Thursday, November 24, 2022

A (Kind Of) Successful Overbreadth Attack

Rovi Guides, Inc v BCE Inc 2022 FC 1388 Lafrenière J

2,336,870 / 2,339,629 / 2,425,482 / 2,514,585 / Interactive Television Program Guide / Internet Protocol Television

As discussed in my last post, in this decision Lafrenière J held that all of the asserted claims of Rovi’s patents were invalid for anticipation and / or obviousness. The last post focused on remedial issues. This post discusses overbreadth.

As Lafrenière J put it, “A claim is overly broad and invalid if it asserts ‘exclusive property or privilege in something the inventor did not actually invent’” [298]. In my articles on Overbreadth in Canadian Patent Law (2020) 33 IPJ 21 and (2021) 33 IPJ 147, I showed that overbreadth is functionally redundant in almost all cases in which it is invoked; that is, in almost all cases, the factual basis and reasoning leading to a finding of invalidity for overbreadth is exactly the same factual basis and reasoning leading to a finding of invalidity on some other standard ground, such as anticipation: see eg BVD v Canadian Celanese [1937] SCR 221. (The historical exception is Amfac (1986) 12 CPR(3d) 193 (FCA), which I argued was wrongly decided.)

Now, even though overbreadth is functionally redundant in almost all cases, that does not mean it is necessarily functionally redundant all cases; so, in my article I gave the example of the ‘roads to Brighton’ problem in which overbreadth might in principle play an independent role. In Seedlings 2021 FCA 154 [50] the FCA held that “overbreadth remains a proper ground of invalidity,” but at the same time Locke JA acknowledged that “the overlap with other grounds of invalidity makes it difficult to define circumstances in which a patent claim would be invalid for overbreadth but not also invalid on other grounds” [52]. Since Seedlings, parties attacking a patent have attempted to give some independent force to an overbreadth attack, so far without success.

In this case the defendants made overbreadth attacks on two patents, one of which failed and one of which succeeded, albeit in obiter.

The overbreadth attack on the 870 patent was novel. The factual basis for the Defendants’ attack was that the 870 patent claims priority from US provisional application 60/092,807 (the ‘807 Application’). One of the inventors of the 870 patent admitted that “the invention he made in July 1998 was limited to that which he described in the 807 Application” [300]. It was undisputed that the matter disclosed in the 807 Application has very limited overlap with the 870 patent: [301]–[302]. What the inventors claimed in the 870 Patent was therefore significantly broader than the 807 Application, and the Defendants argued that the 870 Claims are invalid on that basis [303].

Lafrenière J rejected this; he agreed with Rovi that “the ‘invention made’ is to be determined as of the filing date, not the priority date” [304]. This must be right, or overbreadth would swallow the claim date rules. The claim date is the filing date unless priority is claimed to a prior application “disclosing the subject-matter defined by the claim”: s 28.1(1)(a)(i),(ii). If the subject-matter disclosed by the application as filed is broader than that in the priority document, the only consequence is that priority cannot be claimed. And that is on a claim by claim basis, as the term “claim date” indicates. Claims that are directed to subject-matter that was disclosed in the priority document can still claim priority. To invalidate the entire patent would upend that well-established system.

The overbreadth attack on the 482 patent was different in nature. The 482 patent relates to a system and method for automatic retrieval of on-demand media data so as to avoid interruption to prevent unacceptable delays in the display of the video-on-demand listings during periods of higher traffic or low bandwidth connections. Lafrenière J rejected a construction which would have included “manual only” retrieval of on-demand data, but nonetheless considered and accepted the Defendants’ argument that if the 482 claims covered manual retrieval, they would be overbroad: [556]–[563].

On the claim construction point, as I understand it, Rovi wanted to argue that the claims included user-initiated download of on-demand data, presumably to bolster their infringement argument: see [562]. The claim at issue expressly requires “automatically retrieving” the on-demand data [473]–[474], and Rovi’s expert had admitted that user initiated retrieval was not automatic retrieval [500]. So, the argument that the patent covered manual retrieval was a complete dud. This makes it a bit difficult to grasp the overbreadth argument, since the premise that the claim encompasses manual retrieval is without foundation.

Lafrenière J’s discussion was understandably brief, given his holding on claim construction. Much of his discussion reviewed the evidence to the effect that the patent was entirely addressed to technology using automatic retrieval, which is of course consistent with Lafrenière J’s finding on claim construction. Manual retrieval simply was not addressed.

With this background, the key discussion was that

[561] I agree with [the Defendants’ expert] that the Skilled Person would understand that the invention described in the 482 Patent requires “automatically retrieving” at least some of the on-demand media data without a user indication to access it because “manual only retrieving” the on-demand media data causes the very undesirable consequences that the 482 Patent seeks to avoid.

[563] Dr. Robinson’s uncontested evidence is that there is no disclosure in the 482 Patent about manually retrieving on-demand media data. Therefore, to the extent such a claim is made in the 482 Patent, the claim is broader than any invention disclosed.

So, the gist of it seems to be that the 482 patent made no mention of manual downloading, so if it had claimed manual downloading, it would have been overbroad, in much the same way as the 482 patent would have been overbroad had it claimed a compound that cures cancer on the basis of a description focusing entirely on interactive television guide technology.

More broadly, a claim is overbroad if it claims subject-matter that is entirely unrelated to the disclosure. That seems reasonable enough, but the question is whether this raises overbreadth as an independent ground of invalidity. I suspect that in the vast majority of cases, a patent that claims subject-matter unrelated to the disclosure would be invalid for insufficiency. Put another way, is it possible for a patent to have claim to subject-matter that is entirely unrelated to the disclosure, in which the claim is nonetheless not invalid on any traditional ground? What if the 482 patent in this case had omitted the word “automatically” in the claim, and it was held that the claim, properly construed, encompassed manual retrieval. Suppose also that automatic retrieval was technically difficult, and the disclosure exclusively addressed automatic disclosure without mentioning manual retrieval; but at the same time, manual retrieval was not technically difficult and could easily be implemented by a skilled person in light of their common general knowledge. Would that patent be invalid simply on the basis the description did not mention manual downloading ie even if manual downloading was novel and non-obvious? I’m not sure.

My sense is that the success of the overbreadth argument in this case turned on the fact that it was premised on a claim construction that had no foundation in the patent. This makes it difficult to draw broader lessons from the analysis. It’s not very surprising that if we give a claim a construction that has no basis in the patent, the claim will be invalid in light of that patent. I doubt that observation is going to generate any independent role for overbreadth in the context of a properly construed claim.

No comments:

Post a Comment