out in the
course of their duties. Infringement is generally treated as a species
of tort, and when a
corporation practices the invention, all of the various elements of the
invention are, in
the simplest case, physically put into effect by a single employee of
the corporation
acting within the scope of their duties. In that case, the employee, a
separate legal entity,
is a direct infringer, but the infringement of the employee is
attributed to the
corporation by the doctrine of vicarious liability, with the result that
the employer is
liable as a direct infringer, not for inducement. (There is a separate
question as to
whether the employee is liable personally: see Mentmore (1978) 89 DLR(3d) 195 (FCA);
Liability of Corporate Officers in Intellectual Property Law, (2020), 63 CBLJ 159.) This
is routine, but it nonetheless illustrates the point that the direct infringement of one
party may be attributed to another, so as to make the other a direct infringer. In these
circumstances it is evidently not unreasonable to task the employer with knowledge of
the acts carried out by the employee at the employer’s instructions or at least on its
behalf.
But there is a puzzle even in this seemingly straightforward example. Suppose that the
patent claims a method. In the case where a single employee carries out all the steps of
method and so is a direct infringer, it is clear that vicarious liability allows the wrong of
the employee to be attributed to the employer. But what if the same invention is put into
effect by two different employees, one of whom carries out some of the steps, and
another who carries out the remaining steps, so that neither of the employees is a direct
infringer. It would seem absurd to say that the employer is not liable simply because it
directed two employees to carry out the method instead of one. But neither employee
commits a wrong which can be attributed to the employer; the employer will only be
liable if we attribute the acts of the employees to the employer.
Does the doctrine of vicarious liability attribute the wrongs of the employee to the
employer, or their acts? It appears that it is the former in US law: so, in Akamai v
Limelight 786 F3d 899 (Fed Cir 2015) 909 the Court stated that “Under the principles of
vicarious liability, direct infringement does not occur unless all steps of a method claim
are performed by or attributable to a single entity.” Similarly, in the en banc Akamai
decision, the Fed Cir stated that to determine attribution, they would consider “general
principles of vicarious liability” (1022), not vicarious liability as such, explaining in a
footnote that “previous cases' use of the term ‘vicarious liability’ is a misnomer. . . . In
the context of joint patent infringement, an alleged infringer is not liable for a third
party's commission of infringement—rather, an alleged infringer is responsible for
method steps performed by a third party. Accordingly, we recognize that vicarious
liability is not a perfect analog” (1022 fn2). For that reason, the Court held that the acts
of the customers should be attributed to Limelight, not on the basis of vicarious liability
as such, but rather in light of a “direction or control” standard derived from the “general
principles” of vicarious liability (1022).
To my surprise, I have not been able to
find a ready answer to this question in Anglo-Canadian law. The standard
way of describing it is to say that the law of vicarious
liability holds the employer liable for the “misconduct” of the employee
(see eg Sagaz
2001 SCC 59 ¶ 2; London Drugs [1992] 3 SCR 299, 334; Bazley v Curry [1999] 2 SCR
534 ¶ 19), which implies that the employee’s conduct is itself wrongful. But we shouldn’t
place too much weight on an isolated phrase. Read holistically, the cases on vicarious
liability generally refer to acts and wrongs more or less interchangeably, as does
Glanville Williams, Joint Torts and Contributory Negligence (London: Stevens & Sons,
1951). Presumably this is because in the great majority of cases it is a single employee
who acts wrongfully, so that it makes no difference whether the acts or the wrongs are
attributed to the employer. There is some parallel with cases of concurrent liability, as
when two rioters combine to overturn a car which neither could have overturned if
acting alone; but in that case, each commits an independent tort of trespass to chattels.
I have no doubt that if a method claim were put into effect by two employees of the same
employer, both acting in the course of their duties, the courts would attribute their acts
to the employer. Vicarious liability has its basis in “a combination of policy
considerations” (Bazley v Curry [1999] 2 SCR 534 ¶ 26) and whether one or two
employees put the invention into effect clearly doesn’t matter from a policy perspective;
the employer has the same control and knowledge whether the acts which jointly
constitute the wrong are performed by one employee or two. This doctrinal difficulty
illustrates that divided infringement of method claims really does raise some unique
issues; even in a simple case of divided infringement by two employees of the same
employer, some extension, or at least clarification, of standard vicarious liability law is
required. Then the question is where to draw the line. What if the alleged infringer
carries out some steps of the method and contracts with a third party to carry out the
rest? What if it induces its customers to carry out the rest? In any of those cases the
alleged infringer would have the same ability to determine whether its acts infringe as
would a single direct infringer. Thus, it seems clear that the cause of action should be
allowed to proceed to develop a fuller factual matrix in which to determine whether
there is a principled line to be drawn.
I’ll end this section by noting that the general concept of attribution is not entirely novel
in Canadian patent law. In Bauer v Easton 2010 FC 361 the action was based on
inducement, which Gauthier J held to have been established. But she suggested in obiter
at ¶ 190 that the defendant’s involvement in making the infringing product was so
substantial as to amount to direct infringement, even though it had not itself carried into
effect every element of the invention.
Common Design
The second cause of action at issue was infringement by common design, which Aalto
CMJ defined as follows [8]:
Infringement by common design arises where one party is found to be a joint
tortfeasor when another party commits the tort in furtherance of a common plan.
An essential element of the common design is that the parties must agree on a
common action and the act of infringement must be in furtherance of that
agreement. There must be a common design to do the act that is alleged to
infringe.
Aalto CMJ noted that the UK Supreme Court had accepted the concept of infringement
by common design in Sea Shepherd [2015] UKSC 10. While Sea Shepherd was not a
patent case, the UKSC at [23], [37] endorsed the summary by Mustill LJ in the patent
case of Unilever v Gillette [1989] 106 RPC 583 (CA) 608.
Aalto CMJ noted that the FC has considered common design for patent infringement in
Packers Plus 2017 FC 1111 [48], noting there was no Canadian authority but adverting to
Sea Shepherd before dismissing the argument on the facts; and in Genentech v Celltrion
2019 FC 293 permitting a pleading of “acting in concert,” which was considered
equivalent to infringement by common design. Given the authority of the UKSC and the
ambiguous state of the law in Canada, it was relatively straightforward for both Aalto
CMJ and Brown J to conclude that the pleading of infringement by common design
should be allowed to stand and indeed Videotron accepted that it was a viable cause of
action on appropriate facts.
Videotron’s main argument was that the doctrine required identification of a “primary
tortfeasor” [51]. Brown J did not lay out Videotron’s argument in much detail, but if I
understand correctly, Videotron was arguing that the doctrine requires a primary
tortfeasor that would itself be liable, in the same manner that inducement requires a
direct infringer [51]. Brown J rejected this saying that the term “primary tortfeasor” is
used merely “to identify the party to the common design that joins the defendant in the
tort (but is not named as a defendant)” [52]. I’m not sure Brown J is right about
this—though I’m not sure he’s wrong either. In the cases reviewed by Mustill LJ in
Unilever v Gillette, it appears to me that there was always a single direct tortfeasor and
the question was whether another could be joined or otherwise also held liable for the
same tort. In Unilever v Gillette itself the question was whether the parent of a domestic
defendant could be joined, and it is clear that the domestic defendant was an infringer
on the facts as pleaded. In Sea Shepherd, it is clear that the primary defendant was itself
a tortfeasor, and the question was whether its parent company could be made liable.
Brown J also referred to ICBC v Stanley Cup Rioters 2016 BCSC 1108 and Montréal
(Ville) v Lonardi 2018 SCC 29. He noted that neither decision required a “primary
tortfeasor” who performed all acts necessary to damage any vehicle; that is true, but
neither did they say the opposite. In both the question was whether individual rioters
should be held liable for all the damage done by riots, and in both the answer was no.
This is not to say that I agree with Videotron’s argument. So far as I can tell, at least
from the cases I looked at, the question never directly arose. The cases concerned
scenarios in which there was in fact a single direct tortfeasor and the question was
whether a second party should also be made liable. The references to a ‘primary
tortfeasor’ may have indicated a direct infringer, but only for convenience on the facts. I
certainly don’t see any case holding that there must be a single direct tortfeasor. Certain
statements by Lord Neuberger in Sea Shepherd at [55], [60] might be seen as suggesting
a requirement that there be a single direct tortfeasor, but I think this is merely a
reflection of the facts of the cases he is summarizing, and not a considered holding on
the issue.
With all that said, as a matter of principle, there is no evident reason why infringement
by common design should require a single direct infringer. If two parties agree with each
other that one should carry out two elements of the patented method and the other
should carry out the remaining elements in coordination, with the intended and actual
result that the invention is practiced, both parties are in as good a position to know that
the invention would be practised as a single party which put all the elements into effect
itself.
As with vicarious liability, the tort cases on infringement by common design are not
entirely helpful on the unusual problem raised by divided infringement of method
claims, which seem to have no exact parallel in tort law. This is all the more reason to
allow these causes of action to stand. I might add that while infringement by common
design is clearly established in English law, it does not seem to be very well elaborated;
in Sea Shepherd the action was dismissed on the facts with only a brief summary of
principles at [21]. While Aalto CMJ’s definition strikes me as a very good summary, I’m
not sure it is right, at this point, to treat it as definitive.
Overlap
In my view, Aalto CMJ and Brown J were unquestionably right in allowing these causes
of action to proceed. It is sometimes desirable to deal with a novel question of law on a
motion to strike, as in Atlantic Lottery 2020 SCC 19, this is more appropriate when the
cause of action itself is well defined. The law on divided infringement is still developing
in both the US and the UK, in respect of both apportionment, as was expressly
recognized in Akamai, and, in my view, in respect of infringement by common design in
the UK. (And see Medgraph v Medtronic (Fed Cir 2016) discussed here, clarifying the
Akamai standard.) This means that the matter should be allowed to proceed in order to
develop a fuller factual background to provide context.
By the same token, we should not accept these causes of actions exactly as they are
established in the US and UK, with the only question being whether to adopt them in
Canada. Unlike Canada, the US has codified infringement provisions in 35 USC §271,
with separate provisions for direct infringement (§271(1)), inducement (§271(b)) and
contributory infringement (§271(c)). The distinction between direct infringement and
inducement was central in the Akamai litigation.* In Canada infringement is defined as
interference with the exclusive rights conferred on the patentee: Monsanto v Schmeiser
2004 SCC 34 [34]. While we distinguish doctrinally between direct infringement and
inducement, these are not distinct torts: McCain Foods 2021 FCA 4 [62]. Therefore, as
an early leading case put it, the question is whether the defendant’s act might “with
propriety” be termed an infringement: Copeland-Chatterson (1906) 10 Ex CR 224, 247.
For that reason alone, the distinctions drawn in US law should be treated with caution.
Moreover, it is not entirely clear whether these causes of action are distinct. It might
seem reasonable to distinguish them as follows: in attribution, A is not necessarily a
direct infringer, and the question is whether the acts of A should be attributed to D so as
to make D a direct infringer; in infringement by common design, neither A nor D is a
direct infringer, and the question is whether they might both be considered infringers by
virtue of their joint acts; in inducement (or indirect infringement more broadly, ie
including contributory infringement), A is a direct infringer (who may be such by virtue
of attribution), and the question is whether D should also be considered an infringer by
virtue of the aid provided to A.
However, this schema is difficult to reconcile with the cases. The three traditional forms
of attribution recognized in the Akamai litigation were principal-agent relationships,
contractual arrangements, and joint enterprises (1023). The last of these bears a striking
similarity to infringement by common design. At the least, common design and
attribution overlap in some cases: if A and D work closely together, it may be difficult to
distinguish a case where A’s acts should be attributed to D, and a case where A and D
should be considered to work together. Common design and inducement may also
overlap, as was pointed out in Unilever v Gillette, 608, where Mustill LJ remarked that
there may be doubt “as regards the relationship between indirect infringements by
procuring and by participation in a common design. There may still be a question
whether these are distinct ways of infringing, or different aspects of a single way. I prefer
the former view, although of course a procurement may lead to a common design, and
hence qualify under both” (608). Bauer v Easton illustrates the potential overlap
between inducement, which was established, and direct infringement, arguably by
attribution, which Gauthier J indicated she would have accepted.
It is of course possible that these three doctrines (assuming they were accepted in
Canadian law), are in fact distinct and simply happen to overlap on some sets of facts.
But the extent of the overlap, and the fact that the Canadian Patent Act does not have an
infringement provision as such, suggests that it might be desirable to develop an
overarching approach to infringement, perhaps focusing on knowledge and control or
influence. This is not to say that the courts should develop and apply an overarching
definition in this particular case; a more incremental approach is probably preferable.
But we should at least be open to the possibility that a unified test will develop; again,
this is a reason to be cautious in drawing a sharp distinction between attribution and
infringement, as was done in Akamai.
Other cases
I’ve already noted that Bauer may also be relevant to this debate, and I’ll finish this post
by pointing out a couple of other cases that are potentially relevant. In Valmet v Beloit
(1988) 20 CPR(3d) 1 (FCA), one question was whether Valmet, the parent of VDI,
breached an injunction “because it must, by reason of the circumstances, answer for the
acts of V.D.I. (which clearly would have breached the injunction had they been
accomplished by Valmet)” (12) and see similarly (16). This sounds like a form of
attribution argument, though not under that name. The FCA considered inducement
had not been established, and continued by saying “there is, in my view, only one other
possible situation in which Valmet would have had to answer for V.D.I.'s conduct,
namely, the situation where Valmet's control of V.D.I. would have been so complete that
V.D.I. would, in fact, have been only a puppet in Valmet's hands” (16), which was not
made out on the facts. This seems like a very stringent test for attribution—perhaps
more stringent than the “direction or control” test from Akamai. But on the facts, the
real complaint was that the parent had not acted to prevent infringement by the
subsidiary, and the FCA held that “[t]he mere fact that Valmet did possess the power to
influence the decisions of that board did not render Valmet liable for those decisions if,
in effect, it did not influence them” (16), which is entirely consistent with the Akamai
approach, and with a less stringent test for attribution generally. It is similarly
consistent with Bauer on the facts.
The case of Incandescent Gas Light Co Ltd v New Incandescent Mantle (1898) 15 RPC
81 (QB) (cited in Copeland-Chatterson (1906) 10 Ex CR 224, 243–44) is perhaps an
early example of infringement by common design, though not under that name. The
invention related to mantles for incandescent gas lighting. The defendant sold fittings
for incandescent gas lighting from the ground floor of a building, with another company
selling infringing mantles from the first floor. There was a speaking tube between the
two and customers knew that fittings were to be supplied on the ground floor and
mantles on the second floor. The defendant was held liable for infringement. Another
arguable example is American Arch Co v Canuck Supply Co [1924] 3 DLR 567 (QSC) in
which the patent at issue was for a locomotive fire-box, consisting of a particular
arrangement of fire bricks and other components (570, 573). The court, finding the
defendants jointly and severally liable, stated that “[t]he defendants acting in concert,
one as a manufacturer and the other as sale agent, have adopted, to the smallest detail,
the method employed by the plaintiff in installing its fire arch; . . . in short, they have
jointly put in practice the invention covered by the patent in suit, in the only manner in
which it could be put in practice and. as such, are infringers.” This could be understood
as a case of contributory infringement, except that it is generally thought not to be a
good cause of action in Canadian law; an alternative would be to consider this a case of
infringement by common design.
*The Akamai litigation has a convoluted history. In Akamai v Limelight 629 F3d 1311
(Fed Cir 2010) a panel of the Fed Cir held that Limelight was not liable for direct
infringement under 35 USC §271(a) because Limelight did not perform all of the steps of
the asserted method claims and the actions of its customers could not be attributed to
Limelight. In a rehearing en banc, the Fed Cir found it unnecessary to address direct
infringement, as it held that Limelight was liable for inducement under 35 USC §271(b),
even if no one would be liable as a direct infringer—that is, there can be direct
infringement for the purposes of inducement without there being a direct infringer:
Akamai v Limelight 692 F3d 1301 (Fed Cir 2012) (en banc). That decision was appealed
to the US Supreme Court, which reversed and remanded, holding that for there to be
direct infringement, all of the steps of a method claim must be attributable to one
person: Limelight v Akamai 572 US 915 (2014) 922. At the same time, the Supreme
Court noted that the question of direct infringement was not at issue on the appeal. On
remand, a panel of the Fed Cir held that there was no direct infringement because the
acts of Limelight’s customers could not be attributed to Limelight: Akamai v Limelight
786 F3d 899 (Fed Cir 2015). Finally, in a rehearing en banc, the Fed Cir held that the
acts of Limelight’s customers could be attributed to Limelight, so Limelight was directly
liable: Akamai v Limelight 692 F3d 1301 (2012) (en banc).